"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 Tan said.
Lip‑Bu Tan’s wager: inference, agents and “physical AI”
On Intel’s Q1 2026 earnings call, CEO Lip‑Bu Tan laid out a clear strategic pivot: the company is betting that AI inference, agentic workloads and what he called “physical AI” — agents, robots and edge devices running learning and inference — will reposition the CPU at the center of compute. He told analysts that AI is pushing the total addressable chip market toward $1 trillion and that Intel is well placed to capture share.
Tan argued the demand profile he hears from customers supports that view, calling inference “a much bigger market” and adding that the physical-AI opportunity is “another big market.” Those are the commercial premises behind Intel’s push to re‑emphasize CPUs after years when discussion focused largely on GPUs and other accelerators.
Product delivery is the bottleneck: Intel 14A and the foundry bet
Tan and CFO David Zinsner acknowledged the central constraint: Intel must still build the chips that will realize the strategy. The company has suffered delays to key chips in recent years and cancelled projects, including its “most recent effort to build a credible GPU to challenge AMD and Nvidia,” the call noted.
Intel is positioning its Intel 14A process node as the linchpin for turning Intel’s foundry business into a commercial success, with the company saying it “expects to see earlier design commitments emerge beginning in the second half of 2026 and expanding into the first half of 2027.” That schedule is the company’s public timetable for when outside customers and internal teams will begin committing designs to 14A.
Financials, momentum and market reaction
CFO David Zinsner reported Q1 revenue of $13.6 billion, above expectations, and said AI‑driven business lines accounted for 60 percent of that figure — a 40 percent year‑on‑year increase. Zinsner pointed to wins such as Xeon 6 being selected as the host CPU for Nvidia’s DGX Rubin NVL8 systems as evidence that Intel is resurgent in AI infrastructure.
The market reacted strongly: Intel’s share price rose by as much as 20 percent in after‑hours trading, reaching a five‑year‑plus high after the call. That bump reflects investor confidence in the narrative — but also places pressure on Intel to follow up with product deliveries and commercial traction consistent with the financial commentary.
Partnerships: Nvidia DGX Rubin, Google IPUs, and a cautious mention of Terafab
Executives named concrete partnerships and selections on the call. Zinsner highlighted Xeon 6’s selection as the host CPU for Nvidia’s DGX Rubin NVL8 systems. Tan also referenced a recent long‑term deal with Google for co‑development of infrastructure processing units (IPUs) to offload networking and other tasks, framing that arrangement as “a good example of how we win in AI infrastructure build‑out.”
The call touched, more briefly, on Elon Musk’s “Terafab” project. Tan said he and Musk “both share the vision” that the global supply chain is not keeping pace with demand and that they will “learn a lot together, exploring the innovative way in the process of the manufacturing,” adding “We’ll update you when can.” The Terafab project aims to produce very large volumes of AI chips — the source described it as “a terawatt’s worth of computing power each year.”
How technologists, procurement leaders, and end users will respond
- Technologists: Engineers and system architects will watch the Intel 14A timeline and vendor design commitments closely, since successful delivery is necessary for CPUs to reclaim a leading role in inference and agentic deployments.
- Procurement leaders and infrastructure teams: Organizations buying AI infrastructure will take note of concrete selections such as Xeon 6 in Nvidia’s DGX Rubin NVL8 and the Google IPU co‑development deal when planning next‑generation deployments and vendor relationships.
- End users and operators of edge or robotic systems: The company’s emphasis on “physical AI” signals that more inference capability aimed at agents and edge devices is anticipated; end users will look for chips and systems that match those requirements as Intel’s product roadmap materializes.
David Zinsner’s color on CPU:GPU ratios framed the opportunity: training solutions “are generally running at 8 GPUs to 1 CPU,” he said, while inference may be “3 or 4 to 1,” and agentic or multi‑agent workloads could “potentially even flip in the other direction a little bit.” Those ratios encapsulate why Intel is betting the farm on inference and agentic AI — provided it can deliver the silicon on the timetable it has set.
Intel’s argument is plain: the market is moving toward inference and physical AI, customers are signaling the need for CPUs in new roles, and partnerships are emerging. The proofpoint will be whether design commitments to Intel 14A arrive in the second half of 2026 and expand into the first half of 2027, and whether those commitments can be converted into timely, competitive product shipments. Until then, the rally in shares reflects belief; the next quarter will show whether Intel can turn belief into chips.
