Intel just delivered the most emphatic earnings beat of Lip-Bu Tan's tenure, and possibly the most significant quarter in the company's recent history.
Two days on from Thursday's Q1 numbers, the dust is settling on results that exceeded Wall Street expectations by a margin rarely seen in large-cap semiconductors.
The numbers that matter
Revenue came in at $13.6 billion, 7% higher than a year ago and roughly $1.2 billion above consensus.
Adjusted earnings per share landed at $0.29, against an analyst estimate of just one cent.
The data centre and AI division posted $5.1 billion in revenue, up 22% year on year and well clear of the $4.4 billion forecast.
Intel guided for second-quarter revenue of $13.8 billion to $14.8 billion, again comfortably above estimates.
The stock surged 24% on Friday, its best single-day gain since October 1987, pushing past the dot-com era peak set in 2000. Shares closed at $82.55.
The CPU finds a new role
The strategic story matters more than the beat itself.
Tan used the earnings call to frame Intel's CPUs as essential infrastructure for AI inference and the emerging category of agentic workloads, where autonomous software agents coordinate complex tasks.
CFO David Zinsner put a useful number on the shift. In AI training, data centres typically run eight GPUs for every one CPU. In inference, that ratio drops to three or four to one. In agentic workloads, he said, it could reach parity or flip in the CPU's favour.
Intel does not need to beat Nvidia at GPUs. It needs the industry's compute mix to broaden, and the early evidence suggests it is.
Google has committed to using multiple generations of Intel's Xeon processors for AI workloads. Xeon 6 was selected as the host CPU for Nvidia's DGX Rubin NVL8 systems. Demand is running ahead of supply.
Tan's $1 trillion bet
Tan told analysts that AI is pushing the total addressable chip market toward $1 trillion.
It is a bold claim from a company effectively left out of the first AI hardware cycle. But the thesis is specific: as AI moves from training toward real-world deployment, workloads diversify and hardware requirements broaden.
That plays to Intel's strengths in general-purpose compute, packaging and manufacturing. The Terafab partnership with Elon Musk, announced earlier this month, adds another dimension, with Intel contributing its 18A process technology to a facility producing chips for SpaceX, xAI and Tesla.
What could go wrong
The risks have not disappeared. Intel Foundry posted an operating loss of $2.4 billion in the quarter. External foundry revenue remains tiny at $174 million.
Yields on 18A wafers have improved but remain a work in progress. The next-generation 14A node, planned for 2028 or beyond, still needs to attract major external customers.
Bank of America raised its price target after the results but kept its sell rating, valuing the stock at $56, well below the current price.
At more than 120 times forward earnings, Intel is priced for a transformation that is underway but far from complete.
Related reading
- Google in talks with Marvell to develop two new AI chips as it builds multi-supplier TPU strategy
- Nvidia launches open-source quantum AI models as it deepens grip on chip design workflows
- Meta partners with Broadcom on MTIA AI chips
The takeaway
Tan has made Intel relevant to the AI conversation and delivered numbers that back the narrative. The CPU-to-GPU ratio shift is real, the customer wins are tangible and the manufacturing roadmap has commercial logic.
The recap
- Intel targets AI inference and agentic workloads to boost CPU demand
- Remarks were made on Intel's Q1 2026 earnings call
- Intel acknowledged ongoing chip manufacturing struggles during the call
