The AI boom has showered some of the nation’s most prominent companies in market value. OpenAI and Anthropic are now the two most valuable private companies in the world. Google, Microsoft, and Nvidia have become larger than ever. But among the biggest winners has been Caterpillar, a purveyor of yellow trucks and cranes. Caterpillar’s stock has more than doubled in value over the past year, making the company worth six times as much as Nike. Its ascent has little to do with the construction vehicles (both real and toy) that it is known for. Instead, Caterpillar’s success comes from its giant gas-powered engines that are helping power the nation’s data-center build-out.
For all the attention given to improving model capabilities, the most important inputs to AI are not bits but atoms and electrons. To train and run its models, the AI industry needs to build. Data centers are incredibly complex technological and industrial operations—they require tech companies to erect power plants, build wastewater facilities, and set up all manner of electrical equipment, power lines, and advanced-cooling equipment. That calls for concrete, steel, silicon, glass, copper, and liquefied natural gas.
This data-center construction has long been frenzied, but now it is fast approaching an inflection point that is both awesome and alarming. The major players—Amazon, Google, Microsoft, Meta, and Oracle—are on track to spend more on data centers by the end of the year than they bring in from their operations. That means tech giants whose non-chatbot businesses bring in tens of billions of dollars every year will likely need to take on debt to keep paying for their AI ambitions. From the launch of ChatGPT, in late 2022, through the end of last year, these firms’ capital expenditures—most of which go toward data centers—well exceeded half a trillion dollars. The companies intend to spend a similar amount in 2026 alone. And next year, their AI investments will likely exceed $1.1 trillion, according to a recent report from J.P. Morgan.
[Read: Inside the dirty, dystopian world of AI data centers]
Silicon Valley has long enjoyed a straightforward advantage: Its products and services cost relatively little to make but generate tremendous revenues. But operating a data center is in many ways more like running a steel mill than a smartphone app. In particular, data centers require stunning amounts of electricity, especially as AI models have gotten more powerful. Five years ago, a standard data center might require 10 or 50 megawatts of power, enough for tens of thousands of homes. A year ago, data centers demanding a gigawatt or two of power were considered gargantuan. This week, Meta announced that it is more than doubling the size of its flagship AI data center, which will increase its peak power needs to five gigawatts. A data center proposed in Utah, if completed, will demand nine gigawatts of power. That’s several large cities’ worth of energy pulsing through a few big warehouses, all dedicated to bots.
Electrical grids typically don’t have enough energy to supply these megaprojects, and so the fastest way to bring a data center online is to build your own power plant. That’s where Caterpillar comes in: The company has an enormous number of back orders for its power equipment, as does every major natural-gas-turbine manufacturer in the world. In May, Elon Musk reportedly spent at least $1 billion to buy an energy company with a fleet of combustion turbines, likely to power his AI model, Grok.
Power is only one part of the challenge. A data center requires wiring together dozens of cutting-edge AI chips onto a refrigerator-size rack, then linking thousands of those racks in a building. Those chips run hot, up to 200 degrees Fahrenheit, and are typically cooled by some combination of water and industrial fans; get too close to one of the racks, and you may break a sweat.
OpenAI’s self-proclaimed goal of making “intelligence cheaper, better, and more abundant over time” really amounts to powering and cooling data centers as efficiently as possible—in other words, getting really good at HVAC. And then there’s the complex global supply chain of mirrors and lasers and rare-earth minerals that go into the chips used to train and run AI models.
As the AI industry pours more money than ever before into data centers, executives have begun sounding like they are in the construction business. OpenAI CEO Sam Altman has said that the biggest constraint on his company is “electrons”—that is, power. Musk, Microsoft CEO Satya Nadella, former Google CEO Eric Schmidt, and others have said the same. Meanwhile, Nvidia CEO Jensen Huang recently said that a lack of skilled “plumbers and electricians” will be the AI boom’s “hardest” bottleneck to overcome.
Still, Silicon Valley seems to have only just begun reckoning with what it means for its primary business operations to be in the material realm. You cannot vibe-code a data center. In perhaps the clearest sign of the AI industry’s desperation to keep building, Anthropic, the most safety-conscious firm, is spending $1 billion a month to rent a data center from Musk, who just a few months ago deemed Anthropic “evil.” Anthropic CEO Dario Amodei, for his part, has accused Musk of “disturbing negligence.” Meanwhile, the tech industry is struggling to quell a mounting AI backlash, which has largely taken the form of opposition to data centers. A few days ago, New York became the first state to institute a moratorium on new “hyperscale” data-center construction.
[Read: The AI backlash could get very ugly]
If anything, AI is poised to become even more like drilling an oil well. All the copper, silicon, electricity, and labor that go into building and running a data center are mustered to process and spit out “tokens”—in essence, words—that are the basic unit of AI. The State Department recently announced an international pact to bolster the AI supply chain, called Pax Silica. “If the 20th century ran on oil and steel, the 21st century runs on compute and the minerals that feed it,” Jacob Helberg, the undersecretary for economic affairs, said in the announcement. Financial markets and investors are already beginning to treat tokens as the raw inputs to modern civilization: Various firms are launching such markets for compute, akin to how you can trade oil, steel, and corn on a stock exchange.
By dominating the digital world, Silicon Valley has already exerted indirect control over the physical one. Today’s data-center build-out is a whole new beast. From the most concrete aspects—cooling, power, land—to the most abstract financial instruments, AI has morphed into a heavy industry.