good news and bad news

Intel is allocating more of its own production to its money-making server chips.

Credit: Intel

Credit: Intel

Intel reported its earnings for the fourth quarter of 2025 yesterday, and the news both for the quarter and for the year was mixed: year-over-year revenue was down nearly imperceptibly, from $53.1 billion to $52.9 billion, while revenue for the quarter was down about four percent, from $14.3 billion last year to $13.7 billion this year. (That number was, nevertheless, on the high end of Intel’s guidance for the quarter, which ranged from $12.8 to $13.8 billion.)

Diving deeper into the numbers makes it clear exactly where money is being made and lost: Intel’s data center and AI products were up 9 percent for the quarter and 5 percent for the year, while its client computing group (which sells Core processors, Arc GPUs, and other consumer products) was down 7 percent for the quarter and 3 percent for the year.

That knowledge makes it slightly easier to understand the bind that company executives talked about on Intel’s earnings call (as transcribed by Investing.com). In short, Intel is having trouble making (and buying) enough chips to meet demand, and it makes more sense to allocate the chips it can make to the divisions that are actually making money—which means that we could see shortages of or higher prices for consumer processors, just as Intel is gearing up to launch the promising Core Ultra Series 3 processors (codenamed Panther Lake).

Intel CFO David Zinsner indicated that Intel is “prioritizing [its] internal wafer supply to data center” and having more of its consumer chips made externally. Substantial portions of the Core Ultra Series 3 processors are manufactured at Intel, after the company primarily used TSMC’s chip factories for the Core Ultra Series 2 designs.

“We can’t completely vacate the client market,” said Zinsner, but Intel is “shifting as much as we can over to data center to meet the high demand.”

Compounding this is that yields for the new 18A process—that is, the number of chips in a wafer with no defects that can work as designed—are still improving. Intel CEO Lip-Bu Tan said that “while yields are in line with our internal plans, they are still below what I want them to be.”

Yields are currently improving by 7 or 8 percent every month, according to Intel. But that could be building on pretty low initial yields—reporting from last summer suggested that just 10 percent of the chips coming off of the 18A production lines were meeting Intel’s requirements at the time. Intel predicts that its supply will have ramped up enough within the next few months to help alleviate shortages.

“I do believe that the first quarter is the trough,” said Zinsner. “We will improve supply in the second quarter.”

Intel is selling everything it can make

When Intel can start making enough chips to meet its demand, it ought to help brighten the company’s earnings reports.

“We delivered [our Q4 2025] results despite supply constraints, which meaningfully limited our ability to capture all of the strengths in our underwriting markets,” said Tan. “We are working aggressively to address this and better support our customers’ needs going forward.”

Intel has been signaling for a while now that it was selling essentially all of the chips it could get its hands on. Intel investor relations VP John Pitzer said last month that Intel would be selling more of both its Lunar Lake and Arrow Lake Core Ultra Series 2 chips for consumers, as well as its Granite Rapids chips for data centers, if it could get more of them.

As Intel seeks to improve its position in the short term, the company also says that it’s still making progress on its future manufacturing nodes, including different versions of the 18A process and the upcoming 14A process. Intel is working to engage “potential external customers” who would use the 14A process to make their own chips. If these third parties decide to use Intel’s manufacturing facilities, Intel expects to know about it “starting in the second half of this year and extending into the first half of 2027,” and then it expects to build out manufacturing capacity based on the number of external customers it finds.

On the chip design side, Intel also expects to have its first next-generation Nova Lake chips ready “at the end of 2026.” We don’t know much about Nova Lake yet, but it should be Intel’s next architecture to cover both desktop and laptop processors, while Panther Lake chips are intended mainly for laptops. At least part of the chip will also be manufactured using the 18A process.

Andrew is a Senior Technology Reporter at Ars Technica, with a focus on consumer tech including computer hardware and in-depth reviews of operating systems like Windows and macOS. Andrew lives in Philadelphia and co-hosts a weekly book podcast called Overdue.