AI companies are boasting about power access to signal dominance

In short: Some AI companies are talking up how much computing power and electricity they can access to look ahead of competitors.

What's going on

Building and running modern AI systems often takes large data centers, which are warehouses full of computers. Those computers need two things in huge amounts, chips to do the work and electricity to keep them running.

As that demand grows, boasting about access to “compute” (the raw computer power used to run AI) is becoming part of corporate messaging. The idea is simple, if you can convince people you have the biggest engine, you look harder to catch.

A high-profile example showed up in the legal fight involving Elon Musk and OpenAI. A commentary describing the case quoted Microsoft CEO Satya Nadella saying that even if OpenAI “disappeared tomorrow,” Microsoft has “the people, we have the compute, we have the data, we have everything.” In plain terms, it is like saying, “We have the factory, the workers, and the fuel, so we can keep going without you.”

At the same time, there are signs that the “only giants can do this” story may be overstated. The New York Times points to DeepSeek, a Chinese startup, as an example of a smaller player training competitive AI with a fraction of the usual computing budget. That suggests big public claims about needing massive resources are not always a reliable guide to who will win.

What to watch

Expect more arguments about power and data centers, including scrutiny from regulators and local communities. Also watch whether smaller labs keep showing they can build strong models without the biggest budgets, which would weaken “we have the most power” as an intimidation tactic.

Source: NYTimes