The AI Bubble Isn’t Popping. It’s Plugging In.

Markets took a tumble this week as AI-company Anthropic released new add-ons to Claude that can perform a range of functions typically filled by software providers. 

Shares of software-as-a-service companies like Adobe, Intuit, and Salesforce declined sharply on fears that AI tools might chip away at their business. (TIME co-chair and owner Marc Benioff is the CEO and founder of Salesforce). Legacy tech giants with large AI businesses like Microsoft, Amazon, and Google were also hit hard. A trillion dollars in market cap was wiped out in a week before regaining some ground on Friday. 

[time-brightcove not-tgx=”true”]

Analysts are still unpacking what it all means given the various factors at play. Anthropic pulling ahead in the race to build the best AI models may not be great news for legacy competitors building their own models, but it’s also not an existential threat. Anthropic partners with Amazon and Google to deliver the data center capacity necessary to bring its product to market. In any event, analysts are seeing the news as a clear indication that AI is here to stay. 

What does it all mean for energy and climate? Last week, I wrote about how the capital investment in AI has the power to unlock new energy technologies—and fundamentally reshape how we think about bringing clean energy to market. To my mind, this week’s news only underscores that dynamic. 

Most obviously, markets are signaling that faith in AI is more than hype and that the trillions in investment isn’t likely to be a bubble. By extension, that means that the market may be able to sustain the growing investment in electricity. Indeed, utility company stocks rose this week despite the broader market downturn.

It’s also an important moment for the hyperscalers—the large tech companies that provide cloud services—that have plowed billions into the future of AI. Much of the coverage of their AI efforts focuses on the push to build the best models. That’s important, of course, but this market reaction suggests a more complicated reality: in addition to racing to improve their models, the legacy tech companies have also become AI infrastructure companies—offering chips, land, and power to build data centers to serve not just their own needs but those of their customers. Anthropic is one such customer, and it seems safe to assume it will be a growing one.

As AI becomes more entrenched, the need for this infrastructure will only expand. To understand the dynamic, it’s helpful to pay attention to how hyperscalers are talking about power. On an earnings call last week, Microsoft CEO Satya Nadella referred to “tokens per watt per dollar” as a key performance metric. Including electricity alongside cost as a key performance metric is an indicator that the focus on power is here to stay.        

This has direct implications for decarbonization. In the short- to medium-term, it means more power usage, including fossil fuels. In the long-term, it means more investment in a slew of low-carbon energy technologies—from advanced nuclear to geothermal power. This week’s market disruption may have been about software, but it will have important implications for the future of physical infrastructure and power.

To get this story in your inbox, subscribe to the TIME CO2 Leadership Report newsletter here.

This story is supported by a partnership with Outrider Foundation and Journalism Funding Partners. TIME is solely responsible for the content.