American hyperscalers are borrowing heavily to fund artificial intelligence expansion, and rising yields signal investor caution about the debt binge underway. The U.S. government awarded GlobalFoundries $300 million one day ago to develop silicon photonics technology for more efficient AI data centers—a direct acknowledgment that the private sector's infrastructure race needs government backing to maintain America's technological edge.
The numbers reveal both opportunity and risk. Amazon, Google, Microsoft, Meta and Oracle are projected to spend about $750 billion this year on data centers, chips and other AI infrastructure, up from roughly $400 billion last year. By 2029, AI infrastructure investment is forecast to top $1 trillion globally, up from $318 billion last year. Yet Reuters reported that as hyperscalers borrow to fund this buildout, yields are rising as investor demand cools—a warning sign that markets are pricing in real concerns about returns on these massive capital commitments.
America's Computing Dominance—For Now
The United States controls about 80% of global computing power that drives AI and is home to about 5,500 data centers, about 10 times the next closest country. U.S. companies like Amazon, Google, Microsoft and Meta control about 80 percent of global computing power that drives AI, according to Epoch AI data cited by the New York Times. This advantage rests on physical infrastructure: there are about 20 million AI chips crammed into the data centers that underpin the technology's growing abilities and usage worldwide. That figure is expected to double roughly every nine months, putting the world on pace to have about 200 million of the chips by the end of 2028—10 times current levels.
But China is closing the gap faster than many realized. The world had about 2.4 million "H100 equivalent" AI chips in March 2024. Chinese companies had roughly 1.16 million H100-equivalent chips at the end of 2025, up from roughly 244,000 at the beginning of 2024, though those figures exclude smuggled chips and other offshore computing resources used by Chinese firms, Epoch AI estimated. This year Huawei, ByteDance and Alibaba are expected to spend $111 billion on data centers and other AI investments, according to Bernstein Research.
The Power Problem
The infrastructure race is colliding with a hard physical constraint: electricity. Last year data centers consumed 64 gigawatts of electricity globally, roughly as much electricity as Germany consumes. By the end of 2030, that is expected to quadruple, eclipsing the power used by all countries in South America and Africa combined. At the most advanced AI data centers, every gigawatt of power equates to roughly $40 billion to $60 billion in costs, including servers, land, connectivity and utility hookups, according to industry estimates.
China's National Energy Administration has estimated the country's electricity use for data centers will reach the equivalent of around 91 gigawatts by 2030, or about 6 percent of total use, up from 19 gigawatts last year. China's top tech companies are building new AI chips and data centers at scale, suggesting Beijing views this infrastructure as strategic national capacity rather than purely commercial investment.
The Global Gap Widens
Meanwhile, the rest of the world is falling further behind. France, Germany and other nations are trying to encourage data center construction across the European Union, which has 5 percent of global AI computing power, according to a report by AI developers and policy experts in the region. Europe has been hampered by electricity and land access, permitting and financing challenges that American and Chinese competitors don't face to the same degree.
The Persian Gulf has pledged billions to build data centers, but the war in Iran has affected plans. The U.S.-China race threatens to leave the rest of the world behind entirely.
Why This Matters:
The AI infrastructure buildout reveals fundamental questions about market efficiency and government's proper role. Private companies are making massive capital bets, yet rising yields suggest markets are skeptical about returns—a signal that shouldn't be ignored. The GlobalFoundries award shows government stepping in to fund specific technology development, raising questions about whether markets alone can sustain American leadership or whether strategic subsidies are necessary. More pressing: China's rapid chip accumulation and massive data-center spending suggest a competitor willing to absorb infrastructure costs as a national priority rather than a profit center. The power constraint isn't theoretical—it's becoming a binding limit on expansion. If American companies face permitting delays, electricity shortages, or capital constraints while Chinese competitors operate under state direction, the competitive advantage that's driven U.S. dominance could erode faster than current projections suggest.