Intel CEO Lip-Bu Tan used the company’s Q1 2026 earnings call to argue that CPUs will regain relevance as inference and agentic workloads move beyond the data center, according to The Register’s coverage of the call. The Register notes that Intel still needs to build the chips it is describing, and that its manufacturing track record over the past several years has been a constraint on forward-looking claims.
This article relies on a single Tier-3 source and has not been independently corroborated.
Q1 revenue came in at $13.6 billion, beating analyst expectations. Chief financial officer David Zinsner reported that AI-driven business lines accounted for 60 percent of that figure, up 40 percent year-on-year.
The GPU-to-CPU ratio argument
Tan offered a ratio to support his argument. Training solutions currently run at roughly 8 GPUs to 1 CPU. Inference workloads narrow that to 3 or 4 to 1. And in agentic and multi-agent architectures, according to Zinsner, the ratio may flip — CPUs potentially outnumbering GPUs in some configurations.
Tan argued that AI is moving out of the data center and into what he called “the physical world” — 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,” he said, adding that this reflected what customers were telling Intel and what demand profiles for its products showed. The Register reports Tan also declared that “the CPU is reinserting itself as the indispensable foundation” of AI.
One win cited on the call: Intel’s Xeon 6 was selected as the host CPU for Nvidia’s DGX Rubin NVL8 systems. Tan also referenced a recently announced long-term deal with Google for co-development of infrastructure processing units — silicon designed to offload networking and similar tasks from compute accelerators.
Manufacturing progress and the 14A node
Intel’s ability to deliver on any of these claims depends on its foundry operations. The company has faced chip manufacturing delays and cancellations in recent years. Tan said progress on Intel’s 14A process node is the mechanism that could turn Intel Foundry into a commercial business — producing chips for external customers as well as Intel’s own product lines.
“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 — a timeline The Register notes is consistent with comments Zinsner had made the previous month.
Tan was also asked about Elon Musk’s “Terafab” project, described during Tesla’s earnings call as an effort to produce large volumes of AI chips. Tan confirmed only that he and Musk share a view that the global supply chain is not keeping pace with AI demand, and that the two companies are “exploring the innovative way in the process of the manufacturing.” He declined to provide specifics, saying Intel would update investors when it could.
Market reaction
Intel’s share price rose by as much as 20 percent in after-hours trading following the earnings call, reaching what The Register describes as a five-year-plus high.