Technology companies are clamoring for processors, especially hyperscalers looking to power their data centers.
The demand is so strong and supply is so tight that Intel redirected some of its production lines from consumer chips to Xeon server chips. On an earnings call Jan. 22, CFO David Zinsner said he expects wafer supplies to improve over the next six months, noting that supplies are most constrained in the first quarter.
“It’s just literally hand-to-mouth, what we can get out of the fab and what we can get to customers is how we’re managing it,” Zinsner said.
Intel, unlike fabless companies AMD and Nvidia, can leverage its own production facilities as demand outpaces supply. When asked about the imbalance, Zinsner saw it as “largely a win.”
Then again, there’s only so much available capacity. Entering 2026, he said, buffer inventory is depleted and the shift in wafers toward server chips won’t come out of the fab until later in the first quarter.
In short, Intel is missing out at a critical point in the market. On the call, CEO Lip-Bu Tan said his team is “working tirelessly” to drive efficiency and increase output, but he is also mindful of the challenges ahead.
During the fourth quarter, Intel’s data center segment grew 9% to $4.7 billion over last year. At the same time, consumer sales, which account for the majority of Intel’s revenue, declined 7% as the company looks to ramp up its foundry business and capture AI-driven trends.
“I’m disappointed that we are not able to fully meet the demand in our markets,” Tan said. He also noted that Intel is on a “multiyear journey” and “it will take time and resolve” to rebuild the company.
Investor sentiment wavered following Intel’s latest earnings call and results. Brian Colello, a senior equity analyst at Morningstar, said the thought on Wall Street was that Intel’s supply shortages would result in higher pricing and perhaps a stronger quarter. Many people were also expecting the company to announce a large foundry customer such as Apple or Nvidia, he added.
“I think Intel let them down on both fronts,” Colello said. “It’s always great when you can sell everything you can make, but it could have been better than probably the most optimistic of assumptions.”
The surge in data center demand is also affecting other U.S. chip companies. Nvidia’s CFO Colette Kress said on a recent earnings call that she believes inventory and supply commitments are in place to address growing demand, including shipments through next year. However, she said that supply tightness in server chips will likely persist.
Lisa Su, AMD’s chair, president and CEO, noted similar activity on a separate call, saying the company is working with supply chain partners to increase output. AMD, one of Intel’s biggest competitors, reported data center revenue of $5.4 billion in the quarter, up 39% from a year ago.
AMD and Nvidia both rely on Taiwan Semiconductor Manufacturing Co. or Samsung as manufacturing partners for their computer chips. Intel does as well for most of its advanced current and future products, the company disclosed in its latest annual report. The difference is that Intel is facing higher overhead costs as it works to further establish its foundry business and grow its customer base.
Bank of America analyst Vivek Arya wrote in a research brief that Intel’s foundry business could need another two to three years to reach its full potential. He noted that Intel’s current 18a node, or process for making semiconductors, delivers low yields. If the company is currently struggling with output from its current-generation node, “it may not be able to promise flawless execution to external customers at the more advanced 14a process.”
On the call, Tan said Intel recently started shipping out its products built on 18a, and yields are steadily improving.
In addition to explosive server chip demand, advancements in agentic AI are also driving a comeback for general-purpose processors because the technology does not require as much processing power, Manoj Sukumaran, a senior principal analyst at Omdia, said in an email.
“The market was not prepared for this kind of explosive [central processing unit] demand,” Sukumaran said. “So everyone is facing the heat now.”
Currently, all of TSMC’s leading edge nodes are booked by companies like Nvidia, Apple, AMD, Broadcom and more.
“Even if Intel would like to increase CPU shipments, they can’t do it because TSMC capacity is not available,” Sukumaran said. “The bottleneck for Intel is the compute chiplets manufactured by TSMC rather than their own manufacturing constraint.”
