Amazon, Google, Meta, and Microsoft have spent more than a trillion dollars on AI infrastructure, including data centers, the chips inside them, and the power needed to run the facilities, since 2023. The Financial Times said that these four big companies have already hit $1.1 trillion in capital expenditure based on their latest earnings reports, and that an additional $745 billion is expected to be added to this figure just this year.

“There is basically no end in sight for the growth in capex,” RBC Capital analyst Rishi Jaluria told the publication. “Investors need these companies to toe the tight line between investing in AI and not compromising the things that have made them successful.” This massive investment has upended several other industries — namely electricity prices and memory and storage chips. The massive power demand that data centers have put on the power grid has forced many U.S. utility companies to spend billions of dollars to upgrade their respective infrastructure, which they then passed on to all consumers, not just the big ones that forced the upgrade.

This, alongside other environmental issues, has caused many Americans to push back against data center projects near their communities. The White House instituted the “ratepayer protection pledge” and made AI hyperscalers, utility operators, data center companies, and individual states promise that they will protect the average consumer from electricity cost increases. But so far, no state has taken a step to codify this pledge into law. Oregon actually enacted the POWER Act, which resulted in a 30% increase in the power bill of users that consumed more than 20MW while slashing the bills of residents by 1.3%, but the state did this in 2025, way before President Donald Trump called the tech giants into the White House and told them to “pay their own way.”

The mountains of cash that these tech giants are pouring into AI are also affecting the memory and storage chip industry. Since these AI hyperscalers have a lot of liquidity from investors, they are willing to pay top dollar for the HBM they need to run their data centers. Because of this, it made sense for Micron, Samsung, and SK hynix to prioritize them over DRAM, especially as they can charge a premium for these chips and there are customers who are willing to pay at those prices. This resulted in a shortage of consumer memory that started in 2025 — while this initially affected PC builders and enthusiasts, it has started to affect other industries that require memory as well, including cars and smartphones. Even Apple, which historically had huge sway over its suppliers, was forced to increase prices because of the shortages.

Aside from skewing other industries, the massive CAPEX the big four are going into is alarming some experts, warning that the promises and contracts they’re making are leading to “hidden debt” not listed in their balance sheets. The amount, worth around $1.65 trillion, is annotated in their quarterly financial statements as future obligations that will only come into play as the related asset or service comes online. The current value is 122% of the actual debt reflected on their balance sheets, which could give investors the wrong impression that they have fewer obligations than they actually have.

While the amount of money that the big four are spending on AI might seem dizzyingly high, we must note that these companies are raking in massive amounts of cash quarterly themselves. Microsoft’s latest quarterly revenue is $90 billion, while Meta made $60 billion in the same period. Alphabet (Google) announced revenue of nearly $120 billion, while Amazon made $200 billion. That is a total of nearly $470 billion for these companies in just the last quarter.

Still, that does not mean that they can just keep on spending on AI. For example, even though Google’s cloud business had a revenue of $11 billion last year, its price dropped after it announced that it spent more than it made last quarter — the first time this happened in the 20 years since it went public. Meta is also planning to rent out its AI compute, apparently following in the footsteps of Amazon, Google, and Microsoft, which have growing cloud businesses. However, this announcement caused a drop in its stock price. “They are a bit all over the place,” SLC Management managing director Dec Mullarkey told FT. “For investors it’s no longer growth at any cost; they want to see the spending flowing through to results, like at the Big Three.”

Microsoft announced that it’s increasing Minecraft’s minimum and recommended system requirements for the first time in 17 years. According to the company’s blog post, Minecraft: Java Edition will get a spec bump to ensure that the game runs smoothly today and in the future. While it’s understandable that the game’s developers would want to increase the performance of the hardware that the title runs on to implement more advanced features, the new requirements are significantly higher than the previous minimum specification. This could negatively affect players on older hardware that barely hits the minimum requirements, who would have no choice but to upgrade their PCs at a time when RAM prices are at a record high.

The new minimum specifications are quite reasonable, with the recommended specs falling squarely within the performance of the most common hardware in June’s Steam Hardware Survey (which also indicates that AMD now has more than 45% market share compared to Intel). If your system used to be at or higher than Minecraft: Java Edition’s previously recommended specifications, but no longer hits the current minimum requirements, you can switch over to the Bedrock Edition to continue gaming, although you might have to forgo some of your favorite mods.

If you’re running the bare minimum hardware to play Minecraft, Microsoft said that you could still possibly launch the game, although it warned that you might run into some issues. “These updated recommended system requirements won’t stop you from playing Minecraft: Java Edition today. If your computer falls below the new minimum requirements, Java Edition should still launch,” the company said in the Minecraft blog. “However, its performance and visual quality are no longer guaranteed on below-spec hardware, and it may struggle or fail to launch Java Edition in the future – especially as we transition away from OpenGL and towards Vulkan. We will keep players informed of these changes, especially changes that will sunset hardware that previously supported Minecraft: Java Edition.”

This means that you’ll have to spend some money to upgrade your system to just to continue playing your favorite game. You don’t have to spend thousands of dollars to get a new PC or join a giveaway to be able to play, though. If you know what you’re looking for, you can plausibly buy a used gaming laptop or desktop PC for around $350 to $500 — that should give you about a decade or more to enjoy the game if Microsoft keeps its spec upgrade cadence of 17 years for future versions of Minecraft.

Xbox has announced a new Xbox Backward Compatibility for PC program, which will let gamers play a selection of classic Xbox titles on PCs and handhelds, a move the company says marks the beginning of a bigger push into Xbox game preservation. The announcement begins with four titles that can now be played on PCs and handhelds, including the ROG Xbox Ally and Ally X.

"This marks the beginning of a broader effort to preserve XBOX games from the past and bring them to PC over time," the company said in a blog post authored by Xbox "VP next generation" Jason Ronald. Alongside backward compatibility, the company says games will also include new features. The four titles in the announcement are:

• BLiNX: The Time Sweeper 
• Conker: Live and Reloaded 
• Crimson Skies: High Road to Revenge 
• Fuzion Frenzy

You can now buy all of these games on PC, and they're also included in all Xbox Game Pass plans. Anyone who already owns these titles digitally on console can also now play them on PC or handheld, with support for Xbox Play Anywhere and Xbox Cloud Gaming. Crucially, this appears to be a digital-only preservation effort, so it won't help anyone who only owns physical copies of these games. Xbox has reportedly been testing a way to digitize physical games as far back as Xbox One, but that hasn't yet materialized yet.

Alongside the preserved original gameplay, Xbox claims these games will let users customize graphics settings, with up to 4x resolution scaling, VSync support, Fullscreen and Windowed modes, anisotropic filtering, and enhanced anti-aliasing, with more features to come in the future. Notably, Xbox will add achievements to select original Xbox games on console and PC.

Users who want to play these titles will need to meet the following system requirements:

Microsoft's emphasis on preserving older games is particularly noteworthy given recent backlash against Sony over its decision to end physical games with its next console generation, and the shuttering of some of its online stores. While Xbox's digital emphasis on game preservation still can't match the solidarity of physical media preservation, the timing of this announcement seems like more than just a coincidence.

Five U.S. tech giants heavily invested in AI and its related infrastructure reportedly have an estimated $1.65 trillion in hidden debt, with the figures annotated in their quarterly financial statements instead of being listed in their balance sheets. According to Nikkei Asia, this is higher than the $1.35 trillion officially listed, meaning investors could be caught unaware once the hidden figures come to light.

The publication says that Meta has a high off-balance-sheet-to-recorded-debt ratio, with the company owing $420 billion in unlisted debts compared to the $140 billion written on the balance sheet. Oracle purportedly also has a massive $273.3 billion of hidden debt, which is a 2,900% jump from the hidden debt it had from 2022.

This may sound strange, but it’s actually an accepted accounting practice. The “hidden debt” stems from long-term contracts that have been signed but have not come into force yet, which, Nikkei says, is mostly related to the billions of dollars promised to data center operators. The AI race has got many hyperscalers signing contracts and agreements with data center operators, saying that they will pay for the compute they generate once their project goes online.

While any institution promising to pay any amount of money for services or goods delivered is obliged to list them as a liability, the fact that these data centers haven’t started operations means that these agreements are off-the-books at the moment. But when these projects come online, the contracts that the tech giants have signed will come into force, and they’ll have to pay for the compute that these sites will deliver, no matter if there is demand or not.

Nevertheless, these tech companies aren’t just pouring money into future contracts just for the sake of it. Alphabet, Amazon, and Microsoft reportedly have a cloud service backlog worth $1.45 trillion, meaning these are services yet to be rendered and paid. Amazon Web Services CEO Matt Garman also told the publication that the investments that the company is getting into are “not speculative.”

While this may seem like a good way to secure capacity — sign customer contracts that guarantee demand and then enter into long-term agreements with data centers to get the compute needed to deliver the services- it opens these tech giants to massive amounts of risk. That’s because if the demand fails to materialize, then they’d be left paying for excess compute without having any customers to sell them to. What’s more alarming is that Nikkei says that these investment expenditures are exceeding their earnings, meaning these big tech companies are increasingly relying on corporate bonds and new shares to fund them.

Even though demand for AI compute is increasing, it’s still a relatively new and unproven technology, with many experts saying that it should benefit more people to avoid a bubble. The cost of using AI for nearly everything, called “tokenmaxxing,” has also caught some companies by surprise, with agentic AI eating up annual AI budgets in a matter of weeks. Because of this, some companies are reducing their use of AI or are switching to more affordable models from China. This uncertainty, paired with the way tech companies “hide” these liabilities, is quite concerning, as they would appear to have less long-term obligations than they actually do.

This isn’t the first time that an industry giant has used similar accounting techniques. The publication cited Enron’s 2001 collapse, which was due to the company hiding its troubled assets through special purpose entities and marking unrealized gains from trading contracts into its current income statements. While the tech giants are not hiding underperforming assets off their balance sheets and committing fraud, they’re still using a similar mechanism to list their upcoming obligations. Although these are technically not debt, they still behave like one, and the way they’re reported is what’s concerning some experts.

The demand for AI compute is already insatiable, and it seems only poised to grow in the wake of the introduction of frontier-class open models like Kimi K3 that anybody can potentially fine-tune and serve. Against this backdrop, Microsoft and AMD are teaming up to get Redmond more AI FLOPS for both internal and external use. The two companies announced this morning that Microsoft will commit to adding AMD's Helios rack-scale AI accelerator in volume to run frontier-model workloads in its own data centers, as well as for Azure AI infrastructure customers and services.

The partnership makes next-gen AMD AI compute available to Azure customers like AI labs for AI training and inference serving workloads, and it’ll also underpin managed compute for enterprise customers looking to deploy AI workloads through Microsoft Foundry.

The two companies didn't indicate the exact size of Microsoft's Helios deployment in either watts or dollars, but the commitment would seem to be another major win for AMD as it seeks to grab data center GPU share from Nvidia. AMD has struck massive partnerships with OpenAI and Meta in the past year with gigawatts of compute installations and hundreds of billions of dollars potentially hanging in the balance.

For a quick refresher, the Helios rack-scale accelerator will take the fight to Nvidia’s Vera Rubin NVL72 system when it arrives later this year. Helios joins together 72 next-generation Instinct MI455X GPUs with an aggregate of 31.1TB of HBM4 memory capacity across the system. Those GPUs offer as much as 1.4 exaFLOPS of FP8 compute and 2.9 exaFLOPS of FP4 for AI models using those OCP AI data types.

AMD is targeting 260 TB/s of scale-up bandwidth within the rack, on par with Nvidia’s Vera Rubin NVL72 rack-scale system, and 43 TB/s of scale-out bandwidth using UALink over Ethernet, or about twice that of Vera Rubin, although the performance of UALink over Ethernet in practice remains to be seen.

Microsoft and AMD also announced that Azure will add two new VM series built on AMD's upcoming sixth-gen Epyc Venice CPUs: the HDv2 series for "agentic AI and data pipelines," and the HXv2 for semiconductor design workflows. Microsoft will also leverage its existing deployment of AMD Pensando DPUs to integrate that hardware into its Azure Boost offerings to accelerate networking and storage processing operations.

Tom’s Hardware will be on the ground at AMD’s Advancing AI event this week, where we expect to learn more about AMD’s AI ambitions for the second half of this year and beyond. Stay tuned for our coverage from that event.

Windows 95 worked out whether a setup program had run by reading the executable's name and checking it against a short list of hard-coded words, according to Microsoft engineer and semi-official Windows historian Raymond Chen, sharing the details via his Old New Thing blog. A filename containing "setup," "install," or "inst" flagged the program as an installer, which triggered the operating system's routine for repairing system files that installers had damaged. Three non-English entries appeared on the same list, which Chen identified as his own guesses at Italian, Turkish, and Hungarian.

The full match list ran to six terms: setup, install, inst, imposta, ayarla, and felrak. Chen wrote that "install" was redundant, because any name containing it already contains "inst," and speculated that the shorter entry was added later to catch installers named along the lines of "blahinst" without anyone deleting the original. A program whose own name produced no match got a second test, with Windows 95 checking whether the word "Setup" appeared anywhere in the path to the executable. A separate live check ran after any multimedia driver was installed through an INF file, added because those drivers frequently overwrote system DLLs.

This heuristic gated a recovery mechanism Chen described in March. Installers of the period overwrote system files without checking versions, disregarding Microsoft's rule that a file should only ever be replaced by a newer one. An installer carrying Windows 3.1 copies of shared DLLs, for example, would bury the newer Windows 95 versions underneath them, and every program that relied on the current files would break.

Windows 95 kept backup copies of commonly clobbered files in a hidden C:\Windows\SYSBCKUP directory. It would then let each installer finish, check its work, and restore the correct versions where the installer had downgraded them. This safety net depended entirely on correctly guessing that an installer had run, so a setup routine with an unusual name slipped past it, while an ordinary program named something like instant.exe tripped it for nothing.

The file check was often deferred until the next boot rather than run immediately, with Chen explaining that some installers, unable to replace a file already in use, would drop back to MS-DOS, run a batch file to swap the file, and restart Windows, so the cleanup pass had to wait for the reboot to catch anything the batch file had altered.

Windows 2000 debuted a new approach

Microsoft dropped this approach in Windows 2000, which introduced Windows File Protection. That system registers for file-change notifications through Winlogon restores protected files from a cache at %WinDir%\System32\dllcache, with no filename guessing and no waiting for a setup program to complete.

Then, Windows ME shipped a comparable System File Protection, and Vista onward moved to Windows Resource Protection, which guards files using access-control lists instead. The sfc /scannow command that Windows 11 users still run to repair system files descends directly from that work, having replaced a 1995 mechanism that decided when to act by scanning filenames for the word "setup."

A Brazilian gamer who lost his Microsoft account and all his digital games has won a court order forcing the company to hand it all back, after support staff told him to simply repurchase his library. The user, who posts as Ordo_Liberal on Reddit, shared a screenshot of a small-claims ruling in the Xbox subreddit that gives Microsoft 15 days to restore access or face fines and roughly $400 (R$2,000) in damages. He said the account had two-factor authentication enabled when it was flagged and permanently suspended, and that none of Microsoft's recovery options allowed him to regain access before he filed his lawsuit.

According to the support emails he posted, Microsoft suspended the account after detecting what it described as unauthorized access, and said an investigation found the account's security information had been changed. The company treated the suspension as permanent and directed him to buy his games again rather than reinstating the ones already tied to the account.

Anyone with a large Microsoft account might want to take note here, because the suspension didn’t target a single game or console but the user’s entire account. Microsoft accounts can carry Xbox purchases, Windows licenses, store apps, 365, and OneDrive data, and a permanent suspension blocks all of it, even, as in this case, when you’ve got strong security and two-factor authentication set up.

Brazil's Consumer Defense Code and its small-claims track allow individuals to bring cases without a lawyer and without paying court costs, meaning Ordo_Liberal was able to take Microsoft to court without incurring any costs. If Microsoft misses the payment deadline, an additional 10% penalty will be imposed. While the damages awarded in this case are minor, it's notable that Microsoft has been compelled to restore the account and its contents, which runs counter to the standard platform position that customers license digital games rather than own them.

This, of course, is a first-instance small-claims judgment, not binding precedent, and it covers one account in one jurisdiction. Courts elsewhere have gone the other way. Chinese courts have recognized game accounts as inheritable property, while U.S. rulings generally treat games as revocable licenses. The case comes as both Microsoft and Sony push players away from physical media, with Xbox testing a way to convert discs into digital entitlements and Sony ending new PlayStation disc production in 2028. Microsoft hasn’t publicly commented on the ruling.

Microsoft’s emissions for fiscal 2025 (FY25) rose by 25% from the previous year, even as the company’s 2030 deadline to become carbon-negative draws closer. According to the company’s 2026 Environmental Sustainability Report, released on Thursday, July 9, the backward step was driven primarily by the rapid expansion of its data center infrastructure and its decision to stop using short-term renewable energy certificates, which reduced its reported footprint without necessarily adding new clean electricity to power grids.

