Data Center Investment Trends and Insights

Traditional CRE underwriting for data centers is dead. Industrial comps. Flex frameworks. Rent per square foot. Here are 6 things you need to know according to the country's most experienced data center developer: Daniel English is Managing Partner at Legacy Investing, where he's developed data centers for 15+ years. We spoke to him and Ivo van Breukelen from The Proptech Connection about what matters in this space. Here are 6 shifts investors must understand: 1/ Data centers are civic infrastructure now: Cities evaluate them like ports and power plants. Planning departments look at grid impact before land use. "They were always considered a niche asset class, like a marina or an RV campground," says English. But not anymore. 2/ The market is splintering: Hyperscale campuses in rural areas still dominate headlines. But AI inference needs low latency, not cheap land. That's pushing 10-50MW facilities into urban and metro buildings. Legacy is converting a Chicago office building into a 30-50MW AI center. There's a "missing middle" of these mid-sized metro facilities that's emerging. 3/ Power is the constraint: The question used to be: do you have the land? Now it's: can you secure power on a real timeline? English saw a developer misread utility sequencing. The tenant realized power wasn't coming on schedule and walked. "They didn't do their load letters right and wiped out $60M of stabilized value with two sentences." 4/ Governments control supply: Northern Virginia recorded an 80% increase in under-construction capacity in H1 2025. Dublin froze development and data centers now draw 21% of Ireland's electricity. Amsterdam's moratorium protects power for housing. The same 40MW project can be welcomed or blocked depending on the jurisdiction. 5/ Easy money is gone, but yields vary: Compression is concentrated in long-term credit-backed hyperscale deals. Outside that segment, yields remain variable. "Where you make your best money is when the markets are opaque," English says. 6/ Expertise is the moat: Most LPs can't evaluate transformer lead times, cooling topology, or interconnection strategy. "There's a huge divide between LPs that are doing data centers and those that aren't," says English. "And I'm not seeing much overlap." The investors winning are bringing operating partners and engineers into the process early. Everyone else is bidding on land with "data center potential" and wondering why they never close. Full letter linked in comments.

Data centers, compute, and energy have become a bottleneck and a cash cow. Companies that once discussed software margins in earnings calls now debate cooling technologies and power procurement. The numbers tell a story of infrastructure at an inflection point: Data centers consuming 460 TWh in 2022 (pre-ChatGPT) will exceed 1,000 TWh by 2026 and the global data center market size is projected to approach $1T within 7 years. Behind every earnings mention are two core realizations: 1) Data center infrastructure is foundational to everyone’s AI aspirations Meta increased CapEx by $5B to $72B, citing "substantial internal demand for GPU resources." Microsoft warns AI demand will exceed supply through 2025. Dell raised AI server guidance to $20B. Google is acquiring stakes in crypto miners for GPUs. Hyperscalers are going nuclear with Google signing with Kairos Power for 500MW, Amazon buying Talen Energy's 960MW campus, and Microsoft partnering with Constellation. Every tech giant's earnings call now reads like infrastructure procurement because when one GPT query burns 10x the energy of a Google search, training frontier models requires city-scale power, and AI ambitions die without compute. 2) There is a SH*T TON of money to be made across the data center value chain CyrusOne raised $9.7B specifically for AI infrastructure. Blackstone paid $1B for a Pennsylvania gas plant. Traditional utilities like PPL now build generation exclusively for data centers. Power isn't infrastructure anymore – it's the business model. Cooling specialists like Submer and Green Revolution tackle 300% power increases from new chips. Edge players like Armada can deploy modular centers anywhere. AI-native infrastructure companies like VAST Data ($9.1B valuation) rebuild the stack from scratch. Nscale raised $1.1B in another play from crypto miners turned infra provider. The gold rush extends everywhere, with NVIDIA projecting that the AI infrastructure market will hit $4 trillion by 2030 and a $1T+ buildout underway – every layer of the stack is capturing value. And... with that… coming soon… the full CB Insights’ data center value chain report.

