The infrastructure backbone of the AI revolution is straining under unprecedented demand. Data centers, once the invisible plumbing of the digital economy, have emerged as one of the most capital-intensive and resource-constrained sectors in technology. For venture investors focused on the built environment, this creates both urgent challenges and extraordinary opportunities across cooling technologies, energy systems, and an emerging frontier that once seemed relegated to science fiction: orbital computing.

The opportunity has not gone unnoticed by the venture community. From specialized space-focused funds like Space Capital, Lux Capital, and Seraphim to generalist powerhouses including Andreessen Horowitz, Khosla Ventures, Founders Fund, and General Catalyst, capital is flowing into data center infrastructure and adjacent space technologies at an accelerating pace. Corporate strategic investors—Airbus Ventures, NEOM Investment Fund, and the venture arms of major hyperscalers—are placing bets across the value chain. In 2025, private investment in the broader space sector grew 48% year-over-year to $12.4 billion (Reuters, 2026), with data center-enabling technologies capturing a growing share of that capital.

The scale of growth is staggering. According to the International Energy Agency, global data center electricity consumption is projected to grow by approximately 15% annually through 2030, more than four times faster than total electricity consumption from all other sectors combined. Gartner estimates worldwide data center electricity consumption will rise from 448 terawatt-hours (TWh) in 2025 to 980 TWh by 2030, with AI-optimized servers driving nearly all incremental growth. The United States alone saw data centers consume 183 TWh in 2024, roughly equivalent to the annual electricity demand of Pakistan.

The main culprit is artificial intelligence, although demand had already grown during the cloud and crypto cycles. AI-specific servers consumed between 53 and 76 TWh of electricity in U.S. data centers in 2024, enough to power more than 7 million homes. Training a single frontier model like GPT-4 consumed an estimated 50 gigawatt-hours of energy. The Stargate initiative announced by OpenAI and President Trump aims to spend $500 billion on as many as 10 data centers, each potentially requiring 5 gigawatts of power—more than the entire state of New Hampshire. The numbers are becoming difficult to fathom.

This explosive growth is colliding with grid infrastructure that was never designed for such concentrated demand. BloombergNEF forecasts U.S. data center power demand will more than double by 2035, rising from 35 gigawatts in 2024 to over 100 gigawatts. In advanced economies, data centers are projected to drive more than 20% of electricity demand growth through 2030, putting power systems back on a growth trajectory after years of stagnation.

The costs are already hitting consumers. In the PJM electricity market stretching from Illinois to North Carolina, data centers accounted for an estimated $9.3 billion price increase in the 2025-26 capacity market. Wholesale electricity prices have risen as much as 267% since 2020 in areas near significant data center activity. One Carnegie Mellon study estimates data centers and cryptocurrency mining could lead to an 8% increase in average U.S. electricity bills by 2030, potentially exceeding 25% in high-demand markets like northern Virginia.

The constraints are not merely inconvenient; they are becoming existential for the AI buildout.

Energy is only part of the equation. Data centers have developed an insatiable thirst for water. A medium-sized facility can consume 110 million gallons annually for cooling, equivalent to 1,000 households. Larger facilities can drink 5 million gallons per day, matching the water needs of a town of 50,000 people. Projections suggest water consumption for cooling may increase by 870% as more facilities come online. The Environmental and Energy Study Institute estimates U.S. data centers consume 449 million gallons of water per day.

This resource intensity has triggered growing community opposition. According to Data Center Watch, over $64 billion worth of data center projects were blocked or delayed between H2 2024 and H1 2025 due to local resistance, with $18 billion outright blocked. The report identified 142 grassroots organizations actively opposing projects across 28 states. In Warrenton, Virginia, residents voted out all town council members who supported Amazon's proposed data center. In Prince William County, a $24.7 billion project led by QTS and Compass Datacenters remains mired in lawsuits. Data centers have become the new NIMBY flashpoint.

A Heatmap poll found that only 44% of Americans would welcome a data center in their community, lower favorability than wind farms, solar installations, or even nuclear facilities. The industry that was once invisible infrastructure has become politically contentious.

Fortunately, a wave of innovation is emerging to address these constraints. 2025 has unequivocally become the year liquid cooling went mainstream. Leading-edge GPUs and AI chips now generate heat loads that traditional air cooling simply cannot manage. The average data center rack power density increased 38% from 2022 to 2024, with AI clusters pushing densities from 15 kW to 80-120 kW per rack. Liquid cooling offers power usage effectiveness (PUE) scores consistently below 1.2, compared to 1.4-1.6 for air-cooled facilities.