Microsoft reported approximately 20.3 million metric tons of carbon dioxide-equivalent emissions across its operations and supply chain, up from 16.2 million tons in fiscal 2024 and nearly 58% above its 2020 baseline. Electricity consumption increased by 24% during the year as the company built the computing capacity required for its cloud and AI businesses. Regardless, Microsoft says it remains committed to becoming carbon-negative, water-positive, and zero-waste by 2030. It also reported meeting its 2025 renewable-electricity target, replenishing more water than it withdrew globally, and exceeding several waste-recovery targets

The report’s foreword, written by Microsoft Vice Chair and President Brad Smith and Chief Sustainability Officer Melanie Nakagawa, focused heavily on the collision between the company’s headline sustainability goals and the realities of AI. Microsoft established the goals in 2020, a few years before the current scale of AI’s capabilities and the corresponding high environmental demands began to manifest.

While AI is inarguably a world-changing technological revolution, it is raising serious environmental concerns that begin right at the raw material sourcing and the complex semiconductor fabrication stages. The impact continues even after the processors have been compiled into supercomputers in massive data centers, with issues related to land use, energy consumption, noise pollution, and water consumption. Residents are increasingly opposing the building of these data centers in their communities due to these issues.

Microsoft is exposed at nearly every point of the AI chain. It procures servers and custom AI chips; owns and operates a massive, global network of over 300 data centers across 34 countries that powers the Azure cloud platform; and supplies the computing infrastructure behind products such as Copilot and its partnership with OpenAI. Scope 3 emissions from construction, purchased hardware, suppliers, and other value-chain activities remain the largest part of its footprint. Meanwhile, electricity-related Scope 2 emissions grew from nearly 2% of the total in 2024 to 13% in 2025.

Microsoft acknowledges that environmental solutions are not expanding as quickly as AI infrastructure. “This tension is real,” the foreword states. “It is forcing sharper questions: Where do we need to move faster, invest differently, or rethink our approach?” The company argues that the answer is not to retreat from AI, but to combine carbon-free electricity, carbon removal, sustainable fuels, lower-carbon construction materials, hardware reuse, and efficiency improvements into a single portfolio rather than treating each environmental target separately.

Its decision to stop buying non-additional, unbundled renewable energy certificates forms part of that change. These certificates can allow a company to claim renewable electricity already being generated elsewhere. Microsoft says it will instead prioritize longer-term agreements that help add additional carbon-free generating capacity to the grid, even though doing so will increase its reported emissions in the near term. Its renewable-energy agreements now cover up to 40 GW across 26 countries, with approximately 19 GW operational.

The company is also modifying the data centers themselves. It introduced a closed-loop liquid-cooling design that CEO Satya Nadella says enables AI data centers to use about as much water annually as a restaurant. Microsoft is experimenting with microfluidic channels etched into silicon, zonal cooling that reserves colder liquid for the hottest equipment, and lower-carbon concrete, steel, and mass timber for its construction. These efforts have not exactly quelled anti-data-center sentiment around its data centers. The company faced protests over a planned facility near Granger, Indiana, while residents living near its $7.3 billion Fairwater AI complex in Wisconsin have filed a lawsuit alleging persistent noise, dust, traffic, and light pollution.

Away from carbon, the report records clearer progress. Microsoft replenished 14.2 million cubic meters of water, exceeding its global withdrawals for the first time, and reduced average data center water-use effectiveness by 25% from its 2022 baseline. It achieved a 92% reuse and recycling rate for retired cloud hardware, diverted 90.5% of construction and demolition waste from disposal, and reduced single-use plastics in primary product packaging to 0.07%. It also legally protected 16,266 acres of land, approximately 36% more than the land estimated to be occupied by its operations.

The report is equally candid about where Microsoft is falling behind. The company's most important commitment—becoming carbon-negative by 2030 — is moving further away rather than closer. Total greenhouse-gas emissions climbed 25% year over year and now sit roughly 58% above the company's 2020 baseline, largely because AI infrastructure is expanding faster than its decarbonization efforts can offset. Scope 2 emissions also jumped sharply, rising from nearly 2% of Microsoft's footprint in FY24 to 13% in FY25 as electricity demand from new data centers surged. While Scope 3 emissions remain the company's largest source of carbon pollution, the report says the growing contribution from purchased electricity underscores how increasingly difficult it is to power AI infrastructure with clean energy alone.

The Internet is buzzing over news that 19-year-old Estonian "hacker" Peter Stokes got nabbed by the authorities and extradited to the U.S. on digital crime charges, mostly thanks to Microsoft Windows' built-in telemetry. The FBI seemingly subpoenaed Microsoft, which coughed up telemetry logs that contained both Stokes' GDID (Global Device Identifier) and websites he visited using his main Windows machine.

The existence of GDID isn't new by itself, as Windows telemetry's data collection has been extensively analyzed and reported on. It's also been known, and publicly explained by Microsoft, that the extended telemetry modes (Full/Optional instead of Required/Basic) can upload lists of URLs analyzed by SmartScreen and Defender, together with the GDID. In fact, using the Edge browser in this setup can even send every visited URL. The court documents do not reveal which exact mechanism triggered the telemetry upload, though.

This data collection has long been the source of heated debate and general public disgust. Even though the data is genuinely useful and necessary for debugging (by Microsoft or systems administrators in enterprise environments), the fact that it comes enabled by default in Windows Home and Professional editions is questionable. The fact that those versions don't have a simple, user-facing "Off" switch to fully disable telemetry also adds insult to injury.

The Peter Stokes arrest appears to be the first public case where these Windows GDIDs were both used as a tracking identifier and contained telemetry data including some of the URLs the defendant visited. The case also prompted a renewed analysis of the GDID by a security researcher that you might want to look into. From what we can ascertain, it's likely Stokes had his Windows telemetry set to Optional/Full, as Required/Basic doesn't appear to transmit URLs by default.

Using the telemetry GDID, the FBI easily connected the dashing rogue to his ngrokaccount, because he used that tool in the same session in which he accessed his Facebook and Snapchat accounts. The agents also established a link between travel records, a New York IP address, and a rental at the Empire Hotel, likely facilitated by the photos Stokes posted of his hotel room. The criminal mastermind was equally sneaky (read: not) in his visit to Thailand.

As many hackers do, he enjoyed some time off playing an obscure game, in this case Ubisoft's Growtopia, shortly before accessing his Apple logins, as well as the aforementioned Facebook and Snapchat logins over the following weeks. Besides Microsoft, Google and Apple also collaborated on the hunting effort, with Google linking Stokes' phishing phone number to the same exact IP address and date where he created the ngrok account. Ever the stealthy craftsman, Stokes had created the ngrok account using the same GMail address connected to a second phone number where he made phishing calls from.

While it's easy and arguably quite necessary to hoist pitchforks at Microsoft for collecting detailed information about billions of computers by default, security professionals will be quick to remind users that Windows' telemetry is merely one of the many ways to track a user. Even if not by malice, a lot of software simply requires GDID-like identifiers for things like tracking usage, subscription and licensing limitations, activation requests, and hardware detection. And every company behind such software can be subpoenaed by authorities, as exemplified in Stokes' case by Microsoft, Google, Apple, ngrok, and others. Even privacy-oriented services like Proton are careful enough to describe what they can and cannot reveal to authorities under a court order.

If you're wondering the steps Stokes took to cover his tracks, though, you'd be looking at a small list. He did route his connections through a VPN hosted at servers from Tzulo along with the developer-oriented ngrok tunneling service and teleport.sh. Unfortunately, the modern digital world allows for many forms of identification, and hiding one's source IP address is merely one of them.

Using a VPN is recommended for digital anonymity, but it's merely the first of many necessary steps and can even backfire when not set up carefully. If misconfigured, a VPN may allow certain applications and operating system features to talk to the outside world using the original IP instead of the hidden one. Plus, the VPN will not stop the operating system or any application from sending out identifying information to begin with.

Perhaps more worryingly still, modern-day device and user fingerprinting is far more insidious and hard to counter. For example, plain web browsers are notorious leakers of personal information, as data-harvesting companies can weaponize features like TLS levels, HTML5 Canvas functionality, the fonts list, and even Widevine DRM in a combination that uniquely identifies a visitor. Stokes now has plenty of time to read up on the EFF's surveillance self-defense guides and get acquainted with the scripts at the Privacy Is Sexy website.

Controversy due to AI data center buildouts generally centers around their massive power usage and draining of local water reserves, but noise is a third torment to nearby residents, and one that is arguably much harder to correct. Residents of Sturtevant, slightly south of Milwaukee, Wisconsin, filed a class-action lawsuit against Microsoft due to the excessive noise produced by the company's Fairwater facility. CEO Satya Nadella described the project as "the world's most powerful AI data center," projected to generate 865,000 tokens per second and have a final bill of $7.3 billion.

The Sturtevant residents live just 1.5 miles (2.4 km) from the facility. The lawsuit was filed by three citizens and represents the households within this distance, reportedly amounting to over 1,000 homes, including areas in Mount Pleasant. The filing describes the noise situation as "not only excessive, but consistent and pervasive," and claims Microsoft did not "implement adequate acoustic barriers, shields, or walls that absorb, mitigate, and/or prevent the escape of noise, thereby resulting in the offsite emission of excessive noise beyond its property." One resident claims he had to change his shift at work to be able to sleep at all.

Nearby residents also complain about excessive dust and traffic stemming from the construction work, as well as light pollution. One resident says he often can't see his house coming into town, while another claims that before the data center arrived, the sky was dark and full of stars, now mostly gone due to the bright lights.

To some credit, Microsoft appears to have been trying to improve the situation, judging by a fairly detailed blog post on its community website. On June 18, the company said its engineers applied several measures that "fully resolved the issue," and that it would apply additional mitigations over the following months, including "additional sound reduction components."

The backstory in the post says the firm was aware of the issue back in April, and "[it] did not expect the tonal quality of the sound to travel as far as it has," attributing the decibels to cooling fans operating at too-high speeds, now purportedly corrected. In an earlier project update about the data center, Microsoft says it would have street sweepers working 10 hours a day and limit construction to hours between 6am and 10pm.

The lawsuit was filed on July 1, indicating that either the issue isn't resolved, or that the particularly short distance between Fairwater 1 and Sturtevant might make for a conundrum that's exceedingly tricky to solve. Some Mount Pleasant residents even live across the street from the campus, too. Before the Fairwater data center arrived, the land was already zoned for heavy industrial use back in 2017 to Foxconn, a status that carried over to Microsoft upon purchase. Notably, Wisconsin's "direct legislation" apparently does not allow amending or repealing existing ordinances, and is only available to cities, not towns.

The Department of Justice, with the help of the FBI and Finland's National Bureau of Investigation, has arrested a teenager it says is part of Scattered Spider. 19-year-old Peter Stokes is a dual U.S.-Estonian citizen who was trying to board a flight to Japan from Helsinki, when law enforcement caught up with him. Microsoft's GDID also played a part in the Stokes being apprehended. The accused is now awaiting trial, having been charged with conspiracy, cyber intrusion, and fraud.

Scattered Spider is one of the biggest cybercrime syndicates on the planet, having extorted over $100 million in ransom payments, according to the DOJ. The group also operates under the names Octo Tempest, UNC3944, and Oktapus, and is renowned for its social engineering tactics. As such, the main criminal complaint against Stokes stems from a May 2025 attack on a luxury jewelry dealer based in the United States.

The attackers apparently called the company's IT helpdesk using Google Voice, posing as employees. They were able to convince the help desk into resetting their credentials, which allowed them to infiltrate three accounts, two of which had admin privileges. From there, the group, allegedly including Stokes, stole important data and held the jeweler at ransom, demanding an $8 million payment in crypto.

The company ultimately regained access to their infrastructure and avoided paying the ransom, but the operational disruption still caused a purported $2 million in losses. This served as the spark that led to Stokes' eventual arrest in Helsinki, as the prosecutors slowly followed the paper and digital trail laid by the attackers. Microsoft played a key role in the process by providing GDID data to the FBI to help them apprehend the alleged criminal.

GDID stands for Global Device Identifier; it's a unique identifier assigned to every Windows install that tracks device-specific telemetry. It's the reason why sometimes changing a major component in your PC can revoke your Windows license. Anyhow, the court documents from the case reveal that Stokes used Windows, from which investigators were able to link his physical hardware to specific internet activity and locations.

From what we can tell, GDID pretty much had a comprehensive report on Stokes ready before the prosecution even built its case and it was only a matter of connecting the dots. Stokes' web activity, videogame history, IP addresses, tool usage (including Ngrok), Azure status, and more were logged with timestamps, and were provided to the investigators by Microsoft.

Of course, this raises questions over just how granular and potentially invasive Microsoft's telemetry can be. In this case, it was used to arrest an alleged hacker, but what if someone else, someone with malign intentions, were to get access to all this data instead? Tech-savvy consumers have complained about Windows' excessive telemetry for a long time; the whole debloating culture is a byproduct of this precedent, but GDID is not something you can remove or disable with the click of a button.

Nevertheless, Stokes was carrying two hard drives full of incriminating evidence with him when boarding his flight to Japan, so that helped, too. His real identity has actually been known since 2024, but since he was a minor living across Estonia and the UAE at the time, he could only be monitored until the time was right. Following the arrest, Stokes was extradited to the U.S., where he appeared in front of a federal court in Chicago for the first time on June 30, 2026, and he remains in custody.

Microsoft is further shrinking its Surface lineup of flagship Windows devices. According to a report from Windows Central, citing "sources that are familiar with Microsoft's hardware roadmap," the company is no longer manufacturing its budget-focused Surface Go 4 and Surface Laptop Go 3, and doesn't plan to follow them up with new devices.

Over the last few years, Microsoft has discontinued many of its most interesting designs, including the powerhouse Surface Laptop Studio, the Surface Studio desktop, Surface Headphones and Earbuds, the Surface Book, and the Surface Duo smartphone. The entire line has been simplified down to the Surface Pro and Surface Laptop, with the Surface Laptop Ultra coming later this year with Nvidia's RTX Spark. (Microsoft's RTX Spark Dev Box is also Surface-branded, but isn't consumer-facing).

But the Surface Pro and Surface Laptop each come in more sizes than ever. The Surface Pro comes in 12- and 13-inch sizes. That 12-incher, starting at $849 with Snapdragon X Plus, could arguably be a replacement for the Surface Go, except that it's far more expensive than the Surface Go 4 was priced at $579. On the Laptop side, there are 13-, 13.8-, and 15-inch systems, with the smallest using the same chip at $949.99 that could be considered the entry-level option.

It's possible that you may find these systems in stock at some retailers, but they're hard to find. The Surface Go 4's last iteration was only ever part of Microsoft's Surface for Business line and was seemingly popular with commercial customers that needed a device for fieldwork.

Microsoft didn't immediately return Tom's Hardware's request for comment. A representative pointed Windows Central to Microsoft's Surface website, which doesn't list either the Surface Go or Surface Laptop Go, but just a simplified lineup of the Surface Laptop, Surface Pro, and upcoming Surface Laptop Ultra and the Dev Box.

Windows Central's sources suggested that the decision to end these device lines was made prior to memory price hikes due to component shortages. Either way, with just two main device lines, Surface, which used to be a playground for form factors under former head Panos Panay, has been extremely simplified and far less experimental —if you could call it that at all.

Microsoft has extended its free consumer Windows 10 Extended Security Updates (ESU) program by a year, pushing the cutoff for critical security patches to October 14th, 2027. The change came without any actual announcement, appearing instead in an edit to Microsoft's Windows 10 ESU support page. Devices already enrolled roll over to the new date automatically, with no action required from users. The program was originally set to expire on October 12th, 2026, one year after Windows 10 reached end of support on October 14th, 2025. With the extension, however, users who can’t or won’t move to Windows 11 now have a second free year of patches.

Speaking to BleepingComputer, Microsoft said that the change reflects “our ongoing commitment to helping customers stay secure during the transition,” adding that the extra year gives users “more time and flexibility to find the best PC for their needs while keeping them protected.”

In terms of enrollment, nothing has changed: users can enroll for free by syncing their PC settings to a Microsoft account through Windows Backup, by redeeming 1,000 Microsoft Rewards points, or with a one-time payment of $30. Users in Europe can enroll at no cost simply by signing in with a Microsoft account, a concession Microsoft made after pushback from consumer advocacy groups over its original requirements. A single ESU license covers up to 10 devices tied to the same account.

Microsoft's consumer program is limited to personal devices, however. Systems joined to an Active Directory domain or to Microsoft Entra, or managed through Mobile Device Management, are corporate-controlled and therefore not eligible. Personally owned machines that are only Entra-registered — e.g., a work account added to a personal device, rather than the device itself being organization-owned — still qualify.

Windows 11 has surged to roughly 73% of desktop share against Windows 10’s 26%, so it’s clear that the broader migration is working. Those users still on Windows 10 are increasingly the difficult edge cases, with around 400 million active PCs unable to officially upgrade to Windows 11 because their hardware fails its TPM 2.0, Secure Boot, or supported-processor requirements.

Users of those machines face a choice of remaining on unsupported Windows 10 or upgrading their hardware outright. But thanks to the AI-driven memory shortage, doing so has never been more expensive: DRAM contract prices have roughly doubled since early last year (and are expected to double again) as Samsung, SK Hynix, and Micron divert wafer capacity toward high-bandwidth memory for AI accelerators.