We are living through a fundamental reshaping of infrastructure investment priorities. This chart says it all: in June 2025, U.S. data center construction ($40.1bn) is nearly at parity with general office construction ($44.2bn). A decade ago, office space dwarfed digital infrastructure by more than 10:1. Today, the curves are about to cross. 💡 The implication is clear: • Offices represent yesterday’s economy—centralized, physical, and often underutilized post-COVID. • Data centers represent tomorrow’s economy—digital, decentralized, and powering AI, cloud, and the tokenized world. For investors, this shift raises an important strategic question: how do we allocate capital in an economy where compute is the new real estate? Some reflections: 1️⃣ Digital Infrastructure as Core Real Assets – Data centers, fiber, and energy grids are becoming as critical as roads and airports once were. Expect infra funds to overweight digital assets. 2️⃣ Power & Cooling Bottlenecks – Compute demand is constrained by electricity and water. The most attractive investments may not be in the data centers themselves but in the enabling energy and grid infrastructure. 3️⃣ Location Arbitrage – Offices were built in central business districts. Data centers will cluster near cheap power, connectivity hubs, and tax incentives. Think Virginia, Texas, Nordics—not Manhattan. 4️⃣ Financial Engineering – REITs and infra funds will increasingly spin off digital platforms. We may see the rise of “Digital Infrastructure REITs” replacing traditional office REITs as defensive income plays. 🔮 My position: Investors should treat AI and compute infrastructure as the 21st-century equivalent of railroads. We’re still in the early innings, and the capital intensity is massive. 👉 The real question: Are you still investing in yesterday’s offices, or are you reallocating towards tomorrow’s compute economy? #Infrastructure #Investing #AI #RealEstate #DigitalEconomy #Strategy

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

The global data center boom is one of the defining investment themes of our time—but how real is it, and are we heading toward the next “dot-com” bubble? The honest answer: it’s both real and, in parts, overheated. On the demand side, the shift is undeniable. AI is driving a structural step-change in computing needs. Unlike the late 1990s, this demand is already monetized—through cloud services, enterprise applications, and increasingly through AI-driven workflows embedded across industries. This is not speculative demand waiting to materialize; it is here, scaling rapidly. However, on the supply side, we are beginning to see signs of excess. Developers are racing to secure land, grid connections, and capacity—sometimes without firm customer commitments. In certain markets, projects risk being built ahead of actual demand or without guaranteed access to power. The result: pockets of overbuild and inefficiency. The real constraint, interestingly, is no longer capital—it is energy. Power availability, grid infrastructure, and equipment bottlenecks are becoming the critical limiting factors. In some regions, data center demand is already outpacing grid expansion, forcing operators to rethink how and where they build. This is where the comparison to the dot-com era falls short. Back then, infrastructure was massively overbuilt ahead of demand. Today, we face almost the opposite dynamic: real, immediate demand meeting constrained and uneven infrastructure expansion. The risk is not that the entire sector collapses, but that capital is misallocated in the process. What does this mean going forward? We are likely heading into a phase of consolidation and discipline: - Projects without secured power or customers may not materialize - Prime locations with reliable energy access will become even more valuable - Large, well-capitalized players will continue to strengthen their position In short, this is not a bubble that bursts—it is a market that matures. The opportunity remains significant, but success will increasingly depend on one key factor: Access to power. In the age of AI, electrons are as strategic as algorithms.

