Intel Corp.’s disappointing forecast for the first quarter includes another anticipated double-digit decline in the company’s data-center business, as the chip maker has yet to capitalize on its AI opportunity.
Since its third-quarter earnings, Intel’s stock INTC, +0.94% has surged about 50%, in part due to the potential of providing chips for artificial intelligence. One analyst said in November that Intel represented a potential “under-the-radar AI opportunity.” But on Thursday, its shares tumbled about 11% after hours on its lower-than-expected forecast for the first quarter.
Also read: Intel’s earnings forecast comes up well short and the stock is tanking
The data center is where Nvidia Corp. NVDA, +0.42% A has been cleaning up with its graphics unit processors (GPUs) designed for AI training and inference applications. Nvidia’s data-center business grew nearly 280% in its fiscal third quarter, as it continues to see nonstop demand for its GPUs to accelerate AI applications.
Intel Chief Financial Officer Dave Zinsner told analysts Thursday, though, that the company has a $ 2 billion pipeline for its accelerator product line, code-named Gaudi, a chip that is expected to compete with Nvidia. The next iteration, Gaudi 3, is expected to launch sometime in 2024, but CEO Pat Gelsinger did not give any specifics on timing.
“While the data center has seen some wallet-share shift between CPU and accelerators over the last several quarters, we expect growth in CPU compute cores to return to more normal historical rates and our discrete accelerator portfolio, with well over $ 2 billion in pipeline, to gain traction as we move through 2024,” Zinsner said Thursday.
Marbel Lopez, principal analyst at Lopez Research, said she believes Intel’s AI story is not going to kick in until the third quarter. “Inferencing is Intel’s strength and where they can win,” she said. “It’s just a longer-term game.”
Some analysts are more skeptical. Gene Munster, a managing partner at Deep Water Asset Management, told CNBC after Intel’s call that he does not believe Intel is going to get the big lift from AI that it is hoping for. “There are better ways to play the AI silicon opportunity,” he said.
Lucas Keh, an analyst at Third Bridge, said in a note to clients that he believed Intel’s new products are coming out at a slower-than-expected pace, and that software is one of the bottlenecks. He said investors were disappointed with a slower-than-expected product ramp.
“It also brings the new question of where exactly cloud players’ appetite are for Nvidia alternatives outside of their own custom silicon developments,” Keh said.
In addition to the data center, Intel also has opportunity in the PC space, where it believes that new, AI-ready PCs with Intel chips that can compute AI queries on the device instead of the cloud, will begin to take off starting in the second quarter and running through the second half of 2024.
But the lesson here is that investors got a bit ahead of themselves on Intel’s AI story, and a reminder that there is still a lot of hope surrounding AI — but not every company has concrete numbers yet.