IDC expects PCs, tablets, and smartphone prices to rise 10% to 20% through the end of 2026, with memory now accounting for a far larger share of a system's bill of materials than it did two years ago.

For users unwilling to buy new hardware at current prices, some third-party options remain. The security firm 0patch has pledged to provide unofficial Windows 10 micropatches through 2030, and Linux migration efforts such as the End of 10 initiative continue to court holdouts whose machines can’t run Windows 11.

People seem all too ready to anthropomorphize LLMs and AI chatbots like ChatGPT, Claude, and Gemini. Some humans even admit to ‘relationships’ with one or more of the various examples of machine intelligence. To illustrate how flawed this instinct could be, a Microsoft AI researcher built a tiny neural network inside Age of Empires II using goats, grass, and bridges. Adrian de Wynter shared his work in a paper dubbed If LLMs have human-like attributes, then so does Age of Empires II. The Microsoft researcher, based at the University of York, also talked to 404 Media recently about how he likes to turn absurdism up to 11 to make a point.

In the research paper, De Wynter doesn’t make the argument that LLMs do or do not actually have generalized anthropomorphic attributes. Instead, he illustrates that the AoEII goats can also power the kinds of models that lay behind today's most popular chatbots. That hammers home the argument that “in no case is a machine’s activity to be interpreted in terms of higher cognitive processes, if it can be fairly interpreted in terms of processes which stand lower in the scale of cognitive evolution and development.”

De Wynter also raises the well-known concept of confirmation bias. Those looking for human traits in tech like chatbots will tend to find them, he proposes. However, the big contrast between the absurdist goat example and the commercial LLM chatbot is the way people interact with them, the interface that makes the likes of Claude ‘conversation friendly.’ De Wynter’s research indicates that anthropomizing LLMs is a common trend in computer science papers. From 337 such papers De Wynter looked at, published in the last two years, he says that 57% assumed that LLMs could have human-like traits. This basic assumption could color the research, testing, and, of course, conclusions of these papers.

So, how did the Microsoft AI researcher build the goaty AoEII LLM? Well, he didn’t quite go as far as developing a full-blown LLM. Instead, De Wynter thought it sufficient to use AoEII’s scenario editor to build a working NAND gate, with a 1-bit perceptron, where the goats act as bits. This crude perceptron and the circuit to train it in-game are enough to demonstrate that the simplest building block of a modern neural network could be made this way. And if you think it is absurd that AoEII goats can embody consciousness, then it should be equally absurd to regard any of the well-known chatbots as anything more.

Companies behind the AI boom aren’t discouraging people from anthropomorphizing their wares. In many ways, they might benefit from these human perceptions. Chatbots they deploy are trained with natural language and use techniques to mimic the shape and tone of natural conversation. This makes it easy for users to project personality, emotion, or even consciousness onto them. Top AI company execs have leaned into the perception of their customers, publicly entertaining the idea that their systems could or might be exhibiting signs of consciousness. In his 404 Media interview, De Wynter also highlighted research indicating that people buy more products when they can empathize with them, and that includes AI/chatbot/LLM subscriptions.

Microsoft has quietly updated its consumer-oriented Surface lineup with Qualcomm's Snapdragon X2 chips. The Surface line, often announced with its own event, has been detailed in a blog post by corporate vice president for Surface Brett Ostrum. All three systems are on sale from Microsoft's website today. The 13-inch Surface Pro starts at $1,499, while the Surface Laptop line starts at $1,599 for the 13.8-inch machine and $1,699 for the 15-incher.

The new systems aren't particularly different on the outside, coming in platinum, black, and dune colors that Microsoft has used before. The 13.8-inch Laptop option will have one brand new color: a standout jade green.

All of the systems will have options between 10-core Snapdragon X2 Plus or 12-core Snapdragon X2 Elite processors, and range from 16GB to 64GB of RAM (some of the Surface for Business models Microsoft announced with Intel's Panther Lake processors start with just 8GB of memory).

The Surface Pro starts with 256GB of storage and goes up to 1TB, while the Surface Laptop 13.8 goes up to 2TB. The 15-inch Laptop starts at 512GB and goes up to 2TB.

While all of the devices use Microsoft's PixelSense Flow displays with variable refresh up to 120 Hz, but only the Surface Pro includes both LCD and OLED options. The touchpad on the Surface Laptop has new haptic feedback technologies, working with Windows for small pieces of feedback throughout Windows. On the Surface Pro, you can get similar feedback with the Slim Pen.

The Surface Laptops have 1080p cameras that Microsoft boasts are best in class with No. 1 rankings from DXOMARK. That judges color, texture, noise, and artifacts, but doesn't seem to consider resolution. Some premium PCs have 4K or higher webcams now.

Microsoft is promising up to 20 hours of battery life on the 13.8-inch Surface Laptop when measured with local video playback. That drops to 16 hours when tested with web usage. We'll have to see how that holds up in our labs.

All three devices maintain Microsoft's Surface Connect port, and come with 39W chargers in North America. Microsoft notes that these aren't available in Europe, the Middle East, and Africa, so buyers in those locales will need to bring their own.

Business-focused models using Intel chips will launch on July 14.

Update, June 16, 10:26 a.m.: Microsoft shared a change to their spec sheet. The 13.8-inch Surface Laptop Pro will start with 512GB in the United States.

Microsoft CEO Satya Nadella has claimed that the company's newest generation of AI data centers uses so little water that its annual consumption is roughly equivalent to that of a single restaurant. Speaking at Microsoft Build 2026 on June 2, Nadella made the claim while outlining Microsoft's "Community-First AI Infrastructure" strategy aimed to address growing concerns about the impact of data centers on local communities.

"The cooling loop is filled once, and the data center can operate effectively with zero water consumption," Nadella said during his keynote, while describing how new architectures are redefining data center water consumption. "The daily water usage over the course of an entire year is roughly equivalent to what a single restaurant would use."

While Nadella did not specify the size or type of restaurant used for comparison, the claim is striking given the scale of modern hyperscale AI facilities. Industry estimates suggest that restaurants can consume anywhere from several hundred thousand to more than a million gallons of water annually, while 100MW+ hyperscale data centers have historically required tens to hundreds of millions of gallons of water annually for cooling, depending on climate, workload, and cooling design.

Nadella's statement — not the first time Microsoft has used a restaurant comparison — is based on Microsoft's new closed-loop liquid-cooling architecture. Traditional data centers often rely on evaporative cooling systems that continuously consume water to remove heat generated by servers and networking equipment. In Microsoft’s new approach, over 90% of the facility's cooling relies on a closed-loop liquid cooling system that is filled during construction and then continuously recirculates the same water rather than constantly consuming a fresh supply. The remaining portion of the cooling system primarily relies on outside air and only uses additional water in the hottest conditions.

The system works by circulating cooled water through heat exchangers connected to the AI hardware. As the water absorbs heat, it is routed to a massive chiller plant where large cooling fans dissipate the heat before the water is recirculated back through the facility. Because the water remains in the cooling loop rather than evaporating and being discarded, ongoing water consumption is dramatically reduced. This can translate to savings of multiple billions of gallons of water across Microsoft’s data centers.

The current reality is much more subdued. The new cooling design and its water savings are currently implemented only at Microsoft's Fairwater AI data center campus in Mount Pleasant, Wisconsin. On the other hand, the company has stated that the new cooling approach will become the standard design for future AI-focused data centers. Microsoft has confirmed that multiple identical Fairwater facilities are already under construction elsewhere in the United States as part of a broader expansion of its AI infrastructure footprint. The rollout forms part of Microsoft's longer-term goal of becoming water positive by 2030, meaning the company aims to replenish more water than it consumes globally, as part of its community-first strategy.

While the new cooling design will substantially reduce water use in future facilities, critics note that Microsoft already operates a vast global network of data centers. Azure now spans more than 500 facilities across 80 regions worldwide, many of which were built before Fairwater's cooling architecture was introduced.

Microsoft has not announced a comprehensive retrofit program that would convert existing data centers to the new closed-loop design. As a result, the "restaurant-level" water footprint applies primarily to newly built facilities that use the Fairwater blueprint, rather than to Microsoft's entire global infrastructure estate.

The debate reflects a broader challenge facing the AI industry. Companies, including Microsoft and Google, are racing to build ever-larger AI clusters while facing mounting public pressure to reduce their environmental impact, as communities grow increasingly disapproving of large-scale AI infrastructure projects. Google recently announced a series of water stewardship commitments of its own, including a pledge to replenish more water than its data centers consume by 2030.

Microsoft has unveiled Project Solara, a chip-to-cloud platform designed to power a new generation of "agent-first" enterprise devices — hardware designed to run AI agents instead of traditional apps. Announced at the Microsoft Build 2026 Developer Conference on the 2nd of June, the platform, developed by Microsoft's Applied Sciences Group, features a lightweight edge OS called the Microsoft Device Ecosystem Platform (MDEP). Interestingly, the OS is built on the Android Open Source Project (AOSP) rather than Windows.

MDEP is paired with Azure-hosted agent services and persistent cloud-based state, meaning devices act as interfaces to AI agents running across Microsoft's cloud infrastructure rather than as fully self-contained computers. Together, the software stack forms what Microsoft describes as a chip-to-cloud architecture for enterprise AI devices, combining cloud-hosted agents with centralized security, management, and orchestration capabilities.

"The 'operating system' is liminal, transcending the device and the cloud. The system brings a lightweight window to the edge, where the agent manifests and where the state, via Azure, can encompass a constellation of specialized devices,” explained Steven Bathiche, Microsoft's Corporate Vice President and Technical Fellow in the Applied Sciences Group.

To populate that ecosystem with hardware, Microsoft has partnered with Qualcomm and MediaTek as its first silicon partners — Qualcomm for portable and wearable form factors and MediaTek for stationary devices. The company has no plans to manufacture end products itself. Instead, the company is releasing reference designs for OEMs to build from, alongside an "approved chipsets" requirement that gives Microsoft certification-level control over which hardware qualifies for the platform, similar to Google's GMS certification model for Android.

To demonstrate Solara, Microsoft unveiled two concept reference designs built on the platform. A stationary desk-mounted AI hub built around MediaTek IoT silicon and a wearable AI badge powered by Qualcomm hardware. The desktop companion features a display, a camera, a UWB (ultra-wideband) presence sensor that handles automatic login and lock, dual far-field mics, and two USB-C ports. Connected to an external display, the device can double as a Windows 365 cloud PC client.

Meanwhile, the wearable badge is equipped with a touchscreen, Hello for Business fingerprint sensor, far-field high-SNR microphone array, side-facing camera, and 5G, WiFi, Bluetooth, and GNSS connectivity — targeting front-line workers such as nurses, retail staff, and field workers. Microsoft confirmed that both devices are intended as reference designs for OEM partners rather than retail products.

Central to the platform is what Microsoft calls just-in-time UI — an adaptive interface layer that allows a single agent to render appropriately across different screen sizes and input modalities without requiring developers to rebuild the experience for each device. Microsoft positions this on a spectrum between conventional responsive design and fully generative UI, where AI constructs interfaces dynamically with no predefined structure; Solara currently targets the middle ground, prioritizing consistency while avoiding per-device redesign overhead.

“The same agent can render a custom experience on multiple screen sizes and modalities with little or no additional work from the developer. For us, that is the first proof point: a path to specialized devices without requiring developers to rebuild the experience from scratch each time,” said Bathiche.

Another notable aspect of Solara is Microsoft's decision to build MDEP on Android rather than Windows. AOSP scales naturally to the lightweight, constrained hardware that wearables and embedded devices run on — something Windows, with its memory and processing overhead, was never designed to do. It also sidesteps the application compatibility expectations that come with Windows. Because Solara devices are built around cloud-hosted agents rather than traditional software, Microsoft can optimize the platform for dedicated AI hardware without carrying decades of legacy PC baggage.

To manage multiple agents running simultaneously, Microsoft is also working on an agent dispatcher and agent task manager — components that automatically surface or activate the right agent based on context, rather than requiring users to launch each one manually. Neither component is shipping yet. Early agent integrations include Dragon Copilot for healthcare workflows and GitHub Copilot for developer task tracking — both exploring how persistent, context-aware agents behave differently on dedicated hardware than they do inside a browser or IDE.

The platform appears to be aimed at enterprise buyers in retail, healthcare, and field service sectors, where dedicated agent hardware makes more sense than repurposing a smartphone. Microsoft has already lined up pilots with Best Buy, CVS Health, Levi's, Target, and AccuWeather, with broader OEM deployment targeted across healthcare, hospitality, financial services, legal, and industrial verticals.

The Great Exploitation of 2026 continues apace, with security vulnerabilities being published at an alarming rate, and more often than not, being exploited in the wild almost before anyone has any time to react. Today, Microsoft in the unfortunate limelight, with a 9.8-rated remote execution vulnerability affecting Windows Server domain controllers (DC), versions 2012 to current. The exploit and its explanation are simple: any unauthenticated user in the same network can send a malformed UDP packet to a DC and potentially get system access — no previous access required. Even if an attacker doesn't go that far, it's trivial for anyone to force the DC to reboot, creating potential denial-of-service scenarios.

The vulnerability is CVE-2026-41089, and it's mercifully not a zero-day this time. The vulnerable service is Netlogon, and there's apparently no mitigation, with the only solution being to patch the affected systems. The patch itself arrives in the May 12 Patch Tuesday, but there's a fair chance that a lot of DCs remain unpatched, particularly but not only older versions. Systems administrators might also find specific patch links and remediation scripts handy.

If an attacker can finagle this vulnerability to get System-level of access to domain controllers, the consequences are pretty up to the imagination. The malfeasant can create any number of accounts with all sorts of access levels, including Kerberos Ticket-Granting Tickets, enabling access to most all data across the entire domain. Since DCs often operate as part of a larger network in medium-to-large enterprises, just one vulnerable machine is enough to make the entire network insecure. Cybersecurity boffins recommend that administrators treat this as a worm-style threat and patch all their linked DCs at once, to avoid playing a game of whack-a-mole with high odds for the moles.

Microsoft stated that the vulnerability was not made public at the time, and that no ongoing attacks were using it, but the situation has changed since the discovery date, as recent reports have confirmed that it's now being exploited in the wild. As far as proof-of-concept goes, there's a GitHub repository with some sample code that forces the LSASS service to crash after a minute or so.

The technical details are simple and somewhat facepalm-inducing. The crafted network packet that triggers the vulnerability doesn't have anything all that fancy about it; it just contains one field that's larger than it should be. Data serialization logic in the Netlogon service combines the attacker-supplied data with the server's hostname, resulting in a classic buffer overflow — the most straightforward type of vulnerability.

Microsoft has been in security news quite often recently, mostly thanks to its ongoing spat with security researcher Chaotic Eclipse (aka Nightmare Eclipse), who published a heap of zero-days exploits after apparent negotiations with the company broke down. The situation is unclear, but has escalated to the point where Microsoft is now threatening Eclipse with legal action.

Microsoft has just announced the Surface Laptop Ultra, powered by Nvidia’s latest RTX Spark Superchip for laptops and desktop PCs, at Computex 2026. Claimed to be the most powerful Surface laptop device yet, the new Surface Laptop Ultra is essentially Microsoft’s answer to the current-gen MacBook Pro. The company has positioned the notebook primarily for AI development but also for creators and developers who seek powerful performance in a portable package.

Made in collaboration with Nvidia, the Surface Laptop Ultra will be one of the first devices featuring the RTX Spark Superchip. Nvidia’s new Windows on Arm platform is claimed to be more powerful and capable than any other on the market featuring 20 Arm CPU cores, a Blackwell GPU with 6144 CUDA cores, 128GB of unified LPDDR5X RAM, and up to 300 GB/s of memory bandwidth. The CPU and GPU are interconnected using NVLink C2C, while the large pool of unified memory can be dynamically allocated between the CPU and GPU. Offering up to 1 petaflop of AI compute, the RTX Spark is capable of running up to 120B parameter models locally.

As for the device itself, the new Surface Laptop Ultra will feature a 15-inch mini-LED PixelSense Ultra display with a resolution of 2880 x 1920, offering 262 pixels per inch. Microsoft claims that it can reach up to 2,000 nits of peak HDR brightness, which should be excellent for creators on the go. It also features a large haptic touchpad along with a good selection of ports, including HDMI, USB-C, USB-A, SD card reader, and a headphone jack. Weighing under just 4.5 pounds, it will be available in Platinum (silver) and Nightfall (black) color options. Microsoft claims all-day battery life on the laptop and an optimized internal layout with dual-fans to maintain proper thermals for sustained performance.

The new Microsoft Surface Laptop Ultra is expected to arrive later this year, with final pricing and availability to be announced closer to launch. With the ongoing RAM-apocalypse, it seems like a smart move by Microsoft to adopt Nvidia’s new Arm platform with unified memory. Beyond reducing the need for large pools of dedicated system and graphics memory, the architecture should also help improve efficiency and enable larger AI models to run locally without relying heavily on cloud resources. If Nvidia's performance claims hold up in real-world workloads, the Surface Laptop Ultra could emerge as one of the most compelling Windows alternatives to Apple's MacBook Pro for AI developers, content creators, and power users alike.