🚀 Data Centers Are Becoming the New Oil Fields – The Numbers Are Staggering The global race to build AI capability is reshaping the #energy and infrastructure landscape faster than anyone expected. In 2025 alone, the world is set to invest USD 580 billion in new data center capacity – more than the total global investment in #oil supply. Yes, you read that right. Data centers are now out-investing oil. Why? Because the surge in large-scale AI models demands an unprecedented expansion of compute #power. And that expansion comes with major implications for energy systems, grids, supply chains, and urban infrastructure. This signals a seismic shift in the drivers of the modern economy: digital infrastructure is now just as essential as energy itself. 📊 A few standout insights from the latest pipeline analysis: ·      🇺🇸 🇨🇳 🇪🇺 ~82% of all global data centre capacity is concentrated in the U.S., China and Europe ·      Over 85% of new capacity additions in the next few years are in the same three regions ·      55% of upcoming data centres exceed 200 MW, each consuming the annual electricity of 200,000 households ·      Nearly two-thirds of these new sites are built in or near existing urban clusters – adding strain to already congested grids ·      Connection queues are becoming a critical bottleneck: o  1–3 years in the U.S. (up to 7 years in Northern Virginia aka as Data Center Alley) o  7–10 years in parts of Europe o  Dublin has paused new connection requests until 2028 ·      Supply chains for transformers, turbines, cables, and critical minerals are under severe pressure –putting 20% of projected new capacity by 2030 at risk of delay 💡 What this means: We are entering an era where compute availability becomes as strategic as energy availability. Data centres are no longer a “tech sector issue” – they are rapidly becoming a defining feature of national energy policy, industrial strategy, and urban planning The AI revolution is not just digital. It’s physical. It’s accelerating It’s geographic. It’s energy-intensive. and it’s about where and how that compute can be powered. What are your thoughts? Are we ready for the next wave of digital infrastructure? What policies or innovations can keep pace with data center growth? 🔍 At SDI, this is exactly where we help. Our work supports governments, utilities, and technology companies to: ✔️ Assess grid readiness and identify constraints before they become bottlenecks ✔️ Evaluate optimal locations using geospatial, cluster and infrastructure analysis ✔️ Navigate permitting and regulatory timelines ✔️ Understand supply chain risks for transformers, cables, turbines and critical components ✔️ Build realistic, data-driven roadmaps for scalable AI-age infrastructure 💬 If this outlook resonates, challenges your perspective, or sparks ideas – share your thoughts below! Please like, comment, and repost to keep our conversation going. 📊 Source: IEA and SDI Management Consulting

🚀 Excited to share that our latest Data Center Physical Infrastructure (DCPI) 5-Year Forecast has just been published at Dell'Oro Group — and the numbers tell a compelling story. 📈 The worldwide DCPI market is projected to grow at a mid-teens CAGR through the end of the decade, surpassing $80 billion by 2030. But what's really interesting isn't just the size of the opportunity — it's the structural transformation happening underneath. ⚡ Power scarcity has shifted from a background concern to a binding constraint. Grid interconnection delays and transmission bottlenecks are pushing operators toward on-site generation — natural gas turbines and reciprocating engines — not as a backup plan, but as a practical necessity to bring multi-hundred-megawatt and gigawatt-scale AI campuses online. 🌡️ Cooling is no longer a supporting function — it's becoming a primary determinant of data center design. Thermal management is projected to grow at a 20% CAGR and reach the value of the UPS market by 2030. Direct liquid cooling alone is forecast to surpass $8 billion, especially as NVIDIA's latest Vera Rubin compute tray designs go fully fanless with liquid-cooled configurations. 🏗️ The capex story is equally telling: service providers — cloud, colocation, and neo cloud operators — are forecast to grow DCPI spending at roughly 5x the rate of enterprise. The complexity and capital intensity of AI-ready infrastructure is reshaping who spends, how much, and on what. This is a market in the middle of a profound architectural shift, and I'm proud of the work our team has done to size and forecast where things are headed. 📰 https://lnkd.in/eGRV25K6 💬 Want to dig deeper into the data behind these trends? Feel free to reach out — I'm always happy to discuss how the DCPI landscape is evolving and how Dell'Oro Group's research can help you stay ahead of the curve. #DataCenter #DataCenterInfrastructure #AI #LiquidCooling #PowerInfrastructure #HyperscaleDataCenters #MarketResearch #DellOroGroup #DCPI #EnergyTransition #DataCenterDesign

