Intel was losing the AI race and is somehow making money from it anyway
Intel beat expectations, raised its outlook and lifted planned 2026 capital spending from $18bn to $20bn. It did not sell a single GPU to get there.
For two years the story told about Intel never changed: the company that showed up late to artificial intelligence and watched Nvidia take the lot. Its latest results force a correction to at least half that sentence.
What did Intel report?
Quarterly results ahead of expectations, its fastest revenue growth in years, and third-quarter guidance above Wall Street estimates. The company also raised planned 2026 capital spending from $18 billion to $20 billion to meet demand for data centre processors and expand manufacturing capacity.
How does Intel win from AI without selling the GPUs?
Because an AI data centre is not made of accelerators alone. Large server clusters need general-purpose CPUs to coordinate workloads, manage data and run the supporting applications that keep the whole thing upright. Demand for those processors reportedly outstripped Intel’s production capacity at points this year.
That is precisely the argument chief executive Lip-Bu Tan has been making: Intel can benefit from the AI buildout without having to dethrone Nvidia first. Selling shovels instead of gold remains a perfectly good business.
Are the old problems solved?
They are not. The hard question is still whether Intel can attract enough external customers to its foundry business and hold its manufacturing roadmap to schedule. Neither gets fixed by one good quarter.
What the numbers do show is that AI money is spreading to a wider set of suppliers than investors expected — even as it pushes Alphabet into negative free cash flow and drives AMD toward 256-core server chips. The filings sit in Intel’s investor relations pages.
By Beatriz Mota
Image: Coolcaesar / Wikimedia Commons (CC BY-SA 4.0)