Some people are pretty excited that we are on the cusp of a new Windows PC era ignited by technologies from the mighty Nvidia and Arm. Our article on the teasers for the rumored wave of new N1X laptops makes it clear that companies like Nvidia and Microsoft are excited, at least. However, Microsoft veteran Steven Sinofsky is here to remind folks that it has been done before, with a similar level of simmering excitement, when the first Surface hybrid PC ran on Nvidia Tegra Arm silicon back in 2010/11.

Sinofsky shares some interesting media in the above embedded Tweet. There’s a slide from the promotional deck, shown at CES 2011, where the “strong partnerships” behind this Windows on Arm thrust would surely lead to unstoppable momentum. That’s what you may have believed if you swallowed the effusive presentations at the time.

This prior push for Windows on Arm didn’t exclusively support Nvidia Tegra. Sinofsky points out that Windows 8 on Arm also supported Qualcomm and TI processors. However, we saw Microsoft switch to a Qualcomm partnership focus in the years following. That collaborative effort also didn’t break through, even when the much-lauded Nuvia Oryon architecture arrived with the Snapdragon X family.

The former-President of the Windows Division (July 2009 to Nov 2012) also shares a video that should be preserved for posterity. Check it out to witness “an old school Windows Phone video of the first time Windows ran on Nvidia Tegra Arm using the desktop compositor which was a BIG deal. Sept 2010.”

The video appears to show a Tegra dev kit running Windows 7, with multi-window and Start menu manipulation, which looks rather lethargic to my eyes. However, that was a “first” achievement several months before CES 2011, and it would be over a year and a half until the Surface RT shipped to customers (October 2012).

Will the latest Nvidia and Arm thrust into Windows work out better? We’ll probably have a much better idea about that in the coming days at Computex. Sinofsky replies to some comments on his Tweet to say the result could be “a reliable platform for graphics compute,” using Windows. Pricing is also obviously going to be a sticky issue for any new platform looking to attract buyers right now.

Microsoft has uncovered an ongoing cryptojacking campaign that used SEO poisoning and, in some observed cases, AI chatbot-generated software recommendations to lure users into downloading GPU mining malware disguised as popular PC utilities. According to a detailed threat report published Tuesday by Microsoft Defender Experts and the Microsoft Defender Security Research Team, the operation specifically targeted users who likely own high-performance graphics cards, including gamers, hardware enthusiasts, AI users, and overclockers.

The campaign impersonated widely used utilities such as CrystalDiskInfo, HWMonitor, Display Driver Uninstaller (DDU), FurMark, K-Lite Codec Pack, and PDFgear. Victims searching for the software on traditional search engines — and, in some cases, via AI chatbot recommendations — were reportedly redirected to attacker-controlled download pages hosting malicious ZIP archives.

Microsoft says the attackers appear less interested in maximizing infection volume and more focused on compromising systems with powerful discrete GPUs suitable for profitable cryptocurrency mining. Once installed, the malware deployed persistent remote-access software using the legitimate ScreenConnect remote-management tool before silently loading GPU mining payloads such as lolMiner, gminer, and SRBMiner-MULTI.

The attack chain relied heavily on stealth techniques typically associated with more advanced malware operations. The downloaded archives bundled legitimate software installers alongside malicious DLLs that were automatically loaded through DLL sideloading. From there, the malware established six separate persistence mechanisms, added Microsoft Defender exclusions, checked for virtual machines and security-analysis tools, and used process hollowing to inject mining code into trusted Microsoft-signed .NET utilities such as MSBuild.exe, InstallUtil.exe, and RegAsm.exe.

Perhaps the most unusual aspect of the campaign, however, is Microsoft’s observation that some malicious domains may have surfaced through interactions with AI chatbots. According to the company, users requesting software download recommendations from large language model (LLM)-based assistants were, in some cases, presented with links to attacker-controlled domains embedded in generated responses. Microsoft stressed that the example was illustrative and “does not indicate a systemic issue with any specific AI service,” but noted that the activity appears consistent with emerging AI-assisted search-poisoning techniques.

According to Microsoft’s analysis, the operation has been active since at least March 2026 and involved more than 150 malicious domains masquerading as trusted utility-download portals. Many of the downloads were hosted on subdomains of gleeze.com, infrastructure linked to the Dynu dynamic DNS service, which has frequently been used in past phishing and malware campaigns.

The initial infection process itself was deceptively simple. Victims downloaded ZIP archives containing both the legitimate utility executable and a malicious DLL named autorun.dll. When the legitimate application launched, Windows automatically loaded the malicious DLL from the same directory via DLL sideloading — a long-standing Windows abuse technique that requires no software exploit and often produces no visible signs of compromise.

From there, the malware silently installed ScreenConnect, a legitimate enterprise remote-management platform also known as ConnectWise Control. Microsoft emphasized that ScreenConnect itself is not malicious, but rather is being abused by threat actors in the same way attackers increasingly misuse legitimate remote monitoring and management (RMM) tools to evade security scrutiny.

Once remote access was established, attackers deployed a binary called SimpleRunPE.exe, which Microsoft believes may partially derive from a publicly available GitHub proof-of-concept process hollowing project. The malware copied itself into hidden Windows directories as RuntimeHost.exe, created scheduled tasks and startup entries for persistence, and repeatedly re-added Microsoft Defender exclusions even if users or administrators attempted to remove them.

The malware also appeared engineered specifically to avoid detection by performance-conscious PC users. Microsoft says the miner monitored GPU utilization, system idle time, gaming activity, and streaming workloads, shutting down mining operations whenever heavy GPU activity was detected. In practice, this likely reduced obvious warning signs such as sudden frame-rate drops, overheating, or persistently loud GPU fans that might otherwise alert users to a compromise.

To further evade detection, the malware performed extensive anti-analysis checks before activating. The software scanned systems for virtual-machine artifacts, debugging tools, reverse-engineering platforms, packet analyzers, and forensic utilities, including Wireshark, ProcMon, x64dbg, dnSpy, IDA, and Ghidra. If any such tools were detected, the malware terminated itself.

Microsoft says the malware’s operators ultimately used the compromised systems to deploy one of several GPU-focused cryptocurrency miners, including lolMiner, gminer, and SRBMiner-MULTI. Rather than embedding the miners directly into the malware, the payload dynamically downloaded the most appropriate mining software after conducting extensive reconnaissance on the victim system, including GPU model, CPU specifications, installed antivirus software, memory configuration, and overall system activity.

The campaign highlights an alarming development in which attackers are now targeting not only search engines but also AI-assisted discovery systems. While traditional SEO poisoning has existed for years, the growing use of AI chatbots and LLM-powered assistants for software recommendations may be creating a new attack surface where malicious sites gain additional visibility through generated responses. Users need to be extra cautious, as even highly familiar utilities downloaded from seemingly convincing websites may carry hidden malware payloads, particularly when obtained through third-party mirrors or AI-provided links rather than official vendor pages.

Major tech companies are struggling to justify the skyrocketing prices of heavy AI usage, with even major tech firms like Microsoft and Uber looking at changes to their AI process. Following the recent viral post from Uber CTO Praveen Neppalli Naga that the company had blown through its entire 2026 AI budget in just a few months, Uber's Operations chief, Andrew Macdonald, said that token usage just didn't seem to have a direct correlation with useful consumer features.

Microsoft began revoking its developers' access to the Claude Code programming assistant earlier this month, with plans to move them over to the internal Copilot CLI tool by June 30. Although that has been framed as consolidating its teams onto the tools it's developing, it also comes right at the end of Microsoft's fiscal year, suggesting it may have also been a move to cut costs before the new year.

Worsening matters, Goldman Sachs estimates that Agentic AI could see token use increase by over 24 times in just the next few years. There appears to be a growing disconnect between AI needs, AI wants, and the reality of what AI companies can actually afford as costs mount.

Tokens and trade offers

We've been hearing reports for months about how companies and CEOs are struggling to find the tangible benefit of heavy AI deployment. Uber appears to be the latest AI boosting company to have this come to Jesus moment, following the CTO's explosive claims of annual budgets being wiped out in mere months. In the interview with Business Insider, Andrew Macdonald lamented that there just wasn't a clear correlation between the money Uber was investing in AI use and real consumer feature development.

Having talked to the senior engineers, he said there was no link between higher token usage and a proportional increase in consumer features with real benefits for their customers. Although he admitted more code was being shipped, it "was very hard to draw a line" between that and improvements in the software.

Meanwhile, after opening up its workers to Claude Code subscriptions in December last year, Microsoft is now clawing that back in what's seen by many as a financial move, as much as a consolidation. Microsoft also recently announced the switch of Copilot on GitHub to token-based billing, as the cost of running the tool ballooned earlier this year.

A major reason for this is the explosive growth in agentic AI use. These agents can eat up more than 1,000 times the tokens of a single AI chatbot.

Are more tokens really the answer?

Nvidia CEO Jensen Huang famously said in March this year that if an Nvidia engineer on $500,000 a year wasn't using at least $250,000 of tokens in that same period, he'd be alarmed. This isn't a rare sentiment, though. Many company CEOs are now bragging about the extent of their AI use, as if that alone equates to performance increases.

As Business Insider reports, Airbnb's CEO proudly told investors that 60% of the company's code was now AI-generated. Chime claimed it was shipping 84% AI code earlier this year, and even Google is claiming 50% of its code is AI-generated (though crucially, always checked by a human engineer).

Yet these numbers sound very similar to those of Uber. In the CTO's shocking report of budget runaway, they claimed over 80% of Uber software engineers were using agentic AI, and over 60% of the code was AI-generated. Even then, it's not worth the cost.

And those costs can be extreme if the guardrails are removed. OpenClaw creator and now OpenAI employee, Peter Steinberger, recently announced his team of three people had spent over $1.3 million in tokens in a single month running a suite of agentic AI tools.

This very much reinforces the idea that the cost of AI is rising above that of the workers it's supposed to be replacing. That makes many of the layoffs laid at the feet of AI efficiency and productivity increasingly shaky, unless these companies are simply racing to the bottom.

Or at least racing to new hardware. Goldman Sachs' recent AI agent report suggests that the massive efficiency gains coming from next-generation inferencing chips would make AI use so much cheaper that investment can continue unabated, and profit should follow, with AI agents increasing the revenue at AI companies enormously.

Faster, more efficient hardware will take too long

Nvidia will talk up its Vera Rubin platform at Computex and will officially launch it later this year. It improves AI performance by several times over, uses a new process node, and will reportedly offer as much as 10 times the performance per watt, making it dramatically more efficient than its predecessors.

Such huge gains would give the AI companies that first deploy these cards an enormous advantage over the companies still running Blackwell hardware, and even more so over older Hopper designs. But over 50% of the data center projects announced with Blackwell hardware in mind have been cancelled or delayed, and of those that do complete in the next year, just how keen are the developers going to be to replace those GPUs after they've barely gotten started?

In late 2025, Google, Oracle, and Microsoft all adjusted their plans for hardware in the other direction entirely, suggesting they would make it run for six years before replacing it. That seems impossible to square away with ambitious AI plans and hardware leaps every year.

More tokens on less efficient hardware

The reality is, even as some token costs are falling, the explosion in the number of agentic AI requires cannot be offset by hardware efficiency gains that are many years away from reaching effective deployment, if they ever get to the scale needed to catch up with this ramp-up in AI demand.

That means in the short term, even major companies like Microsoft and Uber are restructuring their use of AI to figure out how to continue using it at scale without nuking their budgets in the process. If those companies can't figure out how to afford it, it's increasingly difficult to imagine how the rest of us will be able to.

And if usage drops because of rising costs, the AI companies are never going to find the short-term profit they need to offset the enormous infrastructure spending they're still trying to justify.

There's been some drama unfolding lately in the Windows security world, and today's episode comes from yet another apparent run-in of researcher Nightmare-Eclipse (aka Chaotic Eclipse) against Microsoft. The company saw fit to ban Eclipse's GitHub account for as-of-yet unspecified reasons, forcing them to pack up and move shop to GitLab instead. Additionally, the Redmond firm had allegedly already deleted the Microsoft account Eclipse used for reporting the bugs.

In a blog post, Eclipse claims this action was vindictive, stating once again that Microsoft refused communication attempts and that they "got zero pennies from doing so", a likely allusion to unpaid bug bounties from the MSRC program. The initiative pays out up to $30,000 to $100,000 for per end-point zero-day depending on conditions, and a cool $250,000 if you can crack open Hyper-V. Already having six zero-day exploits under their belt, Eclipse claims that July 14 will bring a reckoning of sorts for the company, hypothetically in the form of more zero-day exploits being published.

Eclipse's dramatic dispute with Microsoft has been ongoing since early April, when they published the BlueHammer zero-day without warning. The language in their blog posts is unclear and passionate, directing cargo tanks of vitriol at Microsoft/MSRC. As a broad summary, Eclipse implies that Microsoft ignored or refused their zero-day reports and/or did not pay out bounties as requested, somehow causing financial harm in the process. Among other statements, Eclipse says "[they were] told personally by [Microsoft] that they will ruin my life and they did", that there's a dead-man switch of some sort, and that they "will make sure [Microsoft's] bones are shattered."

The saga has drawn speculation from other experts, like William Dormann from Tharros, who said that "MSRC used to be quite excellent to work with. But to save money, Microsoft fired the skilled people, leaving flowchart followers. I wouldn't be surprised if Microsoft closed the case after the reporter refused to submit a video of the exploit, since that's apparently an MSRC requirement now."

Microsoft has been mum on any details about these matters, so it's hard to tell if the situation is about an uncooperative researcher who doesn't follow standard disclosure rules or a company being difficult about security reports. Regardless, the move to ban Eclipse's GitHub account makes for poor optics, as it is being heavily criticized, and ultimately achieves nothing for security, since the code is out there anyway.

In this day and age, when AI-powered security research has arguably made the standard 90-day disclosure-to-patch window completely obsolete, and both time-until-exploit and unused exploits are both nearing zero, Microsoft and other software players would do well to adjust their policies.

Eclipse's technical track record is impressive. They published a string of zero-day exploits for Windows: BlueHammer gets access to the SYSTEM user via Defender, and RedSun does the same; UnDefend knocks Defender offline; GreenPlasma gets SYSTEM access via the CTFMon service, while MiniPlasma grants similar access via a flaw in the Windows Cloud Filter driver. Finally, there's YellowKey, a vulnerability in BitLocker that lets an attacker open up encrypted drives with next to no effort — precisely the action the technology was designed to prevent.

BlueHammer, RedSun, and UnDefend have all been confirmed to be undergoing active exploitation in the wild, and it's not hard to imagine the others are as well, as Eclipse's publications of full or partial proof-of-concept code made it trivial for an interested party to use them.

Many tech companies are pushing their employees to use AI tools and increase their productivity, but it seems that this initiative has begun to backfire. According to The Verge, Microsoft has been reportedly pushing its people to switch to its own Copilot CLI rather than Claude Code because it wants to use an internal tool rather than a third-party one. However, sources say the primary reason is that the cost of using Claude Code has been steadily increasing as more people use the AI tool.

Microsoft is not alone in this, as Fortune reports that other companies are also pulling back on AI usage. While it’s true that the cost of training AI models is falling, making AI tokens more affordable, people have started using more tokens in their day-to-day tasks. This is particularly true for agentic AI, which can use a thousand times more tokens compared to querying an LLM, depending on the number of steps needed to accomplish your instructions. For example, OpenClaw creator Peter Steinberger claimed that his team spent more than $1.3 million in token costs in just a single month. Because of this, it’s now apparent that using AI is more expensive than hiring people, especially since it offers only limited productivity gains at the moment.

Decreasing token costs, paired with increased usage, reminds us of the Jevons Paradox, in which increased efficiency has led to more people using a particular tool or technology. There are many examples of this throughout history — the introduction of efficient steam engines during the Industrial Revolution led more firms to deploy these tools to increase productivity. This is also evident in the airline industry: as planes became more fuel-efficient, lower ticket prices led to higher demand, and air travel demand is now on track to double by 2050, according to IATA.

It seems that this is also true with AI tools, especially as many companies are deploying them in a bid to increase productivity. Nvidia CEO Jensen Huang famously said that its engineers should use AI tokens worth at least half their annual salary each year to be fully productive, even going so far as to say, “Are you insane?” to managers who discouraged AI use. This phenomenon, called “tokenmaxxing,” has led many employees to use AI for just about anything to hit internal targets. This was evident at Amazon, where some team members admitted to using the tool for unnecessary tasks to inflate internal usage scores, and it has also been reported at other companies, such as Microsoft and Meta. Incidentally, these companies are among the biggest spenders on AI development.

It’s unclear yet whether these companies will change their policies now that increased token use, which comes with associated costs, has become an issue. AI is indeed a useful tool, but some companies are using it to replace people in a bid to cut labor costs. If the number of tokens needed to accomplish tasks outpaces the speed at which these tokens become cheaper, then that move might just backfire.

