Data Center Demand Driven by AI Innovation

Data Center Growth Is Accelerating—But It's What Sits Around the Racks That Wins the Margin The installed global data center capacity is projected to surge to 114.3 GW by 2025, growing at a +17.7% CAGR since 2021 (IEA). That translates to 485.4 terawatt-hours of electricity consumption—or 1.7% of the planet’s total demand. We’re seeing a fundamental reordering of digital infrastructure economics. What’s Driving It? Cloud: Enterprise migration is still in early innings. Gartner estimates that less than 50% of enterprise workloads have moved to the cloud. The runway is long. AI: McKinsey projects that AI workloads alone could require 50 GW of incremental capacity by 2030, adding more demand in five years than all of global hyperscale growth from 2015–2020 combined. Edge—or a logical shift to underserved metros: As Accenture notes, workloads and AI inference engines are driving demand into tier 2 and tier 3 metros, reshaping where capital needs to flow. 🧩 The Investment Insight: The Bottlenecks Become the Profit Pools Yes, installed capacity is rising rapidly—but capital is clustering in hyperscale deployments with increasingly compressed margins. The real margin opportunity is forming around the friction points: 1. Power availability and efficiency With many grids facing constraint, EY notes that renewable-backed and dispatchable power procurement strategies are becoming a strategic differentiator. Developers with energy expertise are now drawing infrastructure fund-level investments, not just REIT or data center capital. 2. Interconnection & last-mile fiber As workloads fragment and move outward, the physical and logical edge gains value. Dense interconnection hubs, metro fiber providers, and programmable routing intelligence are becoming supply-side moats. 3. Market ecosystems & orchestration platforms McKinsey highlights that fragmented value chains in digital infrastructure are creating "integration deserts". As quoting, fulfillment, and SLA management stretch across multiple providers, multi-party platforma and orchestration layers—akin to Amazon in e-commerce—are starting to centralize fragmented workflows. 4. Data intelligence & automation Accenture’s Infrastructure Vision 2025 identifies AI-powered operations and smart procurement systems as key value unlocks. Tools that simplify monetization and delivery will define the operating system for digital infrastructure. The Bigger Picture This isn’t just a bet on data centers—it’s a thesis on the unbundling and replatforming of digital infrastructure. The most compelling opportunities won’t be found solely in the four walls of a data center, or in the chips inside it. Instead, they’ll emerge from the data, software, and services layers that monetize and automate digital infrastructure at scale. I am excited for the ecosystem, there is value to be created at a massive scale over the next 5 years. #DigitalInfrastructure #AI #Cloud #DataCenters #ConnectedCommerce #Fiber

AI’s future depends on energy systems we haven’t built yet. I spent the first half of my career in the #energyefficiency and #renewableenergy field, working with electric utilities and in massive facilities including manufacturing plants, government sites, retail, hospitals, and resorts. In that work, my teams and I designed and installed energy systems for data centers, solar on rooftops and utility-scale farms, worked in geothermal power plants, built the world’s first utility-owned on-site co-generation system, and delivered the first fuel-cell energy solutions for the Department of Defense. The second half of my career has focused on IT and digital transformation. Today at AI Squared, I am seeing AI’s energy appetite firsthand. It is real, and it is growing faster than our infrastructure can handle. By 2030, data center electricity demand is projected to hit 945 terawatt-hours annually, up from 415 TWh today. That's almost as much electricity as the entire country of Japan uses in a year. AI workloads are the primary driver, with demand set to increase by 160 percent in just a few years. With quantum computing on the horizon, the ceiling could rise even higher. Yet only about 30 percent of global electricity today comes from clean sources. Most of the power fueling AI still comes from natural gas plants, not solar or wind. I had the opportunity to speak with Nicole Willing at Techopedia about this growing tension between AI innovation and energy reality. You can check out the full article here: 🔗 https://lnkd.in/eS7ERHH3 #AI #EnergyTransition #NetZero #SustainableAI #DataCenters #AISquared #InfrastructureMatters

The almost overnight surge in electricity demand from data centers is now outstripping the available power supply in many parts of the world, according to interviews with data center operators, energy providers and tech executives. That dynamic is leading to years-long waits for businesses to access the grid as well as growing concerns of outages and price increases for those living in the densest data center markets. The dramatic increase in power demands from Silicon Valley’s growth-at-all-costs approach to AI also threatens to upend the energy transition plans of entire nations and the clean energy goals of trillion-dollar tech companies. In some countries, including Saudi Arabia, Ireland and Malaysia, the energy required to run all the data centers they plan to build at full capacity exceeds the available supply of renewable energy, according to a Bloomberg analysis of the latest available data. By one official estimate, Sweden could see power demand from data centers roughly double over the course of this decade — and then double again by 2040. In the UK, AI is expected to suck up 500% more energy over the next decade. And in the US, data centers are projected to use 8% of total power by 2030, up from 3% in 2022, according to Goldman Sachs, which described it as “the kind of electricity growth that hasn’t been seen in a generation.” Read more here: https://lnkd.in/eQRdGmuW

