Of the 96 AI data centers in Epoch AI’s database, 14 have more than 250 MW of IT power today, by our count of its data file updated on 11 November. The largest is Anthropic-Amazon New Carlisle, which Epoch says Amazon owns in Indiana, at 1,037 MW, the only site above 1,000 MW.
The count matches a post on X by @object_zero_, seen about 29,000 times, which said “Today, there are only 14 data centers on Earth that are bigger than 250MW”. By our arithmetic, those 14 sites hold 7,068 MW, about 47% of the 15,122 MW across all 96.
Business Pill · Concentration Explained
A short explainer on concentration, when a few big pieces make most of the total. It teaches the general idea only and does not discuss Epoch’s data.
The key insight: Epoch AI’s database lists 14 AI data centers with more than 250 MW of IT power, by our count of its data file, led by Anthropic-Amazon New Carlisle at 1,037 MW, the only site above 1,000 MW. By our arithmetic, the 14 hold about 47% of the 15,122 MW across the 96 sites Epoch tracks, and Epoch estimates its coverage at 45% of global deployed AI compute.
Epoch says “The 96 AI data centers we cover have a combined 15.1 GW of IT power capacity”. It names the five largest by IT power as New Carlisle at 1,037 MW, Colossus 2 at 946 MW, OpenAI Stargate Abilene at 824 MW, Microsoft Fairwater Atlanta at 636 MW and DayOne Nusajaya at 473 MW.
The 14th site above the line is CoreWeave Denton TX at 262 MW; the 15th, Bridge Data Centres MY06 in Malaysia, sits just under it at 243 MW. By our count, 12 of the 14 are in the United States and two, DayOne Nusajaya and DayOne Kempas, are in Malaysia. Four sites are above 500 MW.
By our arithmetic, the other 82 sites hold 8,054 MW, about 53% of the total, and 20 of them show 0 MW of current IT power in the file. The chart attached to the X post, titled “AI data centers ranked by IT power as of Nov 11, 2026”, opens with the same 15 sites in the same order as Epoch’s data.
Current IT power of each of the 96 AI data centers in Epoch AI’s database (data updated 11 November 2026). The count of 14 above 250 MW and the 47% share are our arithmetic; the 250 MW line is the top of the SST-800 range on Siemens Energy’s product page.
What Was Posted on X
How big are the data centers today?
Today, there are only 14 data centers on Earth that are bigger than 250MW. For context, 250MW is a single Siemens SST-800 steam turbine.
So whilst the press talk this capex cycle up a lot, so far this is not much energy that is flowing to compute. Not yet.
There are only a handful of big data centers operating today, but there are a lot more big data centers being built right now.
The wave of big new data centers coming online still hasnโt hit. Frontier models are still training on sub-GW machines.
If you build cooling towers, itโs quite possible to build a single terrestrial facility that can emit 5GW of heat. Nobody has done that yet, but itโs theoretically possible (although maybe not politically).
The largest data center is more valuable than 2x data centers that are half its size, because a larger machine can more easily train and output the frontier.
Whoever can build the largest coherent synchronous machine, should be able to train the most powerful model.
Building such a facility, now the supply chain is tight, takes 5-7 years. There are a limited number of sites capable of hosting a GW scale data centerโฆ all of those sites are now bought up.
Many countries do not have a single site capable of hosting a GW scale machine.
Now the shape of AI can vary wildly, you can have many small models, a sort of swarm of AI, all running on lots of little compute machines. Or you could have one giant coherent machine that hosts a single giant super intelligence.
Inference can be a distributed swarm or a consolidated deity, or something in between.
But these things are not the same, they have strengths and weaknesses, they serve different applications better or worse.
Humanity is going to build it all.
But different organisations already have distinct trajectories and strategies for what they want to do, there is much more diversity in the market and the economy than people think. There are also myriad unserved gaps.
@object_zero_ posted on 11 November at 00:37 UTC that only 14 data centers on Earth are bigger than 250MW, comparing 250MW to a single Siemens SST-800 steam turbine, and attached a ranked chart of AI data centers by IT power plus two Siemens brochure pages. The piece checks the count against Epoch AI’s data file, where 14 of the 96 sites it tracks are above 250 MW, and the turbine against Siemens Energy’s own page, which lists the SST-800 at up to 250 MW.