Microsoft announced Advanced Shader Delivery (ASD) for Xbox ROG Ally devices last year, bringing precompiled shaders to the handheld to improve load times. Since then, ASD has been included in the DirectX SDK, with both Intel and Nvidia already releasing their own versions of the tech. Today, AMD joins them as Microsoft expands ASD beyond handhelds to RDNA 3, RDNA 3.5, and RDNA 4 GPUs.

Forza Horizon 6 is the latest game to feature ASD on Windows 11 PCs, but you need the Microsoft Store/Xbox PC app version to take advantage of it. Using an RX 7600 GPU and a Ryzen 7 5800 CPU, Advanced Shader Delivery helped the game boot up 95% faster, taking only four seconds to load on first launch. Without ASD enabled, Forza Horizon 6 took nearly a minute and a half to load otherwise.

This is because every time you install a new game or go through an update, the game needs a fresh shader cache, which can take a few minutes to rebuild. Even if you don't touch the game, but your GPU drivers are updated, you need to recompile shaders for all of your games when you open them for the first time. This process is necessary because shader compilation is performed on-device and must account for hardware variability.

Microsoft's solution is to decouple shaders from drivers entirely and place them in a Precompiled Shader Database (PSDB) that lives in the cloud. Every time you download a game from the Microsoft Store or the Xbox PC app, ASD detects your specific configuration (game, GPU, driver) and downloads the precompiled shaders in advance. So, when you open the game, the shaders are already compiled, and you don't have to wait.

Consoles have done this forever, which makes sense considering they don't have to worry about different hardware configs. Even Valve has a version of precompiled shaders for Linux that it developed for the Steam Deck, but never ported to Windows. Since SSDs have made long load times a thing of the past, this has been one of the last remaining hurdles to instantaneous game launches.

Advanced Shader Delivery is available only on RDNA 3 and later GPUs and only on games downloaded via the Xbox PC app or the Microsoft Store. We don't know when other Windows marketplaces will adopt it, but at least 34 other games should support ASD right away, since they were part of the original announcement for the Xbox ROG Ally handhelds.

Moreover, you need to be enrolled in the Xbox Insiders program to get the Xbox Insiders Hub app, which is required to enable Advanced Shader Delivery, since Microsoft is classifying this as a preview. The company worked closely with AMD to develop it, which is why it's limited to RDNA 3+ hardware for now, because otherwise, it's supposed to be a universal solution for all GPUs on Windows 11 going forward.

Regardless, if you have an Nvidia GPU, the Nvidia app should have a feature called "Auto Shader Compilation" that does the same thing. It gives you more granular control over the size of the shader cache, too. If you're rocking an Intel GPU, then check out "Precompiled Shader Distribution" in the Intel Graphics app instead. Intel has implied that it's using its own cloud database for now, with the ASD standard being adopted later this year.

Microsoft has outlined a new feature of Windows called Cloud-Initiated Driver Recovery (CIDR). This newly introduced capability lets Microsoft remotely roll back a bad driver to a previously known good version on affected PCs. Moreover, it can work without user action or OEM intervention. It sounds like a magic bullet for a long history of Windows Update woes, but we’ll have to see if it works when the rubber hits the road. CIDR will only work with drivers distributed via Windows Update.

Windows Update can cause plenty of problems when a bad driver gets through testing and gets pushed to users. Indeed, buggy drivers have caused many a lost hour, gray hair, wrinkle, high blood pressure, and so on, among Windows veterans. Microsoft also notes that a bad driver often means a user has to manually intervene and roll back to “a low-quality driver for an extended period.” So, the new CIDR is cautiously welcomed.

Microsoft spells out the CIDR process in its Tech Community blog, and there we learn that recovery starts by the Windows developer triggering "a recovery action directly from the Hardware Dev Center (HDC) Driver Shiproom." Once a problematic driver is flagged, the system recovers the previously known-good version of a driver via the Windows Update pipeline. “This is handled through coordinated updates to the PnP driver stack and the driver flighting and publishing services,” says Microsoft.

Importantly, Microsoft notes that “recovery is delivered through the existing Windows Update infrastructure — no new client agent or partner tooling is required.” That should help CIDR work well with what we’ve already got and for it to become an established standard. Likewise, partners don’t need to get involved in CIDR, Microsoft will manage it. However, Microsoft asks that these partners “continue monitoring their driver quality metrics in the Hardware Dev Center dashboard and to respond promptly to any shiproom feedback on rejected submissions.”

CIDR is rolling out now for validation and testing, and it is expected to automatically support the Hardware Dev Center publishing process from September onwards.

Amazon is the latest hyperscaler where employees have been caught inflating AI token consumption to hit internal usage targets, following similar behavior documented at Meta and Microsoft last month, the Financial Times reports.

The company set targets requiring more than 80% of its developers to use AI tools each week and tracked consumption on internal leaderboards. Some employees told FT they had been using MeshClaw, an in-house agent platform that can initiate code deployments, triage emails, and interact with Slack to maximize their token numbers. Amazon said usage statistics would not factor into performance evaluations, but multiple employees said they believed managers were monitoring the data. One said there was "so much pressure to use these tools," another described how tracking created "perverse incentives."

The practice — dubbed "tokenmaxxing" — has become widespread enough to generate its own vocabulary and leaderboards, but beyond workplace culture, if a meaningful share of AI consumption is performative, how reliable are the demand figures that hundreds of billions in AI infrastructure procurement are being allocated against?

Combined 2026 capex from Amazon, Microsoft, Alphabet, and Meta is tracking between $650 billion and $700 billion, with some Wall Street projections exceeding $1 trillion for 2027, and every hyperscaler has told investors that inference capacity is being absorbed as fast as it can be deployed. Internal developer consumption is obviously part of that absorption, and it sits alongside paying external customers in the usage data that informs the likes of capacity planning, GPU orders, HBM procurement, and power infrastructure.

Tokenmaxxing doesn’t mean the demand is fabricated — enterprise AI adoption is broadening, and inference workloads are scaling into production — but there’s a distinction between adoption and consumption intensity. The former is a durable driver of demand, whereas the latter is gameable, and it’s currently being amplified by the incentive structures that these companies built. The water is further muddied by reports that AI is more expensive than actual workers.

Meta's internal leaderboard lasted days after public exposure, and Amazon recently restricted visibility of team-wide usage statistics. And when measurement shifts, the consumption intensity they incentivized will shift with them.

Nvidia CEO Jensen Huang has highlighted per-engineer token consumption as a key metric, stating he’d be "deeply alarmed" if a $500,000-a-year engineer was not consuming at least $250,000 in tokens. Nvidia's inference growth obviously depends on that consumption being a productive workload that persists and compounds because every inflated token is real GPU time.

Angie Jones, formerly VP of engineering for AI tools at Block, told LeadDev she expected the industry to pivot toward measuring efficient token usage rather than celebrating volume. In a cycle where GPU orders and power commitments are being placed years in advance, the quality of the demand projections behind them matters. The hyperscalers are building for a world where every knowledge worker consumes hundreds of thousands of dollars in annual compute. Whether that consumption proves productive or performative will determine how much of this year's $700 billion generates durable returns.

Microsoft Threat Intelligence said in an X post on Monday that it is investigating a compromise of the mistralai PyPI package after attackers reportedly injected malicious code that automatically executed on import, downloaded a secondary payload disguised as transformers.pyz, and launched malware on Linux systems — the latest incident researchers believe may be linked to the broader “Mini Shai-Hulud” software supply-chain campaign targeting developer ecosystems.

According to Microsoft, the compromised mistralai package version 2.4.6 contained malicious code inserted into mistralai/client/__init__.py that silently downloaded a file from a remote IP address to /tmp/transformers.pyz and executed it in the background whenever the package was imported on Linux machines.

The filename appears deliberately chosen to resemble Hugging Face’s widely used Transformers AI framework, potentially allowing the malware to blend into machine learning environments and evade suspicion. Microsoft said the second-stage payload functioned primarily as a credential stealer, but also contained country-aware logic and a destructive branch capable of executing rm -rf / under certain geographic conditions. The payload contained logic designed to avoid Russian-language environments, a behavior commonly observed in some cybercriminal malware campaigns, though such checks are not definitive indicators of attribution.

The disclosure comes amid a growing wave of software supply-chain compromises affecting both npm and PyPI ecosystems. Earlier Monday, security firm Aikido warned that malicious package versions tied to the popular TanStack JavaScript ecosystem had been compromised in two separate attack waves beginning around 19:20 UTC. Affected packages reportedly included @tanstack/react-router, @tanstack/history, and @tanstack/router-core, components collectively downloaded tens of millions of times per week.

Hours later, Aikido said several Mistral npm SDK packages had also been compromised as part of the same ongoing “Mini Shai-Hulud” campaign, including @mistralai/mistralai, @mistralai/mistralai-azure, and @mistralai/mistralai-gcp. The firm warned developers to immediately rotate GitHub tokens, npm credentials, cloud API keys, and CI/CD secrets if affected packages had been installed.

Microsoft has not publicly attributed the PyPI compromise to Mini Shai-Hulud. Still, the incidents share several characteristics, including malicious code inserted into trusted packages, staged payload downloads, credential theft, and automatic execution during installation or import. That overlap has raised concerns that attackers are increasingly targeting developer infrastructure itself rather than end users directly.

Modern development environments often contain high-value credentials, including GitHub personal access tokens, cloud deployment keys, SSH credentials, npm publishing tokens, and CI/CD system access. A compromised developer workstation or CI runner can therefore provide attackers with a path into much larger software ecosystems, allowing malicious updates to spread through legitimate package distribution channels.

The behavior observed in the compromised Mistralai package reflects that escalation risk. According to Microsoft’s analysis, the injected code silently used curl to retrieve the secondary payload before launching it as a detached background process designed to continue operating independently of the original Python session. The malware also reportedly suppressed execution errors and limited activity to Linux systems, the dominant operating system across servers, cloud environments, and many AI workloads.

Supply-chain attacks have become an increasingly serious concern across the software industry because of the sheer scale at which trusted dependencies are reused. A single compromised package can rapidly propagate into thousands of downstream applications, enterprise environments, and production systems. Major incidents in recent years have included the SolarWinds breach, the event-stream npm compromise, the 3CX supply-chain attack, and the XZ Utils backdoor attempt.

The latest wave appears particularly notable for simultaneously targeting AI tooling, cloud SDKs, and widely used frontend development frameworks. Researchers believe the campaign’s primary objective is credential theft, potentially allowing attackers to compromise additional packages, maintainer accounts, and publishing infrastructure in a cascading chain of ecosystem infections.

Microsoft advised organizations to isolate affected Linux hosts, block outbound connections to the malicious IP address, hunt for indicators including /tmp/transformers.pyz, pgmonitor.py, and pgsql-monitor.service, and rotate any potentially exposed credentials immediately. The compromises are still under investigation, and additional affected packages may emerge as maintainers and security firms continue auditing publishing infrastructure and compromised credentials.

A $1 billion data center that Microsoft and Abu Dhabi-based AI firm G42 planned to build in Kenya has stalled after the Kenyan government failed to meet Microsoft's demand for guaranteed annual capacity payments, Bloomberg reported Sunday. Kenyan President William Ruto put the scale of the project's power requirements into clear terms at a recent state event in Nairobi, saying the country would need to "switch off half the country" to keep the facility running.

The project, announced in May 2024 during Ruto’s visit to Washington, was supposed to bring a geothermal-powered data center to the Olkaria region in Kenya's Rift Valley. G42 was to lead construction, with the facility running Microsoft Azure in a new East Africa cloud region. The first phase targeted 100 megawatts of capacity and was expected to be operational by this year, with a long-term goal of scaling to 1 gigawatt.

President Ruto isn’t exaggerating about shutting off half the country’s power. Kenya’s total installed electricity capacity sits between 3,000 and 3,200 megawatts, and peak demand reached a record 2,444 megawatts in January, according to data from KenGen, the country’s government-owned electricity producer.

The full 1 gigawatt build would therefore have consumed roughly a third of the country’s total capacity, and even the first 100 megawatts would have required a significant share of the Olkaria geothermal complex's output, which currently generates around 950MW across all its plants.

John Tanui, principal secretary at Kenya's Ministry of Information, told Bloomberg that the project hasn’t been withdrawn and that talks are continuing, adding that the “scale of the data center they [Microsoft] wanted to do still requires some structuring.” A separate 60-megawatt project with local developer EcoCloud is also still under discussion.

Kenya’s Microsoft campus was set to be the first facility that Microsoft and G42 built together after Microsoft invested $1.5 billion in G42 back in 2024. That deal followed G42's agreement to divest from Chinese holdings and strip Huawei equipment from its systems under pressure from Washington. Microsoft President Brad Smith joined G42's board as part of the arrangement and described the Kenya project at the time as the “single biggest step forward” for digital technology in the country's history.

Meanwhile, Huawei is expanding its presence in Kenya, having launched a new fiber broadband service with Safaricom, Kenya’s largest telecom operator, last week. Africa currently hosts roughly 1% of the world's data center capacity.

Microsoft is spending $190 billion on capex in 2026, and the company adds approximately 1 gigawatt of data center capacity every three months globally. But power constraints are proving to be a universal bottleneck: nearly half of planned U.S. data center builds this year have been delayed or canceled due to shortages of electrical infrastructure.

A top Microsoft exec has admitted that Windows 11 still relies on a bunch of old code from the 1990s. It is refreshing for Microsoft Azure Chief Technical Officer Mark Russinovich to highlight this fact on social media, but it might not surprise as many folks as he thinks.

As Russinovich eloquently puts it, those of us invested in the computer scene in the 90s “were thinking flying cars and moon stations by the year 2026, not Win32.” The admission that such old software tech is still the "bedrock" of Windows today may be the CTO strategically sharing a cold, hard truth, providing a 'let's be real' moment as part of Microsoft’s latest charm offensive. Sharing a candid confession indicates that the corporation is actually aware of the issues in its OS.

Remember, the firm is currently in the midst of a major transformation, targeting enthusiast hot button areas like Windows performance, overhead, and reliability. This drastic pivot was cautiously welcomed in contrast to Microsoft being widely slammed for boasting about Windows “evolving into an agentic OS” last November. Currently, Microsoft seems to be flailing around, trying to stop folks straying to pastures greener like Mac and Linux.

However, Win32 isn’t inherently labeled as ‘bad’ by the Microsoft CTO. Though it is ancient, it has probably stuck around for good reason. “I think one of the reasons it’s got this staying power is just a fundamental layer inside of Windows that so many apps have built on,” notes Russinovich. “So many technologies and ecosystems have been built on top of it that it’s kind of a bedrock.”

The CTO explains that Win32 has persisted even when facing targeted existential threats from within Microsoft, particularly in the Windows 8 era. “There’s been various times in Microsoft’s history where we thought we’d reboot the Windows API surface, like WinRT, that actually didn’t play out the way a lot of people expected it to.”

In closing, Russinovich highlights that Win32 was also the bedrock for tools like Sysmon and ZoomIt, which he actually wrote back in 1996. These tools are now “more relevant than ever in 2026,” as parts of Windows 11 and PowerToys, respectively, reckons the CTO.

Xbox CEO Asha Sharma announced on X on Tuesday that the company is winding down its Copilot for Gaming feature on mobile and canceling its planned console launch.

Sharma also unveiled a broad leadership overhaul, bringing in four senior executives from her former CoreAI engineering group at Microsoft, according to an internal memo reported by CNBC. Sharma called the Copilot pullback part of a push to "retire features that don't align with where we're headed."

Microsoft first unveiled Copilot for Gaming at GDC in March 2025, positioning the AI assistant as a real-time sidekick that offers gameplay tips, coaching, and session recaps. A beta rolled out to the Xbox mobile app, then expanded to the PC Game Bar in September, and the ROG Xbox Ally handheld. Then came a privacy backlash in October after a user discovered that it was sending screen-activity data to Microsoft's servers by default.

Sharma hasn’t addressed the status of Gaming Copilot on the PC Game Bar or the ROG Xbox Ally handheld, leaving the future of those versions unclear. In the memo, Sharma wrote that Xbox needs to "evolve how we work" and acknowledged the division spends "too much time inward instead of with the community."

She appointed four executives from CoreAI: Jared Palmer, formerly VP of product at CoreAI and a senior VP at GitHub, will serve as VP of engineering and technical advisor to Sharma; Tim Allen, previously CoreAI's VP of design, becomes head of Xbox design; Jonathan McKay, who held growth roles at OpenAI and Meta before leading CoreAI growth, takes on the same function at Xbox; and Evan Chaki, a CoreAI general manager, will lead a new team focused on simplifying development workflows. A fifth hire, David Schloss, joins from Instacart to oversee Xbox's subscription and cloud business.

Two senior Xbox executives are departing. Kevin Gammill, corporate VP of gaming ecosystem organization, is leaving the company. Roanne Sones, corporate VP of Xbox devices and ecosystem, will take a leave of absence after the summer, then transition to an advisory role.

