Intel bets the farm on AI inference to drag CPU back to the top table

Chipzilla hopes agents, robots, and edge devices make CPUs cool again...

now it has to build the chips
Intel is betting on AI to reverse its fortunes, wagering that inference and agentic workloads will restore the CPU to the center of compute - even as its chip manufacturing struggles persist.

"For the last few years, the story around high-performance computing was almost exclusively about GPU and other accelerators.

In recent months, we have seen clear signs that the CPU is reinserting itself as the indispensable foundation of the AI era," Lip-Bu said.

AI is moving out of the data center and into the physical world, he added, with inference and learning workloads increasingly running on agents, robots, and edge devices.

"I think the inference is going to be a much bigger market and the physical AI is another big market.

So I think that's an opportunity for us...

This is not just our wishful thinking, it is what we hear from our customers, and it is evident in the demand profile for our products."
However, Intel needs to build the products in order to deliver on the promises, and the past several years have seen the chipmaker suffer delays to key chips and the cancellation of others , notably its most recent effort to build a credible GPU to challenge AMD and Nvidia in the AI training stakes.

Lip-Bu says Chipzilla is making progress with its Intel 14A process node, one that it hopes will turn Intel's Foundry biz into a commercial success by producing chips for other companies as well as its own products.

"We expect to see earlier design commitments emerge beginning in the second half of 2026 and expanding into the first half of 2027," he said, echoing comments by chief financial officer David Zinsner last month.

Zinsner reported Q1 revenue of $13.6 billion, beating expectations, with AI-driven business lines accounting for 60 percent of that figure, up 40 percent year-on-year.

He pointed to recent wins including Xeon 6 being selected as the host CPU for NVidia's DGX Rubin NVL8 systems as evidence that Intel is resurgent in the AI arena.

Lip-Bu also referenced a recent long-term deal with Google for co-development of infrastructure processing units (IPUs) to offload networking and other tasks, saying: "This is a good example of how we win in AI infrastructure build-out.

And then stay tuned - at the right time, we will announce other contracts."
Zinsner added: "One statistic we look at is the ratio of CPUs to GPUs.

And if you look at training solutions, they're generally running at 8 GPUs to 1 CPU.

As we look into inference, it's probably getting into the 3 or 4 to 1 kind of level.

And as you get into agentic and multi-agent, it's one potentially even flip in the other direction a little bit."
Another potential AI win is with Elon Musk and his "Terafab" project, which aims to produce large volumes of AI chips - a terawatt's worth of computing power each year , in fact.

Although Musk himself talked about this during Tesla's own earnings call this week, Lip-Bu was more tight-lipped when asked about it by an analyst.