AI in your toaster: Analyst predicts $1.5T global spend in 2025

And we're paying for it piecemeal through the software, services, and devices we buy

Tech analysts expect worldwide spending on AI to hit nearly $1.5 trillion in 2025, including $268 billion on optimized servers. These investments will also soon appear in even more consumer products.

Tech research company Gartner predicts that money spent on AI servers will more than double between 2024 and 2026, going from $140 billion to $330 billion, outstripping the market for conventional servers.

John-David Lovelock, Distinguished VP Analyst at Gartner, said the figure would increase again in 2027 to $380 billion.

"When we put it in perspective, think of everything currently running on servers in the world, the traditional ones: every bank, every transaction through Visa or ATMs, everything that the government does for you, every retail purchase, the entire coordination of the supply chains around the world, every piece of payroll, unified communication, social streaming services and gaming. They all run on servers. If we add up all the money we've ever spent on servers, from the year 2000 to now, we will have spent more on generative AI servers in five years than has ever been invested in traditional servers to do everything else."

By 2026, overall global AI spending is forecast to top $2 trillion, largely led by AI-integrated products such as smartphones and PCs, as well as infrastructure.

The money to pay for such mammoth spending will come from myriad GenAI services embedded in devices and software. "It's going to be in every TV, it's going to be in every phone. It's going to be in your car, in your toaster, and in every streaming service. Every piece of technology that you consume today will have GenAI in it, and that will be built and trained and run on these big servers that are out there," Lovelock said.

For enterprise software, this means higher prices with the promise that the embedded GenAI will make users more productive. "The software gets more expensive and it does more things," he said.

Lovelock compared the introduction of GenAI in business applications to the rollout of word processing software to the legal profession, allowing lawyers to write their own documents, cutting costs from the profession and allowing them to do more.

"As lawyers started to make more money, because of it, more lawyers came into the world," he said. "So in this case, that enabling technology increased the industry and access to the industry. The same kind of thing is likely to happen for IT services, consultants, and application implementation. They may charge a bit more on an hourly basis or task basis, but it'll take much less time to deliver, which means they can start going down-market. Rather than going to KPMG or McKinsey, for example, to get a full tax audit done, tax audit hooks are built into the software, and you simply offer it up to a service that will run it automated."

Lovelock has previously predicted an extinction among GenAI model providers, as the market will only sustain a few independent vendors with the capital to deliver the compute these models demand.

However, a clear-out of model providers does not mean a drop in investment overall, he argued.

It's more likely that the few model providers that are left after an "extinction event" would draw on so much computing power that the infrastructure spending would continue to increase, he said.

"Some people are writing about it as though only 10 percent of the [AI] startups will survive. While I agree with their number, I don't agree with their analysis. It doesn't mean 90 percent are failing. They are going to be involved in some form of merger or divestiture, morphing from what they were to what they will become. To be acquired is not a failure of the company. It's not the indication of a bubble." ®