AI Data Centers List & Locations Database 2026 – 346 Facilities
Por PPGEO UEMA em 24 de dezembro de 2021
Content
Novel VPP agreements are emerging, such as the Google-Voltus deal that would aggregate up to 100 MW annually over three years in the https://dnray.com/cloud-hosting/best-cloud-hosting-options/ PJM grid. Google, for instance, is funding 1.6 GW of renewables (mostly wind) and 300 MW of storage alongside a new Minnesota data center. Tech companies are also turning to renewables, storage, and virtual power plants (VPPs) that use software to aggregate and redirect power in real time from distributed sources such as batteries and smart thermostats. Pairing it with carbon capture, utilization, and storage could make it cleaner, but that’s unproven at scale.
Microsoft committed $80 billion in FY2025 capital expenditure, largely for data centers. Traditional data centers cannot handle AI workloads without significant and expensive retrofitting. Electricity prices in Northern Virginia, the largest data center market in the world, have risen as utilities struggle to build generation fast enough. The US has the most, with 4,767 listed data centers (Data Center Map, August 2026), and Virginia and Texas lead the new-build pipeline. Global investment through 2030 is projected at $7 trillion, with $3 trillion in the US (IRMI, 2026).
The real estate industry, including asset managers, public companies and private investors, has also invested in AI development. Other AI data center projects include the Colossus supercomputer from xAI, a Louisiana-based project from Meta, Hyperion, expected to use 5 GW of power, and a second Ohio-based Meta project, Prometheus, with a capacity of 1 GW. By July 2026, local community resistance had blocked the construction of AI data centers worth some $130 billion. About 70% of global computer memory production in the 2026 fiscal year has been purchased for AI data centers. Their purchases of memory chips, particularly High Bandwidth Memory, led to a global memory supply shortage amid a broader competition for semiconductors, power, and infrastructure. This view shows the current indexed facility distribution in one linked projection.
The Energy Grid Challenge AI Data Centers Are Creating
AI data centers cluster in the United States, Western Europe, the Gulf states, and emerging Asian corridors. Third-quarter job postings fell year over year in seven of ten major markets, eight of ten industries, and nine of ten functions, while AI-related postings climbed. Hyperscalers and enterprises are shifting from raw expansion to disciplined, power-aware growth across global hubs.
Company
Staying at the leading edge grew so costly (tens of billions of dollars per generation) that missing one generation made catching up effectively impossible. A grand challenge on this scale demands grand solutions, and executives planning on business as usual will get blindsided. Constructing a 1 GW data center typically takes one to four years, even on sites with existing power (see Figure 3), and it can take longer when new generation and transmission infrastructure is needed. Most of the growth will come from the US, but compute capacity will grow by double digits across regions. To overcome it, the tech industry is planning investments at unprecedented scale and speed. On the plus side, they bring construction and operations jobs, local tax revenue, and the computing capacity behind AI tools.
What It Costs to Build an AI Data Center in 2025
- The volume of construction planned through 2030 reflects a bet by the technology industry that AI demand will continue growing at rates that justify the capital commitment.
- China lists only 376 but runs nearly 11 million racks, a gap we explain in our guide to how many data centers China has.
- A single large-scale LLM training run can consume 50 million kilowatt-hours, roughly equivalent to the annual electricity use of 5,000 US homes.
- AI facilities are a growing slice of that total, with nearly 3,000 data centers under construction or planned worldwide by 2030 (IRMI, 2026).
- First, the availability and price of GPU compute, which determines what AI products exist and what they cost, is a direct function of how much AI data center capacity gets built and where.
- Whether that bet is correct will shape the technology landscape for a decade.
Data Center Map lists 4,767 data centers of all types in the US (August 2026), and a separate Cloudscene dataset counts 5,427. Outside the US, China is the biggest other market, with 376 listed data centers (Data Center Map, August 2026) and nearly 11 million racks (Oxford Institute for Energy Studies, 2026). In the US that means Northern Virginia and Texas, plus newer hubs such as Louisiana (Meta's Hyperion campus) and Memphis, Tennessee (xAI). Beyond hyperscalers, xAI built a 100,000-GPU H100 cluster in Memphis, Tennessee, in 2024. Oracle has committed to expanding to 77 cloud regions globally.
- Individual projects are finding workarounds, but they can’t be replicated at scale, and some are controversial.
- The United States leads by both count and aggregate megawatt capacity, followed by other major hyperscale and sovereign buildout markets.
- The Lawrence Berkeley National Laboratory projects US data centers could use 6.7% to 12% of US electricity by 2028, up from 4.4% in 2023.
- Power, hardware, and skilled labor are simultaneously constrained (see Figure 2), and public approvals are increasingly difficult to secure.
- Data centers powering AI have rapidly become a new layer of critical infrastructure worldwide that are central to technology innovation, economic growth, and national sovereignty.
- Colossus (from xAI, now part of SpaceXAI) was built in under a year by repurposing a factory and leasing dozens of gas turbines without federal clean air permits, drawing a lawsuit.
AI data centers, defined
To justify the investment, AI must do more than boost productivity; it will need to unlock new sources of growth and value. Technological innovation, new revenue, and public support may be needed to fund and supply enough electricity. But the same interventions that can unlock power and land can introduce control, oversight, and delays, and companies and investors https://cognixpulse.com/articles/ecm-reviews-tools-insights-trends/ will need to plan around that risk. For example, China’s government is investing $295 billion in AI data center construction and $740 billion in grid upgrades to pipe renewable power from its resource-rich west to data centers. SpaceXAI’s post-IPO valuation reflects investor conviction that space-based compute will eventually matter, but timelines and economics remain unproven. Orbital data centers could theoretically address land, cooling, and power constraints.