Sharma took over as Xbox CEO in February after Phil Spencer retired following 38 years at Microsoft. Since then, she has axed the "This is an Xbox" marketing campaign, cut Game Pass prices, rebranded Microsoft Gaming back to simply Xbox, and outlined a back-to-basics strategy focused on console and community. The leadership changes arrive as Xbox continues to struggle financially: gaming revenue fell to $5.3 billion in the most recent quarter, down from $5.7 billion a year earlier, and hardware revenue dropped 33%.

A new Microsoft-backed AI study claims a 'Transformation Paradox' is holding back the adoption of AI in workplaces, with many users preferring to focus on current goals rather than AI innovation. According to the Microsoft study, only 1 in 5 workers are equipped to use AI tools and are in an environment where management clearly supports it. On the other hand, nearly half of the survey respondents are in an unclear situation, where both individual skill and conditions within their organization are still unclear when it comes to AI policies.

Many companies are deploying AI for the sake of it, with only a few reporting actual gains from its use. The Microsoft study seemingly investigated this phenomenon, and it found that the issue isn’t caused by the lack of adoption, but the policies behind it. “What emerges is a pressure point within the organization where the pull to perform collides with the push to transform,” the research paper said. “65% of AI users fear falling behind if they don’t use AI to adapt quickly, yet 45% say it feels safer to focus on current goals than to redesign work with AI. And only 13% of AI users say they’re rewarded for reinvention of work with AI even if results aren’t met.” Microsoft calls this phenomenon 'The Transformation Paradox' and says, "The same forces accelerating AI adoption are holding it back."

Some experts are hailing AI as the fourth industrial revolution, and say it will change how we execute many of our day-to-day tasks. However, this study is saying that it’s not just enough to give workers access to AI tools — instead, management must also change how it measures productivity and even be willing to revamp existing workflows and policies to take AI tools into account.

This rush to integrate AI without considering how it can be effectively used is leading to a lot of negative effects. Even OpenAI CEO Sam Altman said that many of the reported layoffs across the globe were being falsely attributed to AI, with some experts saying that it was, in fact, due to poor business decisions. Although it cannot be denied that this tool will lead to a massive disruption to the job market, one study suggested that AI firms making the most use of AI are hiring more people.

As the cost of using AI is increasing and is now comparatively more expensive than actual workers, using these tools as efficiently as possible has never been more important. And while executives can easily sign up their people for an AI subscription, Microsoft’s study suggests that until companies change their approach to their business from the top down, they won’t be able to maximize these tools and will be left behind by companies that do. We should note, though, that Microsoft sponsored this research paper and that it has invested billions in Copilot, OpenAI, and AI data centers, so it stands to gain from the wider adoption of AI.

Google, Microsoft, and Elon Musk's xAI agreed today to give the U.S. Commerce Department's Center for AI Standards and Innovation (CAISI) access to their AI models before public release, Bloomberg reports. OpenAI and Anthropic, which had existing evaluation partnerships with the center dating to 2024, renegotiated their deals to align with priorities in Trump's AI Action Plan, the agency said.

The agreements mean that every major U.S. frontier AI lab now participates in voluntary pre-release government evaluations. CAISI has completed more than 40 model assessments to date, including evaluations of unreleased state-of-the-art systems, according to the Commerce Department.

CAISI operates within NIST and was originally established in 2023 under Biden as the AI Safety Institute. The Trump administration renamed it last June, with Commerce Secretary Howard Lutnick calling the rebrand a move away from what he called regulation "used under the guise of national security." Despite the shift in rhetoric, the center's core function has remained largely the same: evaluating frontier models for cybersecurity, biosecurity, and chemical weapons risks.

"These expanded industry collaborations help us scale our work in the public interest at a critical moment," CAISI director Chris Fall said of the new agreements. Fall took over the center after Collin Burns, a former Anthropic and OpenAI researcher, was pushed out just four days into the job. The Washington Post reported last month that White House officials were concerned about Burns's Anthropic ties, given the administration's ongoing dispute with the company. Burns had relocated across the country and given up Anthropic equity to take the position.

The center still lacks permanent legal standing, and some lawmakers have introduced draft legislation to codify it, but nothing has passed. Trump's AI Action Plan, announced in July last year, directs CAISI to serve as part of an "AI evaluations ecosystem" and lead national security-related model assessments. It also instructs regulators to explore using evaluations when applying existing law to AI systems.

Anthropic's renegotiated deal with CAISI sits alongside a separate and hostile set of interactions with the federal government. The Pentagon designated Anthropic a supply chain risk in March after it refused to lower guardrails on autonomous weapons, though a federal judge later called that move "Orwellian." Both Defense Secretary Pete Hegseth and Trump have outlined a six-month phaseout period for government use of Anthropic's tools, and two active lawsuits remain unresolved.

The new CAISI agreements also come one day after reports that the Trump administration was considering a mandatory pre-release review process for AI models via executive order, with Anthropic's Mythos model cited as the catalyst. The voluntary agreements announced Tuesday, and any potential mandatory review framework, would run in parallel, though it remains unclear how they might interact.

Microsoft continues to make some of the earliest chapters of its operating system history open-source and freely available. Earlier this week, it announced that Tim Paterson's DOS listings, containing source code of the 86-DOS 1.00 kernel, various PC-DOS 1.00 pre-release kernels and utilities, and the Microsoft BASIC-86 Compiler runtime library, were available on GitHub. Microsoft VP Scott Hanselman tied the release to 86-DOS 1.00’s 45th anniversary. The exec confirmed that the code, transcribed from reams of old dot matrix printouts found in a garage, was perfect, "and recompiles byte for byte to the original binaries.”

If you head on over to the GitHub page to snag the code, you will see a photo of Tim Paterson standing in his garage with a pile of yellowed dot matrix printouts in the foreground. These pages contain the code for the software mentioned in the intro, and you can even see the original scans in PDF and PNG format via a link to the Internet Archive. These include the coder’s handwritten notes.

Probably more important to tinkerers, though, is the fact that the work of transcribing the printed code has been completed (for those three mentioned wares). Tips to compile and assemble the sources can also be found on Paterson’s GitHub.

From 86-DOS to MS-DOS

In case you aren’t familiar with the place of 86-DOS (or Tim Paterson) in Microsoft’s history, here's a short refresher. Microsoft took a shortcut and gained a foothold in the OS software market by purchasing 86-DOS from Seattle Computer Products and inventor Tim Paterson for a figure in the region of $75,000.

In the GitHub repository, you can see 86-DOS’s transformation into the PC-DOS 1.00 kernel, alongside code for some well-known utilities (still in use today) such as CHKDSK. As the Microsoft blog asserts, this work “offers rare insight into how MS-DOS/PC-DOS came to be, and how operating system development was done at the time, not as it was later reconstructed.”

So, we have another old DOS release to tinker with. In April 2024, we reported on Microsoft releasing the code for MS-DOS 4.00 under the generous MIT License, allowing tinkerers free rein. It did the same with MS‑DOS 1.25 and 2.11 in 2018. Also in 2024, we coincidentally covered a video demo featuring 86-DOS version 0.1C being taken for a test drive (via the Internet Archive), and now version 1.00 of this OS has hit GitHub, straight from the files squirreled away in Tim Paterson's (the creator’s) garage.

We’re still waiting for any version of Windows to be open-sourced. You have to dig through leaks if you are curious enough to want to investigate the source code for Windows XP, for example.

The U.S. Department of War has announced deals with "seven of the world’s leading frontier artificial intelligence companies" for operational use. According to the Classified Networks AI Agreements press release, SpaceX, OpenAI, Google, Nvidia, Reflection, Microsoft, and Amazon Web Services will deploy their LLMs across the Pentagon’s classified networks “for lawful operational use.” The government said that this move will help turn the United States military into “an AI-first fighting force” and will help with “decision superiority across all domains of warfare.”

It seems that the AI tools that these companies offer will, for now, be limited to data analysis and help make decision-making faster and easier as the U.S. faces complex situations. These tools are accessible via GenAi.mil, the Pentagon’s official AI platform, through the Department of War’s network and are widely available for its personnel.

“Over 1.3 million Department personnel have used the platform, generating tens of millions of prompts and deploying hundreds of thousands of agents in only five months,” the Pentagon said. “Warfighters, civilians and contractors are putting these capabilities to practical use right now, cutting many tasks from months to days.”

Nevertheless, there have been concerns about the use of AI in military applications. Anthropic has famously refused to budge on the Department of War’s demand to lower its safeguards, saying that doing so could mean that its AI products could be used for mass surveillance or to create autonomous weapons. This move resulted in President Donald Trump banning the company from federal agencies, even going as far as designating it a supply chain risk for refusing to bow to the federal government’s demands.

While AI is certainly useful for distilling massive amounts of information and spotting patterns that humans can miss, it’s still not a 100% reliable tool for making decisions that could have a global impact. A researcher discovered this when they pitted GPT-5.2, Claude Sonnet 4, and Gemini 3 against each other in a wargame, with 95% of the outcome ending in a tactical nuclear strike. Three scenarios even ended in a strategic nuclear strike that would have ended the world.

But even though these AI tools are limited to analysis and support, with a human operator at the helm still responsible for every decision, there’s also the risk of automation bias. This is a person’s tendency to follow a computer’s suggestion despite contradictory information, especially as AI systems can process a ton of data so much more quickly than any human could. However, the data the AI is relying on could be false, erroneous, or misinterpreted, so it’s crucial that humans apply their intuition and experience before accepting AI suggestions at face value.

The U.S. military isn’t the only one experimenting with and deploying AI technologies in operational use. China, for example, has been showing off a 200-strong AI drone swarm that can be controlled by a single soldier, as well as ground-based drone wolfpacks armed with machine guns and grenade launchers for urban combat. While we cannot stop these armed institutions from deploying AI tools for intelligence-gathering, reconnaissance, and decision-making on the battlefield, we can only hope that they do not ignore safeguards and never give AI the triggers to any weapon.

Google, Amazon, Microsoft, and Meta plan to spend a combined $725 billion on capital expenditure in 2026, a 77% increase over last year's record $410 billion, according to first-quarter earnings reports compiled by the Financial Times.

Google led with 63% cloud revenue growth and an 81% jump in net income to $62.6 billion, while Meta's stock dropped 6% after hours despite a 33% revenue increase, punished by investors for adding $10 billion to its spending forecast and offering no firm timeline on new AI models.

But in the earnings calls, at least two of the four companies explicitly blamed rising memory chip prices for pushing budgets higher, confirming what market data and industry executives have been warning about for months.

Memory costs inside the capex

Microsoft’s CFO, Amy Hood, told investors that rising prices for memory chips and other components accounted for $25 billion of the company's record capex budget. Microsoft set its 2026 spending at $190 billion, far above the $152 billion average analyst forecast. Hood warned that even with the additional investment, Microsoft expects to remain capacity-constrained on GPUs, CPUs, and storage through at least 2026.

Meta cited the same, with the company raising its full-year capex range to $125 billion to $145 billion, up from a prior ceiling of $135 billion. In its earnings release, Meta attributed the increase to "higher component pricing this year, particularly memory," alongside rising costs for land, power, and skilled workers needed to build data centers that now consume 70% of the world's memory output.

The timing of all this is hardly coincidental, with TrendForce having reported DRAM contract prices rising roughly 95% quarter over quarter in Q1 2026, with a further 58% to 63% increase projected for Q2. NAND is following a similar trajectory, with Q2 contract prices expected to climb 70% to 75%. Server DRAM and high-density DDR5 RDIMMs are absorbing the bulk of production capacity, and all NAND output for 2026 is already committed, according to Phison CEO Khein-Seng Pua.

Hood's $25 billion, therefore, helps to put a dollar value on what has previously been an abstract concern: If one company's memory cost inflation alone exceeds the entire annual capex of most semiconductor firms, the pressure on consumer DRAM and NAND supply becomes much easier to quantify.

Google Cloud's contract backlog

Meta and Microsoft aside, Google’s Cloud revenue hit $20 billion in the same quarter, growing 63% year over year, outpacing both Amazon Web Services ($37.6 billion, up $8.3 billion) and Microsoft's Azure-driven cloud segment ($34.7 billion, up $7.9 billion).

Google's cloud contract backlog reached $460 billion, roughly double the $240 billion reported at the end of Q4 2025. Amazon reported $364 billion in its own pipeline, which will expand further after a recent $100 billion computing contract with Anthropic over the next decade. Microsoft's commercial remaining performance obligations hit $625 billion, up 110% year over year.

Cloud boss Thomas Kurian attributed Google's growth to its strategy of building custom AI chips, foundation models, and products in-house, telling the Financial Times that this gives the company a cost and research advantage over competitors that have struggled to develop their own chips and frontier models. Google's 7th-gen Ironwood TPU, which packs 192 GB of HBM3E per chip with 7.37 TB/s bandwidth in pods of up to 9,216 chips, is central to that strategy, and Anthropic has committed to access up to one million of them. Google recently unveiled its 8th-gen TPUs, which are split into two distinct variants for training and inference.

Alphabet raised its capex guidance to between $180 billion and $190 billion, up $5 billion from its previous guidance of $175 billion. CFO Anat Ashkenazi said he expects capex to “significantly increase” in 2027, causing shares to rise by some 7% after hours. It’s worth noting that $37.7 billion of Alphabet’s net income of $62.6 billion came from unrealized gains on non-marketable equity securities, primarily the company's Anthropic stake, according to the earnings release filed with the SEC. Strip that out, and operating performance was still strong, with a 36.1% operating margin, but the total net income number overstates recurring profitability.

Custom silicon and the GPU question

These capex figures reflect more than GPU purchases, because each hyperscaler is now deploying or developing custom accelerators to reduce dependence on Nvidia for inference-based workloads.

Amazon's Trainium3, built on a 3nm process with 144 GB of HBM3E and roughly 4.9 TB/s of bandwidth, is what CEO Andy Jassy described as "nearly fully subscribed" for 2026, and Meta has announced four generations of its MTIA inference chip, all fabbed at TSMC alongside Broadcom, even as it signed GPU deals worth roughly $110 billion combined with AMD and Nvidia. Meanwhile,. Microsoft's Maia 200 is deploying in U.S. Central data centers.

This pattern is likely to extend beyond accelerators as CPU demand for agentic AI workloads drives a parallel supply crunch with CPU lead times currently stretching to six months. Intel has reported billions in unmet Xeon demand, and Arm CEO Rene Haas has stated that agentic workloads require roughly 120 million CPU cores per gigawatt of data center capacity, four times what traditional AI training clusters need. Per Intel CFO David Zinsner, data center CPU-to-GPU ratios have already moved from 1:8 to 1:4, with further convergence expected to reach or go beyond parity.

Despite record spending, all four companies have acknowledged supply constraints that additional capital alone can’t resolve. Nvidia has booked an estimated 800,000 to 850,000 wafers of TSMC's CoWoS advanced packaging capacity for 2026, consuming over half of the total output and leaving AMD, Broadcom, and Google's TPU program competing for the remainder. CoWoS remains oversubscribed through at least mid-2026, and TSMC's U.S. packaging fabs aren’t expected to reach volume until 2028.

Power infrastructure is another bottleneck, with large power transformer lead times extending to roughly 128 weeks, and the IEA estimating that approximately 20% of planned global data center projects could be at risk of grid-related delays. TrendForce recently downgraded its full-year server shipment growth forecast from 20% to 13% because power management ICs and baseboard management controllers needed to assemble complete servers are stretching to 35- to 40-week lead times. Samsung's planned closure of its S7 eight-inch wafer fab in Korea will tighten PMIC supply further.

‘The bear thesis is garbage’

Meta's stock slipped by 6% after-hours following the earnings, erasing roughly $113 billion in market value. That drop reflected both the $10 billion capex increase and CEO Mark Zuckerberg's lack of a firm schedule for releasing improved AI models to follow the recently launched Muse Spark. Dec Mullarkey, managing director of SLC Management, told the FT that investors are concerned about whether Meta's historically capital-light business is becoming far more capital-intensive.

"The bear thesis is garbage," countered Brent Thill, an analyst at Jefferies, arguing that revenue growth across the sector justifies the spending. Zuckerberg offered little to settle the debate. Asked about Meta's AI agent development, he told investors he cared more about quality than deadlines, adding that most AI agents available today are not good enough for everyday users.

Amazon kept its $200 billion capex plan unchanged, and Microsoft CEO Satya Nadella said ending his company's exclusive contract with OpenAI was beneficial, claiming royalty-free access to OpenAI's frontier models and IP through 2032.

Microsoft announced yesterday that Xbox Mode is rolling out to Windows 11 PCs, bringing the controller-optimized, full-screen gaming interface to desktops, laptops, and tablets for the first time. The feature, which consolidates games from multiple PC storefronts into a single dashboard, launched in select markets on April 30 and will expand to more users over the coming weeks.

Xbox Mode replaces the standard Windows desktop with an Xbox-style interface designed for gamepad navigation. It pulls together installed titles from Steam, Epic Games Store, Battle.net, and the Xbox Game Pass catalog into one unified library. Users can browse, launch, and switch between games without touching a mouse or keyboard, then return to the regular Windows desktop whenever they want.

The feature runs on top of Windows 11 and, when active, suppresses background distractions and presents a streamlined UI modeled on the Xbox console dashboard. Microsoft is marketing it as an optional layer that doesn’t lock users out of the full desktop environment, and switching between the two is seamless.