Direct-to-chip (D2C) cooling has emerged as the dominant architecture for hyperscalers. Cold plates mounted directly onto CPUs and GPUs can remove 70-80% of heat loads at the source. Microsoft's advanced AI supercomputer, unveiled in 2025, features exclusively liquid-cooled racks. NVIDIA and AMD now design flagship GPUs with liquid cooling as the preferred thermal management solution.

M&A activity reflects the sector's momentum and the strategic imperative for incumbents to acquire liquid cooling capabilities. Eaton's $9.5 billion acquisition of Boyd Thermal represented one of the largest deals in the thermal management space, signaling that industrial conglomerates view cooling as critical infrastructure. Trane's purchase of Stellar Energy Digital and Carrier's acquisition of ZutaCore for its HyperCool waterless direct-to-chip platform further underscore the consolidation trend. Schneider Electric's $850 million acquisition of Motivair in October 2024 added cold plate technology and CDU manufacturing to its data center portfolio. We expect M&A activity to accelerate as hyperscalers and infrastructure operators seek to vertically integrate cooling capabilities.

Immersion cooling represents the next frontier in thermal management. Submerging entire servers in dielectric fluid enables two-phase systems to handle heat flux up to 1,500 W/cm², versus just 16 W/cm² for single-phase water cooling. The global immersion cooling market was valued at $4.7 billion in 2026 and is projected to reach $18.1 billion by 2032, yielding a 24.9% CAGR over a mere six-year period (Yahoo Finance, 2026). Companies like Submer, GRC, and Asperitas are deploying production-scale solutions, while TSMC's Direct-to-Silicon Liquid Cooling technology promises to cool chips dissipating 2 kW of power with less than 10W of pump power.

Startups Doing Big Things in Cooling

If terrestrial constraints are becoming insurmountable, perhaps the solution lies beyond Earth entirely. This once-fringe idea has rapidly gained credibility among the industry's most powerful players. Jeff Bezos has stated he believes gigawatt-scale data centers will operate in space within 10-15 years. Elon Musk has confirmed SpaceX "will be doing" data centers in space. Sundar Pichai has declared that "a decade or so away, we'll be viewing it as a more normal way to build data centers."

The thesis is compelling: data centers already consume an immense share of electricity and are projected to account for nearly half of U.S. electricity demand growth over the next five years (IEA base case). They strain local grids, swallow land, and extract billions of gallons of water. Orbital data centers could be powered by unlimited, uninterrupted solar energy, with 30-40% higher irradiance than Earth's surface and near-continuous exposure without day/night cycles or weather interference. Cooling could leverage the vacuum of space as an infinite heat sink, using passive radiative cooling to dissipate waste heat without water.

On an operating level, these systems should theoretically deliver dramatically lower energy costs. Starcloud estimates orbital data centers could offer energy costs 10x cheaper than terrestrial alternatives, even accounting for launch expenses. A European Commission study concluded that orbital data centers could significantly reduce greenhouse gas emissions from grid electricity and eliminate freshwater usage for cooling entirely.

The milestones achieved in late 2025 have transformed this from speculation to demonstration. In November 2025, Nvidia-backed startup Starcloud launched Starcloud-1, a 60 kg satellite carrying an Nvidia H100 GPU, aboard a SpaceX rocket. This was the first AI chip 100x more powerful than any previously operated in space. The satellite successfully trained NanoGPT (a large language model created by OpenAI founding member Andrej Karpathy) using Shakespeare's complete works, and ran Google's Gemma model in orbit. "Anything you can do in a terrestrial data center, I'm expecting to be able to be done in space," CEO Philip Johnston told CNBC.

Starcloud envisions a 5-gigawatt orbital data center with solar and cooling panels measuring approximately 4 kilometers in width and height. Such a compute cluster would exceed the capacity of the largest power plant in the United States. The company plans to launch Starcloud-2 in October 2026, featuring 100x the power generation of Starcloud-1, with a module running a cloud platform from Crusoe that will allow customers to deploy AI workloads from space.

Google has entered the arena with Project Suncatcher, announcing plans to launch two prototype satellites by early 2027 in partnership with Planet Labs. The project will test Google's TPU chips in orbit and validate high-bandwidth optical inter-satellite links for distributed machine learning tasks. Google's research envisions 81-satellite clusters operating in dawn-dusk sun-synchronous orbits, with spacecraft flying just hundreds of meters apart to enable multi-terabit communication. Radiation testing of Google's Trillium TPU v6e showed it could withstand nearly three times the expected five-year mission dose without permanent failures.

The Challenges and Investment Implications

While the excitement and strategic sponsorship from tech titans indicates genuine potential, it is imperative that investors navigate through the noise and avoid blindly jumping on the bandwagon. Investing in orbital data center technology is undeniably a long-term play with significant feasibility and viability risks that venture investors must weigh carefully.

