The infrastructure buildout currently bypasses Europe and US allies
In 2027, big tech companies expect to spend nearly $1 trillion on AI data centers around the world. Where they choose to build those data centers could help determine the global balance of power.
To learn more, @alasdairpr, @TawilTeddy, and I created a detailed financial model showing what drives these $10+ billion decisions, and wrote about our findings in a new Carnegie report.
Users are excited to read reports on Big Tech planning to spend $1 trillion on AI data centers by 2027 because the scale and global power implications make the topic compelling.
@SamWinterLevy @alasdairpr @TawilTeddy Excited to read!
Finding data on datacenter buildouts is hard—it's still a surprisingly niche topic.
I was all the more happy by how precise the report is (and how many excellent charts it contains). Some highlights:
- The US remains the most attractive location for buildouts. - Delays are the largest driver of data center value erosion. - Time-to-power is the most relevant variable for buildouts. Compared to US/UAE, time-to-power in Germany is 2x longer. - The Iran War has lowered the UAE's competitiveness substantially.
In 2027, the world will spend $1 trillion on data centers. That's nearly 1% of world GDP.
The resulting infrastructure will largely bypass Europe.
@alasdairpr, @SamWinterLevy and @TawilTeddy have written an extremely detailed report on how to change this. Highly recommended.
New: America can’t build the world’s AI infrastructure alone. We need the scale only our allies can provide, but they are currently missing out on the biggest industrial mobilization since World War II.
@SamWinterLevy, @TawilTeddy, and I go deep into the economics of the AI infrastructure boom and propose a way forward for democracies to shape the trajectory of, and reap the benefits from, transformative AI.
6. The delays holding back most democracies are not laws of nature; they are the result of policy choices that can be revised. Democracies that rise to the challenge will help determine how AI reshapes the world. Those that do not will have little say in the matter.
Full paper here: https://carnegieendowment.org/research/2026/06/the-compute-coalition-how-to-build-the-future-of-ai-in-the-free-world
Finally, the US would benefit a lot from having more infrastructure built in Europe, especially for competition against China.
E.g., while power is scarce in some European countries, others have it in abundance. So scaling to Europe would increase total Western DC capacity!
1. Compute could determine the global balance of power
Compute is essential to transformative AI. The countries that host it can tax it, control who has access to it, and regulate what they do with it. If democracies lead the development of AI, they can shape it for the better. If they don’t, they’ll confirm the charge that democratic governance is too slow and fractious to compete.
4. The Gulf is not the answer either.
The US may be tempted to lean on the Gulf instead, which offers speed and capital. But heavy reliance on the Gulf was always a risky bet, and the war has done nothing to assuage concerns. Our back-of-the-envelope take: prolonged conflict (one-year delay, 20% higher construction costs, 1% annual war-risk insurance premiums) drops the UAE from first to sixth, ~8% behind the US.
5. America needs its allies.
The U.S. alone generates less electricity than China, and the gap is growing. But add in major US allies, and the picture changes. Nordic renewables and Australian solar can help fuel an AI supply chain spanning Canadian critical minerals, Dutch lithography machines, and automated German and Japanese manufacturing. The liberal world can collectively dominate the AI value chain, but for that to be anything more than a slogan, the world needs a democratic acceleration. The obvious lever to pull first is compute.
If you are interested in writing about how we can build more European AI infrastructure, please pitch me!
At Works in Progress, we are interested in much more writing on this topic.
3. America can’t build the AI future alone.
The U.S. hosts ~75% of the world’s advanced AI compute and U.S. projects move faster than those in most other countries. But that lead is fragile. Domestic constraints—grid capacity, permitting rules, political opposition—are tightening. And customers abroad will demand local compute for latency and data sovereignty reasons.
2. Europe and other democratic allies are at risk of falling behind.
Because of slow construction timelines and high power prices, it’s not as profitable to build in Europe. Compared to a similar 100 MW data center in the U.S., one built in the UK or Germany is worth >$1 billion less.
For democratic U.S. allies, this is a problem. The entire continent of Europe has a small fraction of America’s AI computing power, and it isn’t acting fast enough to change that.
For more on the economics of this, @TawilTeddy's thread is excellent:
@Simon__Grimm Datacenter buildouts: constrained by delays, geopolitics, time-to-power.
@Acurast: Compute on smartphones already in billions of pockets globally.
Not competing for the same buildout slots—solving the infrastructure problem differently.
227K+ compute units live.