Xbox Mode originated as the Full Screen Experience, a feature initially exclusive to the Asus ROG Xbox Ally handheld. That version leaked to other handhelds via registry tweaks in September 2025, and Microsoft officially expanded it to all Windows 11 handhelds two months later. The jump to desktops and laptops was confirmed at GTC 2026 in March, alongside the reveal of Project Helix, Microsoft's next-generation Xbox console built on a custom AMD SoC.

The expansion to full-sized PCs puts Xbox Mode in direct competition with Valve's Steam Big Picture mode and, more broadly, with SteamOS itself. Valve's operating system has attracted a following among handheld and living-room PC gamers for its low-overhead, controller-first design. Xbox Mode's advantage is native access to every Windows-compatible game and storefront without the compatibility layers that Linux-based SteamOS requires.

Xbox Mode is also a building block for Project Helix, the next-gen Xbox that Microsoft confirmed will run both console and PC games on a custom AMD SoC, and development kits are expected to ship in 2027. Rolling the same interface across existing Windows 11 hardware gives Microsoft a head start on software maturity before that console arrives.

Xbox Mode is being deployed via Windows Update in a phased rollout. To get access as early as possible, users need to open their settings, navigate to Windows Update, and toggle on "Get the latest updates as soon as they are available." Once the update reaches a given device, Xbox Mode can be launched directly from the PC.

Microsoft didn’t specify which markets are included in the initial wave or provide a timeline for full global availability. The company said it plans to continue evolving the feature based on user feedback.

Google, Amazon, Microsoft, and Meta collectively plan to spend $725 billion on capex in 2026, up 77% from last year's record $410 billion, according to first-quarter earnings compiled by the Financial Times. Google delivered the strongest results, with cloud revenue jumping 63% year over year to $20 billion, while rising memory chip prices pushed spending forecasts higher at both Microsoft and Meta.

"The AI economy is healthy," Brent Thill, an analyst at Jefferies, told the Financial Times, adding that recent revenue growth justified the enormous capital outlays. "The bear thesis is garbage."

Microsoft set its calendar-year 2026 capex at $190 billion, well above the $152 billion average analyst estimate. The company’s CFO, Amy Hood, attributed $25 billion of that figure to rising memory chip and component costs. She told investors that despite the additional spending, Microsoft expects to remain capacity-constrained through at least 2026 as it works to bring GPU, CPU, and storage infrastructure online faster.

Meta increased its full-year projection by $10 billion to a range topping $145 billion. The company cited higher component pricing, particularly for memory, alongside growing competition for land, power, and skilled workers needed to build data centers. Revenue grew 33% to $56.3 billion.

Dec Mullarkey, managing director of SLC Management, told the Financial Times that investors are growing uneasy with Meta's escalating infrastructure costs, questioning whether a historically lean business is becoming far more capital-hungry. “Investors continue to be concerned about how Zuckerberg’s once capital-light money machine may be morphing into a capital-intensive incinerator,” he said.

Alphabet posted an 81% increase in net income to $62.6 billion on revenue of $110 billion. Google Cloud reached $20 billion in quarterly revenue, growing faster than Amazon Web Services ($37.6 billion total, adding $8.3 billion year over year) and Microsoft's Azure-driven cloud segment ($34.7 billion total, adding $7.9 billion).

The company's cloud contract backlog reached $460 billion, roughly double the $240 billion reported at the end of Q4 2025. Google Cloud boss Thomas Kurian credited the company's strategy of building custom AI chips, foundation models, and products in-house for giving it a cost and research advantage over competitors. Alphabet's capex guidance rose by $5 billion to as much as $190 billion, matching Microsoft. Shares climbed 7% after hours, putting Alphabet on track for a record $4.3 trillion market valuation.

CEO Mark Zuckerberg offered no firm schedule for releasing improved AI models to follow the recently launched Muse Spark. Asked about the pace of Meta's AI agent development, Zuckerberg told investors: "There's a lot of agents out there that people are building for different things, but there aren't that many that I would want to give to my mother."

In early 2025, OpenAI announced Stargate, a joint venture with Oracle and SoftBank, which aimed to invest $500 billion in AI data centers in the United States. But after more than a year of challenges and disagreements, it seems that the startup has abandoned the original idea of directly owning infrastructure alongside its two partners. According to the Financial Times, OpenAI now prefers to rely on third-party providers and lease capacity in the long term.

This is a sensible idea for the startup, which is burning through cash and has reportedly missed internal revenue targets in recent months. But it has also caused chaos among its partners and put its reliability into question. According to the report, OpenAI has "in practice... abandoned the joint venture," choosing instead large bilateral deals with Oracle and more. One person involved with Stargate reportedly said the company had "sidelined first-party data centres," while OpenAI itself admitted that Stargate is merely an "umbrella for our compute strategy."

Stargate’s initial goal was to build 20 data centers, with the first project at Abilene, Texas, already operational. However, the three partners reportedly squabbled among themselves for months as they could not agree on who would have ultimate control of the planned data centers. In the end, SoftBank agreed to own and develop the Texas data center, while OpenAI would design and operate it on a long-term lease.

Other Stargate projects located in other areas have also been hit by uncertainties. The UK government signed a deal with OpenAI, among other partners, to build a data center in the UK, but the startup has put it on hold earlier this month. It cited “restrictive regulations” and “high energy costs” as the reason behind the move, but UK AI Minister Kanishka Narayan told the Financial Times that the “only thing that has changed [since] the moment of those commitments…has been the financing environment for OpenAI.”

It has also done the same for another Stargate project in Narvik, Norway, with Microsoft stepping up to take over the lease for the site. OpenAI will then lease compute capacity from Redmond, instead of getting it directly from Nscale, the British company that developed the site and also worked on the canceled UK project.

All these changes have got some partners “feeling let down and misled by OpenAI,” a person familiar with Microsoft’s decision said. Thankfully, the software giant has stepped in on some of the projects that the startup has supposedly abandoned. One source told the publication that money is not unlimited, no matter what Sam Altman might say, while another said that they prefer Microsoft over OpenAI as a tenant, as “they are more creditworthy.”

Even though OpenAI has made a name for itself in AI, the startup has not turned a profit since it was founded in 2015. Many institutions believe in its potential, though, with the firm securing $110 billion in its latest funding round — the biggest amount secured in Silicon Valley history and $10 billion more than what the company initially targeted. Still, some analysts estimate that it could run out of cash by mid-2027 with the massive amounts of money it’s been throwing around to secure more compute.

Anthropic CEO Dario Amodei criticized moves like this, saying that some of his company’s rivals are pushing infrastructure investments too far. However, OpenAI says that it’s ahead of the exponential compute curve, allowing it to have an advantage over everyone else. For example, Anthropic has had to limit access to some features on its various products due to limited resources, and Amodei has had to spend more on securing capacity to satisfy the increasing demand

The biggest difference between startups, like OpenAI and Anthropic, and their more established rivals, like Microsoft, Google, Meta, and Amazon, is cash flow. The startups still rely on external funding to fuel their growth, while the big tech companies have billion-dollar revenue that they can rely on to pour into expensive hardware and infrastructure projects.

Microsoft and OpenAI have once again renegotiated the terms of their deal with one another, but it might be what's best for both of them. OpenAI and Microsoft have announced an end to their exclusive arrangement, and a re-jigging of how they handle model oversight, revenue sharing, and cloud deployments. Microsoft will no longer pay OpenAI for what it makes from Copilot, but OpenAI no longer has to exclusively use Azure servers for ChatGPT, opening it up for further deals with other cloud service providers.

What this means for the ever-nebulous AGI clause that both companies were so keen to retain access to and control over, if and when it materializes, remains to be seen. It's an intriguing move that leaves the immediate future of both companies' AI efforts uncertain, but perhaps it's better than Microsoft's legal department firing all barrels at OpenAI over its recent deal with Amazon.

Where's the ROI?

One of the biggest questions of the AI industry over the past year and a half has been the source of profit. Not the infrastructure investment, or the circular deals and token IOUs, but the real profit. For the investors who pumped tens of billions of dollars into OpenAI, Anthropic, and xAI, and for the shareholders who ballooned Microsoft, Google, and Meta's stock prices off the back of these mega deals and unprecedented investment plans.

Microsoft CEO Satya Nadella hinted at this in January, when he said at the World Economic Forum that AI companies needed to find a clear use for the technology or risk losing the "social permission" to continue the work.

That seems to be more of a pressing issue for Microsoft by April, when it announced that Copilot use on GitHub would move to token-based billing — that is, charging users for the amount of tokens they use, rather than on a per-request basis. No longer would shorter requests with shorter responses cost as much as longer, more in-depth queries. From June, this will result in users paying more when Copilot is verbose in its responses, or when it has to analyze more data before making its suggestions.

Microsoft is already doing that with Azure agents, and it's also set to raise the price of Microsoft 365 with its Copilot integration by several dollars a month for most tiers.

According to internal documents reportedly shared with journalist Ed Zitron, this move came because Microsoft had faced a more-than-doubling of its Copilot-related costs from January this year. He also claims Microsoft will take further steps to tighten controls and increase earnings from individual AI users, including reducing rate limits and forcing users onto different models, which could more than double costs.

Things aren't much better at OpenAI, either. It was projected in January to be on track to run out of money entirely by the end of 2027, and despite announcements of enormous investments in the company, it's projected to burn through tens of billions over the coming years. All while somehow planning to turn a profit by the end of the decade, but to manage that, it would need to earn hundreds of billions of dollars a year. OpenAI's annualized revenue run rate is reportedly sitting at roughly $2 billion per month, or $24 billion a year.

OpenAI also performed several major pivots and navigational shifts in recent months. We learned about its chip manufacturing ambitions in February, it announced it was building a GitHub competitor in March, the company warned that it would shutter the Sora text-to-video generation tool in April, and it bought a podcast for over $100 million that same month.

Even OpenAI's own financial officer has said she doesn't see how OpenAI can afford its own promised infrastructure spending, as it misses key revenue targets in 2026, according to a new WSJ report.

It's very hard to see how any of this takes OpenAI from a heavy-loss-making company to one that's incredibly profitable in just a few years.

Don't drop the bag

OpenAI was under pressure in 2025. To secure the promised investment of billions from Japanese investment firm Softbank, it needed to convert to a for-profit company and settle its disagreements with Microsoft. It managed that just in time, finally securing a long-term partnership agreement with Microsoft in the Fall. The Softbank money came rolling in, and just a few months later, the deal was renegotiated again.

But rejigging the deal may be OpenAI's way of securing the next round of funding — the $50 billion promised investment from Amazon in February, which Microsoft was none-too-pleased about. But in doing so, it's lost one of its limited revenue streams from Microsoft's Copilot earnings, and will still have to pay Microsoft 20% of its own limited earnings.

That Amazon investment could come alongside another $60 billion from Nvidia and SoftBank (though not the $100 billion Jensen originally promised), if all goes to plan. That would also value the company at around $730 billion, making a potential IPO incredibly profitable for Altman and anyone else holding OpenAI shares at the time of a public offering.

But even with OpenAI more than halving its compute ambitions from $1.4 trillion in expenditure to $600 billion by 2030, that's still contingent on increasing its own revenue to $280 billion a year by that same date. As of the time of writing, OpenAI hasn't even managed to earn 10% of that, while having close to a billion active users (though crucially, it also missed that milestone by the end of 2025), and it is losing mindshare to competitors like Anthropic.

Regardless, OpenAI seems keen to push forward with its IPO plans. At this stage, that may be the only real avenue left for it to get anywhere close to its ambitious goals. Even with shifting goalposts, the timeline for its profitability is shrinking rapidly, and it still hasn't made a clear path toward it.

Microsoft is currently facing a lawsuit from the UK Competition Appeals Tribunal (CAT) after it was alleged that the company is charging higher wholesale prices for Windows Server for customers using Amazon AWS, Google Cloud, or Alibaba Cloud. This makes Azure, Microsoft’s cloud services provider, cheaper than the competition as its competitors have to eventually pass on the increased costs to their clients, reports Reuters. Competition lawyer Maria Luisa Stasi filed the case at the Competition Appeal Tribunal in late 2025, representing nearly 60,000 businesses that used Microsoft software on competing cloud services, with her team suggesting that the claim is worth about USD 2.8 billion (or about GBP 2.1 billion).

The company argued that Stasi did not show a concrete way of computing for any alleged losses, so the case should be thrown out. However, the London tribunal certified the case, and it’s now proceeding to trial. Microsoft is planning to appeal the decision and told the publication, “We also dispute the underlying allegations by the class representative, and today’s decision makes no final determination on those claims.”

During a previous hearing, Microsoft said that its strategy of integrating Windows Server with Azure while also licensing it to rivals will help competition. However, the British Competition and Markets Authority (CMA) discovered that this licensing practice is “materially disadvantaging AWS and Google” in mid-2025, with the government body opening another investigation into the company’s licensing practices. Other regulators in the U.S. and Europe are also looking into Microsoft and other cloud computing firms. Some sources said that the Federal Trade Commission (FTC) started scrutinizing claims late last year that the company has put punitive licensing terms in its productivity suite that make it harder for customers to switch to rival platforms from Azure.

This isn’t the only legal trouble that Microsoft faces in the UK. The company is currently appealing a ruling handed down by the same authority, which said that reselling perpetual licenses, including those of Microsoft Office and Windows, is legal and valid. This stemmed from a lawsuit by ValueLicensing, after it argued that Microsoft’s contract, which prohibited reselling of licenses issued before, is against the law. It follows a similar argument in a UsedSoft complaint filed against Redmond more than a decade ago, but Microsoft used a copyright infringement argument this time. It said that Word and other apps contained graphics that are protected by creative work, which the UK tribunal rejected.

After several years of subsequent price rises and management restructuring that led many gamers to believe the Xbox brand was on its way out, Microsoft has pulled off a surprise move. The company announced today in a press release that it would bring the price of Game Pass Ultimate back down from $29.99 to $22.99, while the PC Game Pass goes from $16.49 to $13.99 a month.

It's not all good news, though, as those who subscribed to Game Pass for the sole purpose of playing Call of Duty (CoD) will now have to wait around a year for new titles to appear on the platform. The change does not affect existing CoD releases, and all other day-one game releases will continue.

The now-nullified decision to include CoD in a monthly subscription was widely deemed as particularly gutsy. The game sells tens of millions of copies a year. At $70 a pop for the standalone title, it's possible the company was leaving hundreds of millions, if not billions of dollars, on the table.

Even still, in the broader scheme of things, this development is quite the volte-face for Microsoft, which repeatedly hiked the price of both the consoles and subscriptions over the past couple of years. Those decisions led to wide speculation that the company could be preparing for the end of the Xbox brand, in light of a minority market share versus the PlayStation.

As a quick recap, Microsoft raised the price of the Xbox twice in the 2025 calendar year alone, with the first raise in May and the second in September. The Game Pass subscription had seen a dollar hike in 2024 along with added limitations, and salt was further rubbed on the wound in 2025 with a 50% raise with nary any benefits — a move that led to so many cancellations that the Microsoft website was reportedly having issues processing them.

This Game Pass price drop is the first big commercial move since Asha Sharma took over the division. The choice of executive was a move out of left field, as she was previously managing the Core AI business. Nevertheless, Sharma is reportedly spearheading an effort to revitalize the brand, and bringing down subscription prices to more reasonable prices is a decision that will definitely be appreciated by gamers.

Microsoft really wants users to embrace its Edge browser. Its latest effort involves unveiling a pleasingly rotund $2M prize pot to tempt users over. That figure includes a grand prize of $1M in cash, some alluring Mercedes-Benz automobiles, and “thousands of instant prizes.” Sadly for Microsoft/Edge, it has taken over a month for the media to become aware of this new barnstorming Rewards package with NeoWin flagging it as a surprise “new promotion in Edge” this Thursday.

Head to the rewards.bing.com link at the top of this article, and you will be asked to sign in to join the rewards program. There are T&Cs, privacy, and official rules smallprint links to digest. But basically, you will be agreeing to “receive emails about Microsoft Rewards, which include offers about Microsoft and partner products.” There is no monetary fee for entering the sweepstakes. But you can use Microsoft Rewards points balances to accrue more entries. The unfeasibly large maximum entry limit is 465 entries per person/account/email address/mobile phone number.

Sharing links to the prizes event page can also net you more entries and chances to get something from the prize pot. As this rewards program has been running for a month, we can see multiple folks across the world who have already won this and that. For example, someone in Spain won an Acer Copilot+ PC, an entrant from the U.S. snagged a $50 Xbox Gift Card, and someone from Mexico managed to lasso a ROG Xbox Ally handheld. Is the chance of this kind of prize worth the potential spammy email avalanche? That’s up to you.

The official rules say that the sweepstakes are open to U.S. residents, as well as those in Canada and a multitude of South American and European countries. We see New Zealand, Japan, and South Africa are listed, too. No, we don’t know what happened to Australia.