So far this year, the U.S. data center landscape is shifting fast. As power constraints tighten in traditional hubs like Northern Virginia and Silicon Valley, secondary and emerging markets are stepping into the spotlight. Do you agree? Here’s what’s driving the shift and where the real opportunities may lie: 👉🏽 Key Trends: • Explosive Growth: U.S. data center inventory jumped 43% year-over-year in Q1 2025. Atlanta and Phoenix are now among the top four markets. • Hyperscaler Expansion: AI and cloud giants are pre-leasing space in non-traditional markets like Columbus, Des Moines, and Reno to secure power and speed to market. • Design Evolution: Liquid cooling and high-density racks are becoming the norm to support AI workloads. But what are the challenges? • Power Scarcity: Grid capacity is maxed in many metros. Transmission delays are pushing timelines into 2027 and beyond. • Construction Bottlenecks: Labor shortages, permitting delays, and rising material costs are slowing builds. • Cooling and Density Challenges: AI workloads demand new cooling strategies and facility designs, especially in retrofitted sites. So, where’s the opportunity? • Emerging Markets: States like Iowa, Indiana, and Ohio offer cheaper, more available power and faster permitting. This is ideal for hyperscaler and co-lo builds. • Behind-the-Meter Solutions: On-site renewables, gas peakers, and early-stage nuclear (SMRs) are gaining traction to bypass grid delays. • Sustainable Infrastructure: Clean energy integration is now a competitive edge. Why does this matter? AI infrastructure here in the U.S. is being built in places we’re only beginning to notice. Maybe even in a county near you! It’s a dynamic space, and I’m keeping a close eye on the trends and opportunities. What are you seeing in your region or sector? Let me know! 👇🏽

🚨 One AI Model. More Power Than an Entire State. 🤯 Anthropic just dropped this: 👉 By 2028, a single frontier AI model could require a 5GW data center to train. That’s more electricity than Massachusetts, Colorado, or Iowa use in a whole year. (Yes—one model. Not the industry. Not inference. Just training.) And they’re not alone. Between OpenAI, Meta, xAI, Google, and others… 🧠 Total frontier AI training demand in the U.S. is projected to hit 20–25GW by 2028. 💡 Translate that: • 1 model = 40 TWh/year • 5GW = more than 40 states’ entire residential demand • Massachusetts: 50M MWh/year • Colorado: 55M MWh/year • Iowa: 54M MWh/year • Utah: 33M MWh/year • Rhode Island: 7M MWh/year • AI will soon be one of the largest sources of new power demand in the U.S. economy 🔥 This isn’t just about AI anymore— It’s a new industrial sector with grid-scale energy demands. We're witnessing the birth of compute infrastructure that's: • Bigger than entire state economies • Heavier than manufacturing • More power-hungry than steel or chemicals The U.S. AI sector will need 50GW+ of electric capacity by 2028 to stay ahead. 📌 Now imagine the ripple effects: • Transmission buildouts • Permitting reform • Energy procurement wars • Edge vs. core siting strategies • And a whole new breed of power-aware model design Welcome to the era of AI Energy Realism.

Artificial Intelligence's rapid growth is not just a trend, it's a force that is driving up electricity demand, which is already challenging the power grid and tech companies. The strain is real and immediate. The boom in Artificial Intelligence is leading to a significant increase in electricity usage, putting a strain on the already stressed power grid. From simple ChatGPT queries to complex AI-generated images and videos, the demand for power is escalating rapidly. Data centers, which consumed more power than entire countries in 2023, are at the forefront of this surge. Experts predict that if AI's power needs continue to grow at this rate, it could potentially outpace the grid's capacity, leading to a significant increase in reliance on non-renewable energy sources, a scenario that should raise concerns. ⚡ Soaring Electricity Consumption: Even simple AI tasks, like ChatGPT queries, consume significant power, equivalent to a 60-watt bulb running for 10 minutes, highlighting the intensive energy needs of AI technology. 🌍 Massive Data Center Demand: In 2023, data centers used more electricity than nations such as Italy and Taiwan. Their energy demand has surged over seven times since 2008 despite advancements in energy-efficient chips. 📈 Projected Growth: According to the Boston Consulting Group, data centers' power consumption could rise to 7.5% of the global total by 2030, tripling from current levels. This could overwhelm existing power generation capacities and strain renewable energy sources. 🌪️ Regional Vulnerabilities: In regions like Texas, which experienced deadly blackouts in 2021, the rising energy demands from AI data centers and crypto miners could lead to grid instability and increased risk of outages. ♻️ Energy Source Challenges: While tech companies aim to use green energy, the high consumption by data centers often exhausts available renewable resources. This forces power providers to rely more on non-renewable energy sources to meet overall demand. #AIBoom #ElectricityDemand #PowerGrid #DataCenters #RenewableEnergy #TechIndustry #EnergyConsumption #AIGrowth #SustainableTech #EnergyChallenges 