Checking the Turbine Comparison
The post adds: “For context, 250MW is a single Siemens SST-800 steam turbine”. Siemens Energy’s product page lists the SST-800’s power output as “โค 250 MW”, so 250 MW is the top of its range, not a fixed size. The brochure page attached to the post describes two SST-800 turbines at a Klabin pulp mill in Brazil with 190 MW.
Epoch’s figures are IT power, the load of the computing hardware. It says “Total facility power is 20-50% higher due to cooling and other infrastructure overhead”. By our arithmetic, a site with 250 MW of IT power would need 300 to 375 MW at the facility level, more than one SST-800 at its maximum.
Why the Count Is a Floor
Epoch does not claim to see every site. The FAQ on its hub page says “we estimate this databaseโs coverage of global deployed AI computing capacity to be 45% as of November 11, 2026 (90% CI: 30% to 78%)”. It also says “We continue to expand coverage globally, including Asia, Europe and the Middle East”.
Epoch says it expects that “80% of the time, a given IT power capacity estimate is accurate within a factor of 1.4x”. The 14 is therefore a count of Epoch’s estimates, not a census of every data center on Earth, which is how the X post phrased it.
Where China’s Galaxy Campus Sits
The same file lists Envision Galaxy Ulanqab in Inner Mongolia at 118 MW of current IT power. SemiAnalysis reported on 7 November that Galaxy is “a campus in Ulanqab planned for up to 800MW of IT capacity” and that “Its first 120MW phase was delivered within a year at a fifth of US costs”, a report we covered in China AI Campus Built 120 MW in a Year.
By our arithmetic, Epoch’s 118 MW is about 15% of Galaxy’s planned 800 MW, which keeps it below the 250 MW line for now. We used Epoch’s database earlier for Huawei Horinger’s compute.
The Structural Read
As we read it, the AI build-out is top-heavy: fewer than one in six of the sites Epoch tracks hold almost half of their combined power, and 12 of those 14 are in the United States.
As we read the turbine comparison, 250 MW is the ceiling of the SST-800 range on Siemens Energy’s page, and Epoch counts IT power, so a site at the line draws more at the facility level than one such turbine supplies.
Epoch AI, AI data centers hub page
“The 96 AI data centers we cover have a combined 15.1 GW of IT power capacity”
Three Implications
FOR POWER BUYERS By our arithmetic, 250 MW of IT power means 300 to 375 MW at the facility level, using the 20-50% overhead Epoch gives.
FOR THE COUNT Epoch estimates its database covers 45% of global deployed AI compute, so 14 is a count of the sites it tracks.
FOR CHINA Epoch lists Envision Galaxy Ulanqab at 118 MW; SemiAnalysis says the campus is planned for up to 800 MW.
The framework argues: “The structural bottleneck isn’t computing, it’s electrons.”
As we read it, Epoch’s file is a map of where the electrons already flow: 14 sites above 250 MW hold about 47% of the IT power it tracks, by our arithmetic. The framework predates this story and does not discuss Epoch’s count.
What Is Not Established
Epoch’s figures are estimates from satellite imagery, permits and company statements, and its database covers an estimated 45% of global deployed AI compute, so sites above 250 MW may exist outside it. The documents we read do not say how @object_zero_ arrived at 14. We did not contact Epoch AI, Siemens Energy or the poster for this piece.
This Week In The AI Supercycle
The Great Convergence of Capital, Power, and Geopolitics
This week’s issue: for two years the AI build-out was priced as a supply story, and last week as a credit story. This week it was priced as a permission story, and the answer arrived as a cost, not a ban.
Epoch AI’s database lists 14 AI data centers above 250 MW of IT power, led by Anthropic-Amazon New Carlisle at 1,037 MW, the only one above 1,000 MW. By our arithmetic, those 14 hold about 47% of the 15,122 MW across the 96 sites Epoch tracks, and Siemens Energy rates the SST-800 at up to 250 MW.
A note on sourcing. We read Epoch AI’s AI data centers hub page, including its FAQ, and counted the sites in its data file (ZIP updated 11 November 2026); every site figure is Epoch’s estimate. We read the SST-800 technical data on Siemens Energy’s own product page. The post by @object_zero_ prompted the piece and is quoted as posted; the Galaxy figures are from SemiAnalysis’s 7 November report. Counts, shares and the facility-power range are our arithmetic.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