If you want to convert the $1,000,000 grand prize to your local currency – perhaps you are an optimist – Microsoft has already done the work for you. It is “approximately $1,380,000 CAD / £740,000 GBP / €860,000 EUR / Fr. 800,000 CHF / kr 10,100,000 NOK / $18,000,000 MXN / S/ 3,400,000 PEN / ₡498 000 000 CRC / Q7,000,000 GTQ / $1,730,000 NZD / ¥157,000,000 JPY / Kč 20,800,000 CZK / lei 4,350,000 RON / R 16,500,000 ZAR.” Some unfortunate countries will face Microsoft withholding 30% of the prize value in tax.

Three cars are up for grabs to appeal to the petrolheads among you. The rules suggest that winners will pop to their local Mercedes-Benz dealer, where a budget of $100,000 (or local equivalent) will be all yours to play with. You can’t buy a key fob and get the remainder in cash, though, so canny winners should max out their Merc specs to the last cent.

Installing this month's Windows Server security update has knocked some enterprise domain controllers into continuous reboot cycles, Microsoft confirmed in a release health dashboard entry.

The company says the April 2026 patch, KB5082063, triggers crashes in the Local Security Authority Subsystem Service (LSASS) on non-Global Catalog domain controllers used in Privileged Access Management (PAM) deployments, leaving Active Directory authentication and directory services unavailable on affected servers.

Microsoft's dashboard lists Windows Server 2016, 2019, 2022, 23H2, and 2025 as vulnerable to the fault, with the LSASS crash occurring during the startup sequence, which is what turns the failure into a loop: each automatic reboot re-enters the same faulty authentication codepath rather than recovering into a stable state.

The problem is only affecting managed enterprise environments that run PAM for Active Directory privilege delegation, and Microsoft said personal devices outside IT-managed domains aren’t exposed. The company hasn’t yet published a patch and has instead directed affected administrators to Microsoft Support for Business for mitigation guidance that can be applied if KB5082063 is already deployed.

KB5082063 now has three acknowledged bugs within a week of release, and Microsoft has warned separately that the same update prompts some Windows Server 2025 machines for a BitLocker recovery key after installation. The company is investigating reports that KB5082063 fails to install entirely on a subset of Windows Server 2025 systems.

April security updates have disrupted Windows Server domain controllers for three consecutive years. In March 2024, Microsoft shipped an emergency out-of-band fix after that month's Patch Tuesday caused DC crashes outright. The April 2024 patch cycle then broke NTLM auth across Windows Servers and forced unplanned DC restarts, which Microsoft corrected in a May 2024 rollout.

In June last year, the company released another correction for Active Directory authentication problems introduced by the April 2025 security update. This month's LSASS crash follows the same MO for the third year running: a general Patch release followed by post-deployment failure reports from enterprise admins, and a scramble for mitigation while the fix is prepared.

With KB5082063 still on the release channel and no patch date published, admins have three choices: delay the April update, isolate a test DC to validate patch behavior before wider rollout, or escalate through the Microsoft Support form Business to obtain the mitigation steps the company is providing case-by-case.

Microsoft’s Clippy was put out to pasture a quarter century ago. This hapless, and some would say ‘irritating,’ productivity assistant would no longer be enabled by default in Office, starting April 11, 2001. Nowadays, it is easy to remember Clippy with some fondness through rose-tinted retro spectacles. But, in its era, Clippy’s repetitive catch-all catch phrases such as “It looks like you’re writing a letter” and “Would you like help with that?” would soon erode any tolerance you might have for cute character-based digital assistants.

Clippy (more properly called Clippit) was a digital assistant introduced with Microsoft Office 97. The plan was to bring a friendly agent to the screen to interface with Office help content, as explained by Wikipedia. Several characters were designed to offer this help, with Clippy (Clipit) as the default choice. Among the alternatives were caricatures of Albert Einstein, William Shakespeare, and Rocky the dog, as well as several animated inanimate objects (like the unpopular paperclip).

Some esteemed figures in the computer industry think that the introduction of Clippy might have been a "tragic misunderstanding" of research conducted at Stanford University on breaking barriers in human-machine interaction. Indeed, there must have been something seriously wrong with a ‘helpful’ project like this for it to attract so much ire and ridicule among users and tech commentators.

As per our headline, Clippy was officially retired on April 11, 25 years ago, when Microsoft announced it would be disabled in Office by default. In Office XP, it would still be there as a dormant and optional feature. However, with Microsoft Office 2007, there was no longer any way to summon help from Clippy or his friends.

Rose-tinted retro spectacles

Clippy’s infamy has been sealed with its place in Time magazine’s 50 worst inventions. However, the mists of time have taken the edge off the pain of working with such a useless assistant, as it is now often viewed as being part of an amusing, heart-warming era in computing.

Microsoft has played on this softening of public opinion, or even nostalgia, for Clippy in several marketing campaigns since the animated paperclip and his friends were discarded. Most recently, it resurrected Clippy as an Emoji in Microsoft 365 – after overwhelming popular demand.

Last year, as the wave of new AI assistants began to grate on the public nerves, we also observed some fondness for Clippy being rekindled in a project by software engineer Felix Rieseberg – a locally hosted, LLM-based, AI-enhanced Clippy, complete with Office 97-era-appropriate UI.

Despite the sting of Clippy's clear failure, Microsoft keeps coming back to digital assistants as the future of computing. We had Windows Cortana foisted upon us from 2014 to 2023.

Now we have Copilot everywhere, in every corner of our Windows 11 PCs and Microsoft apps. A recent count indicates that there are at least 80, and probably over 100 Copilot apps…

As optimists, we hope Copilot will be reined in, as far as Windows goes, thanks to Microsoft’s latest stated initiative to focus on OS performance, reliability, and RAM usage. This telegraphed change is also supposed to lead to fewer Copilot interactions.

Microsoft's promises of a better Windows are moving one small step closer to fruition, starting with the Windows Insider Program. In a blog post authored by Microsoft product manager Alec Oot, the company promised a simpler channel structure and more control over which features Insiders get to try.

The channel structure is being simplified down to two primary channels. The previous Beta Channel will go into the more simply named "Beta" branch, while what was previously the Dev Channel will go the "Experimental." These are where you go for first access to the newest features. Though Oot explains that in the Experimental channel, "what you see may change, get delayed, or not ship at all."

If you're in the existing Canary Channel, things will get slightly more complicated, with 29500-series builds going to Experimental (Future Platforms), while 28000-series builds will transition to Experimental (26H1). Future Platforms will be the earliest preview build you can get, and one Oot writes that it is "not aligned to a retail version of Windows."

Features and gradual rollouts

On the Beta channel, there's another big quality of life change for Insiders. Microsoft is stopping gradual feature rollouts. If Microsoft announces a feature and you download the update, you'll get it. This differs from the previous practice of gradual rollouts. While the company says the gradual rollouts are designed to assess how features perform before releasing widely, they acknowledge that it made the program "unpredictable," and it could mean that "you don't get the new features that motivated many of you to join the Insider program to begin with."

For the Experimental channel, there will be flags for new features, so Insiders can enable the ones they want to try and disable those they don't.

Windows Insider for Business will see the same changes as the consumer program, while the Windows Server version will continue unchanged. Release Preview will hang around as an advanced option for those who want to try production builds shortly prior to their general release.

Upgrading in place

Moving between Windows Insider channels or leaving the program has been historically different, with very few points to stop and get off the ride without needing to wipe your PC and clean install Windows.

Microsoft now says it is working on the ability to hop between versions without losing all of your software and settings.

"This will allow, in most cases, Insiders to move between Experimental, Beta, and Release Preview on the same Windows core version, or leave the program without a clean install," Oot writes. "An [in-place upgrade] takes a bit more time than your normal update but migrates your apps, settings, and data in-place."

The Experimental (Future Platforms) build will still require a clean install, as it doesn't line up with any retail production builds of Windows.

Oot writes that these changes are set to begin in "the coming weeks."

Last month, Microsoft promised a slew of features to improve Windows 11, including a more flexible taskbar, a less aggressive Windows Update, improvements with RAM issues, and a better File Explorer. To assess upcoming fixes, Microsoft will need its slew of Insiders. Hopefully, a streamlined program makes it easier for enthusiasts to join the program and leave feedback before changes come to everyone else.

France is accelerating its digital sovereignty plans. In an official press release this week, the country’s DINUM announced its “exit from Windows in favor of workstations running on the Linux operating system.” (machine translation).

The DINUM is an important section of the French state, headed by the Interministerial Directorate for Digital Affairs, so this will mark a key changeover in the machinations of government, eliminating U.S.-based commercial interests from workstation computers. We assume some French flavor of Linux will be adopted to satisfy the stated objective of migrating to sovereign solutions. Joining the DINUM in this mission for digital sovereignty are France’s Directorate General for Enterprises (DGE), the National Cybersecurity Agency of France (ANSSI), and the State Procurement Directorate (DAE).

Moving to Linux is described as one of three “concrete initial steps” that have recently been committed to, to reduce France’s extra-European digital dependencies. The plan is expected to be formalized in the fall. By then, stakeholders should know what “workstations, collaborative tools, antivirus software, artificial intelligence, databases, virtualization, and network equipment” will be needed to move ahead with this digital sovereignty initiative.

On the topic of applications, not just the underlying OS, France recently announced that it had moved 80,000 National Health Insurance Fund employees to open-source alternatives to platforms like Microsoft Teams, Zoom, and Dropbox. These commercial platforms have been put into forced retirement by the new Tchap, Visio, and FranceTransfert services (and others), delivering a set of modern collaborative productivity tools dubbed La Suite.

Last month, the French government also “announced the migration of the health data platform to a trusted solution by the end of 2026,” says the source press release.

We know why – but why?

French Ministers are very keen on reducing reliance on technologies that depend on or are controlled by outside interests. “We must become less reliant on American tools and regain control of our digital destiny,” wrote David Amiel, a Minister of Public Action and Accounts, in a statement pinned to the above-linked PR. “We can no longer accept that our data, our infrastructure, and our strategic decisions depend on solutions whose rules, pricing, evolution, and risks we do not control.” It is interesting to see Amiel single out the U.S. for his statement about how the French state “must break free.”

Anne Le Hénanff, Minister Delegate for Artificial Intelligence and Digital Technology, echoed Amiel’s sentiments by asserting that “Digital sovereignty is not an option, it is a strategic necessity."

One wonders whether the growing cultural rift between the U.S. and its traditional European allies has added momentum to the digital sovereignty movement in France.
The implications for software and services businesses across the Atlantic don’t look great. As a leading member of the EU, France’s decisions and direction can exert a strong influence on others in the bloc. Moreover, if the move to Linux is seen as a success, it could also influence other government departments, and organizations that work closely with the government, and so on, all the way down to individual users.
Perhaps 2026 is set to be l’année de Linux?

The ROG Xbox Ally, one of the first Xbox handheld devices from Microsoft in collaboration with Asus, is now available at a $100 discount on Amazon and Best Buy. Launched at a retail price of $599, you can grab the ROG Ally X for $499, a price which we haven’t seen since January of this year and still matches one of the biggest drops since launch.

The ROG Xbox Ally and the ROG Xbox Ally X are more or less similar to the original ROG Ally and Ally X, with a slightly refreshed design including large Xbox controller-style contoured grips, and updated internals including AMD’s latest-gen Ryzen Z2 series chips optimized for handheld consoles.

• Check out this deal at Amazon

The ROG Xbox Ally is powered by the entry-level Ryzen Z2 A processor, which is the entry point into AMD's Ryzen Z2 lineup with a four-core/eight-thread CPU based on AMD’s Zen 2 architecture, an RDNA 2 GPU with eight cores, boost clock speeds of 3.8 GHz, and a configurable TDP range of 6W to 20W. In terms of memory, you get 16GB of LPDDR5X-6400 along with a 512GB M.2 2280 SSD, and a 60Wh battery.

The display on the ROG Xbox Ally measures 7 inches, offering a 1080p resolution, a 120 Hz refresh rate, 500 nits of brightness, and support for VRR (variable refresh rate) and FreeSync Premium. It is a standard IPS LCD panel, not OLED, but overall, in our testing, we found the display's performance to be fairly good and identical to the original ROG Ally.

The device is powered by Windows 11, but can directly boot into the Xbox Full Screen Experience, which feels significantly better for handhelds than standard Windows. It looks similar to the Xbox app, featuring a home screen with recently played games, along with access to your Game Pass and game library from various platforms like Steam, Epic Games, Battle.net, GOG Galaxy, and Ubisoft Connect.

At $100 off and $499 at Amazon (and Best Buy), the ROG Xbox Ally (see our Ally X review) is now a far more tempting buy than it was at launch. While it may not be the most powerful handheld on the market, it’s more than capable for casual and mid-tier gaming. If you’ve been waiting for a price drop to jump into handheld PC gaming, this deal makes a lot of sense.

If you're looking for more savings, check out our Best PC Hardware deals for a range of products, or dive deeper into our specialized SSD and Storage Deals, Hard Drive Deals, Gaming Monitor Deals, Graphics Card Deals, Gaming Chair, Best Wi-Fi Routers, Best Motherboard, or CPU Deals pages.

An artificial intelligence (AI) aficionado has put together a chart featuring all the Copilot things that Microsoft has released since AI became the next big thing. At the latest count, Ty Bannerman notes that there are 80 different, separately marketed Copilot products and tools. Charting these Copilot things wasn’t a trivial task; even Microsoft doesn’t appear to maintain a definitive list. When I first noticed this story, there were 78 Copilots in Bannerman’s charts, but now it has expanded to 80.

The AI strategy, design, and implementation expert says that the idea of charting the expanse of the Copilot universe came to him when someone asked what Microsoft Copilot is. He knew it meant at least 75 different things, in so many contexts, at the time. “Apps, features, platforms, a keyboard key, an entire category of laptops - and a tool for building more Copilots,” tallied Bannerman in his blog. “All named ‘Copilot’.” His chart contends this is "What happens when you name everything Copilot."

Last week, the AI aficionado charted the number of Copilots as 78. However, since yesterday, I note the number has increased to a nice round 80. Thanks to the power of the internet / social media, Bannerman had learned of the existence of Gaming Copilot and Microsoft Dragon Copilot. The latter of those isn’t designed for residents of Westeros, but an AI clinical assistant.

Copilot says there are ~95 to 120+ Copilots

So, we have a chart of 80 Copilots, and who knows what the final figure may be, and how many more Bannerman can uncover?

Since I’m typing on a laptop with a Copilot key, I prodded it and asked the thing itself. Who better to ask? The answer was that “the ecosystem is well north of 100,” if you include things like every app-embedded Copilot, enterprise, and Azure-adjacent tools, etc. I then asked it to add them all up, and it concluded there were “~95 to 120+ Copilots.”

Is that too many? Well, even on this PC, I was surprised to find two Copilot apps in my system tray a few weeks ago. One pops up the usual chatbot box, the other was actually Copilot 365, which, when clicked, asked me to sign in with my (non-existent) Microsoft 365 credentials before I could use it. It has been eliminated.

The corporation's promise of major improvements to Windows 11 performance, reliability, and fewer Copilot interactions can't come soon enough.

Several delisted Xbox 360 games briefly reappeared on the Xbox Store this week before being pulled again, according to the Better xCloud datamining account, which monitors Microsoft's cloud gaming backend for changes. This follows Xbox Vice President Jason Ronald's commitment at GDC 2026 in March to revive the company's backward compatibility program as part of Xbox's 25th anniversary later this year.

The titles spotted include Aegis Wing, Prince of Persia: The Sands of Time, and Mars: War Logs, all of which had been removed from the store previously. The Prince of Persia title resurfaced with its original release date, along with an apparent placeholder price of $100, and the 2003 shooter Armed and Dangerous also reappeared and vanished shortly before the latest batch. Mars: War Logs was delisted around the same time.

One or two accidental relists might mean nothing, but the pattern across four titles in quick succession has understandably caught attention. The Better xCloud account, run by the developer behind the browser plugin of the same name, first flagged Armed and Dangerous as a possible mistake before noting that subsequent appearances looked intentional.

At GDC 2026 in San Francisco, Ronald told attendees that the game preservation team would be "rolling out new ways to play some of the most iconic games from our past" as part of Xbox's 25th anniversary celebrations. Microsoft discontinued its backward compatibility program in 2021, citing licensing issues and technical limitations that prevented further titles from being added to the catalog.

The program originally launched in 2015 and brought hundreds of original Xbox and Xbox 360 titles to Xbox One and later Xbox Series X|S. At its peak, the catalog included over 600 titles spanning two console generations. Since the hiatus, fan demand for its return has remained high, with the community-run Xbox Game Preservation website tracking which Original Xbox titles players most want to see come back. Top-requested games on the site include Jet Set Radio Future, Sonic Heroes, and The Simpsons: Hit & Run.

All this raises questions about how backward compatibility will work on Project Helix, Microsoft's next-generation console powered by a custom AMD SoC. Microsoft has confirmed that Helix will play both Xbox console and PC games, and Ronald described the machine as delivering an "order of magnitude" leap in ray tracing performance at the same GDC presentation.

Speculation has centered on the possibility that Microsoft's preservation team is working to bring Xbox 360 emulation to Windows PCs, which would keep older titles accessible on both Helix and standard Windows 11 machines, but Microsoft hasn’t confirmed that.