Launch costs remain the critical economic barrier. Current launch costs average $1,500-3,000 per kilogram to low Earth orbit. Google's Suncatcher team estimates costs must fall below $200/kg by the mid-2030s for orbital data centers to achieve cost parity with terrestrial facilities (Google, 2025). SpaceX's Starship promises to slash costs dramatically, potentially reaching $10-20/kg at scale, but this remains unproven at commercial volume.

Computing hardware must withstand extreme radiation. High-energy particles can corrupt data and damage chips. Radiation hardening currently adds 30-50% to hardware costs and reduces performance by 20-30%. Error-detection and correction software, along with physical shielding, are necessary, adding mass and complexity.

Thermal management presents fundamental physics challenges. In vacuum, heat can only be dissipated through radiation, requiring large radiator arrays that add significant mass. NASA studies show radiators can account for more than 40% of total power system mass at high power levels. Some scientists remain skeptical that radiative cooling can scale cost-effectively to match the thermal loads of modern AI clusters (Chaotropy, 2025).

Environmental sustainability is not guaranteed. Some modeling estimates that carbon emissions intensity of orbital data centers could exceed terrestrial facilities after accounting for rocket launch emissions and spacecraft reentry. The European Commission's ASCEND study estimated that for space data centers to effectively reduce carbon emissions would require launchers emitting 10x less carbon than current rockets.

Space debris and orbital congestion pose operational risks. With over 7,500 active satellites already in orbit and growing rapidly, collision risks increase. Large orbital structures require highly responsive spacecraft maneuverability and state-of-the-art tracking systems.

Startups Doing Big Things in Orbital Data Centers

The venture capital community has dramatically increased its exposure to space technologies over the past decade. From 2017 to 2025, venture capital invested in space technologies has nearly quadrupled (Pitchbook, 2025). This growth reflects a fundamental shift in how investors view the risk-return profile of space infrastructure—from speculative moonshots to critical enabling technology for the AI era.

Specialized space-focused funds have scaled significantly. Space Capital, led by Chad Anderson, has emerged as a category-defining investor with a thesis centered on space as critical infrastructure. Seraphim Space, the London-based fund, has deployed capital across the orbital value chain from launch to applications. Lux Capital has backed multiple space infrastructure companies as part of its broader deep tech mandate. These specialized funds bring domain expertise and networks that generalist investors often lack.

Generalist venture giants have followed. Andreessen Horowitz, Khosla Ventures, Founders Fund, and General Catalyst have all made significant space technology investments in recent years. Their participation signals that space infrastructure has crossed the threshold from niche to mainstream venture category. These firms bring larger check sizes, later-stage capital, and the ability to support companies through multiple growth phases.

Corporate strategic investors are increasingly active. Airbus Ventures invests across the aerospace value chain with strategic alignment to Airbus's commercial interests. NEOM Investment Fund, backed by Saudi Arabia's sovereign wealth, has made space technology a priority as part of its broader future cities and infrastructure mandate. The hyperscaler venture arms—Google Ventures, Amazon's Alexa Fund, Microsoft's M12—maintain watching briefs on orbital compute given the strategic implications for their core cloud businesses. Nvidia's direct investment in Starcloud reflects the chipmaker's interest in expanding its addressable market beyond Earth.

In 2025, private investment in the wider space sector grew 48% year-over-year to $12.4 billion (Reuters, 2026). The global space economy is expected to continue its growth trajectory, with estimates indicating it could reach $1.01 trillion by 2034, implying a 12% CAGR (SpaceNews, 2026). We expect technology that either advances or builds the tech stack and infrastructure for the orbital data center ecosystem to capture a significant and growing portion of future investment.

With strong market signaling, massive tailwinds from data center and energy demand growth, and IPOs and M&A activity expected in the coming years, the trend toward space infrastructure investment is poised to accelerate. The liquidity events that validated the commercial launch sector—SpaceX's continued private market appreciation, Rocket Lab's public listing, and multiple SPAC transactions—have created proof points for investors evaluating the broader space tech stack.

From 2017 to 2025, venture capital invested in space technologies has nearly quadrupled. With strong market signaling, massive tailwinds - particularly from data center and energy demand growth - IPOs and M&A deals expected in the coming year, this trend is expected to accelerate. A brief look at the past 8 years gives us some indication into what the future holds:

The data center sector stands at an inflection point. The AI revolution's appetite for compute has exposed fundamental constraints in how we build and operate digital infrastructure. For venture investors, this creates a multi-decade opportunity across the technology stack:

The boundaries between venture development sectors are blending. Cooling technology intersects with materials science. Energy infrastructure merges with aerospace. Data center operations converge with satellite communications. The investors who recognize these convergences early—who can underwrite both the near-term certainties of liquid cooling and the long-term optionality of orbital compute—will be positioned to capture the full breadth of this transformation.