Amazon Web Services (AWS) has announced plans to invest approximately $200 billion in capital expenditures in 2026, marking one of the largest infrastructure spending commitments ever made by a single company. The majority of this investment will be directed toward Amazon Web Services, with a clear focus on expanding data center capacity, AI infrastructure, custom silicon, and the power systems required to support them. AWS continues to report strong growth, with cloud revenue expanding at more than 20 percent year over year, yet leadership has been explicit that capacity remains constrained. New data center capacity is being absorbed almost immediately as it comes online, particularly by AI training and inference workloads. This dynamic has driven Amazon to accelerate investment well beyond historical levels. This scale of spending highlights several structural realities shaping the industry. First, demand for compute continues to outpace supply. Even after years of aggressive buildout, hyperscale data centers remain capacity-constrained, reinforcing that AI and cloud demand is not cyclical but structural. Second, data centers are now core revenue infrastructure. Unlike previous technology cycles where infrastructure was treated as a cost center, today’s data center investments directly underpin AI platforms, enterprise cloud services, and proprietary chip ecosystems. Capital deployment at this scale signals long-term confidence in sustained utilization. Third, the investment reflects a broader understanding that power, cooling, and connectivity are now the primary bottlenecks, not just silicon availability. Significant portions of this capital will be allocated to electrical infrastructure, grid interconnections, onsite generation, and advanced cooling systems. Finally, the macro impact is difficult to overstate. Amazon’s planned capital spending rivals the annual investment budgets of entire industrial sectors. When combined with similar commitments from other hyperscalers, total annual investment into AI and data center infrastructure now measures in the hundreds of billions of dollars globally. While markets may focus on near-term margin pressure, the longer-term message is clear. The next decade of technological growth will be determined by who can deploy physical infrastructure at scale, secure power reliably, and sustain capacity ahead of demand. Amazon’s commitment signals that the race for AI leadership is, at its core, an infrastructure race.

Data center inventory in North America grew by a record 43% over the past 12 months, but the region’s vacancy rate remains near historic lows at just 4.2%, according to new data from the CBRE Global Data Center Trends report for 1Q 2025. In other words, developers are deploying more new data center space than ever, but are still unable to keep up with demand, according to CBRE Data Center Solutions. “Rising demand from AI and hyperscale users is shrinking vacancy and operators with available capacity in key markets are commanding premium rates,” said Pat Lynch, executive managing director for CBRE’s Data Center Solutions. Despite ongoing power supply challenges, Northern Virginia remained the largest market, expanding its inventory by 523 megwawatts (MW) over the past 12 months, with a vacancy rate of just 0.76%. The average rental rate increased by 15%, driven by preleasing of facilities scheduled for delivery as far out as 2028. The Top 5 U.S. markets continued to evolve, as Atlanta and Phoenix are now the second and third largest data center markets in North America, surpassing Dallas and Silicon Valley, now fourth and fifth. “Power constraints in legacy markets are forcing hyperscalers to seek new frontiers for development, spreading workloads across multiple smaller locations with faster power availability timelines,” said Gordon Dolven, Director of CBRE Americas Data Center Research. “While improvements in fiber connectivity have reduced latency concerns and supported this shift, it’s power that ultimately determines where infrastructure can scale.” Here's a chart that illustrates the supply issues: Of the 7 markets with the lowest vacancy rates, all but Singapore are in the Americas. Read the report here: https://lnkd.in/eW3aVhQP

The AI race will drive a fierce data centre land rush. In 2026, the scarcest resource in AI won’t be talent or chips – it will be land for data centres. Industrial land prices for data centres are already surging worldwide. Northern Virginia, the world’s largest data centre hub, now sees plots trading above $6m (£4.4m) per acre. Europe is following suit, with hyperscalers like NVIDIA committing over €50bn to new real estate projects. In the Nordics, demand could quadruple by 2032, fuelled by low energy costs, cool weather and available land. A Deloitte report ranks data centers as a top real estate opportunity for 2026. In nine major global markets surveyed, 100% of new construction is already pre-leased. "AI infrastructure is the foundation that will support the growth of the new economy,” wrote Victor Arnaud, president of Equinix Brazil, announcing a million-dollar investment in clusters for São Paulo and Rio. This land rush will bring ripple effects – commercial zones, housing markets and tech jobs will likely surge near AI hubs. But environmental concerns loom. In parts of Latin America, drought-stricken regions are courting thirsty data centres, sparking alarm among environmental groups and residents. And for investors, the risks are high. Venture capitalist Rahul Mathur warns that tech depreciation means you can’t model cashflows like a rental asset. “Data centres aren’t typical real estate," he says. For professionals, keeping an eye on these dynamics could be critical – and lucrative. “If I were 25 today, what business would I get involved in? l'd focus on two massive opportunities: Al implementation and data centre development,” businessman Kevin O'Leary says in a LinkedIn video. “This is where the future's heading.” ✍ Lucas Carvalho 📷 Getty Images 💡 This is one of a several ideas LinkedIn News is highlighting in our annual list of predictions. Read it here: https://lnkd.in/BI26UnitedKingdom Join the conversation in the comments or share your own prediction in a post or video with #BigIdeas2026.