The AI Power Paradox: Why Energy Will Determine Who Wins the Next Computing Revolution ⚡ 🧠While everyone debates which AI model will dominate, I’m watching a brewing crisis that could determine the real winners: power infrastructure. ⛔️ Here’s the problem: AI #datacenter power demand could surge more than 30x by 2035, but America’s largest grid operator just delivered a stark warning—“There is simply no new capacity to meet new loads,” said Joseph Bowring president of Monitoring Analytics, the independent watchdog for PJM Interconnection. The numbers tell the story: 📈 • Goldman Sachs Sachs forecasts global data center power demand will increase 165% by 2030 • U.S. Department of Energy (DOE) projects data center load could double or triple by 2028 • #Data centers currently consume 1-2% of global power, rising to 3-4% by decade’s end • Between 2024-2025, data center power usage accounted for $9 billion (174%) of increased power costs in #PJM territory alone. This creates a strategic shift: The companies building #AI infrastructure must now become energy companies. EQT Corporation CEO calls America’s permitting delays a “geopolitical liability”—while we’re stuck in regulatory quicksand, #China accelerates AI infrastructure deployment unconstrained. The AI gold rush is an #energy arms race. What’s your take on the energy-AI nexus? Love to hear your thoughts below ⬇️ 💭 #Nuclear #Utilities

"Here’s the thing about AI technology: It requires an enormous amount of energy to develop and run. And that’s where utilities come in. “Power demand from data centers has already been humongous, then came the AI hype and the need for power skyrocketed,” said Manju Naglapur, senior vice president and general manager for cloud, applications and infrastructure solutions at Unisys Corp. “With all the money spent on data centers, the power consumption will increase massively.” Across the US, utilities are preparing for historic increases in electricity demand led by data centers and AI. Even outside Data Center Alley in Northern Virginia, where Dominion Energy Inc.temporarily paused new data center connections in 2022 due to grid constraints, the companies are planning new power plants and transmission lines. Artificial intelligence is poised to help drive a 900% jump in power demand from data centers in the Chicago area, which will potentially require as much electricity as around four nuclear power plants can produce, Exelon Corp. Chief Executive Officer Calvin Butler said recently. Of course, all of this demand can only benefit utilities if they can produce the electricity to meet it. Many energy experts are concerned that the US power grid isn’t prepared to handle the wave coming its way. And that has some investors turning to the companies that will be brought in to strengthen the grid so utilities can adapt to the new high-energy environment." https://lnkd.in/gmEQJZbC

AI data center development is expected to accelerate until the end of the decade, but bottlenecks in data center and GPU deployments are just emerging. Morgan Stanley now forecasts a 23% annual increase in global data center capacity until 2030, with the U.S. at the forefront. However, the main challenge isn’t demand; instead, supply faces ongoing limitations due to government regulations and international supply chain issues, including power, machinery, and materials. Energy remains the main constraint, but it is not the only limiting factor. Permitting processes are slowing down, and building materials are becoming increasingly scarce and costly. Construction expenses are projected to increase by 11% each year, as reported by Morgan Stanley. To meet increasing demand, operators are considering two main options: converting existing Bitcoin mining farms and constructing facilities near nuclear or gas power plants. Sectors and regions such as the Middle East and China, which already have advantages in the data center supply chain, are expected to attract growing strategic interest. Elon and xAI just announced another data center project in Saudi Arabia. Meanwhile, hyperscaler capital expenditures continue to hit new records. Meta has increased its full-year revenue forecast to $64–72 billion. The four largest cloud providers are expected to spend a combined $1 trillion annually by the end of the decade. Additionally, NVIDIA’s GB200/300 order book continues to be revised upward. The worldwide AI infrastructure race is underway. The only true limit is how quickly we WANT to build.