Intel warns of supply constraints for artificial intelligence data center chips as turnaround meets new hurdles

Intel is struggling to meet demand for server processors used alongside artificial intelligence accelerators, issuing a cautious outlook that sent its shares sharply lower despite signs of strengthening product demand.

Intel said it struggled to satisfy demand for its server chips used in artificial intelligence data centers, and forecast quarterly revenue and profit below market estimates, sending shares down 13% in after-hours trading. The company, whose shares have risen 40% in the past month, recently launched a long-awaited laptop chip aimed at regaining ground in personal computers just as a memory chip crunch is expected to weigh on that market. Executives said Intel was caught off guard by surging orders for server central processors that accompany artificial intelligence chips, and that even with factories running at capacity it cannot keep pace, leaving profitable data center sales unrealized while the new PC chip pressures margins.

Chief executive officer Lip-Bu Tan told analysts that in the short term he is disappointed Intel is not able to fully meet market demand. The company forecast current-quarter revenue between $11.7 billion and $12.7 billion, compared with analysts’ average estimate of $12.51 billion, according to LSEG. It expects adjusted earnings per share to break even in the first quarter, compared with expectations of adjusted earnings of 5 cents per share. Finance chief David Zinsner said some cloud-computing giants were also surprised by artificial intelligence demand and had to scramble to upgrade aging chip fleets because of an “erosion in networking performance,” adding that they were all caught off guard. Zinsner told investors that despite owning its own factories, Intel faces a lag in changing product mix and had not been running plants with the assumption that data center demand would shift this way.

After years of missteps in the fast-growing artificial intelligence chip market, Tan is pushing a turnaround focused on cost cuts, slimming management and a new product roadmap. Intel has held off heavy spending on its next-generation 14A manufacturing process while it waits for a large customer, and Tan said two customers are now evaluating the technical details of 14A as a possible step toward test chips, with decisions expected by the second half of this year. Zinsner said capital expenditure could remain steady versus prior expectations for a decline. Investor confidence has been buoyed by a $5 billion investment from Nvidia, $2 billion from SoftBank and a U.S. government stake in Intel, and Michael Schulman of Running Point Capital said the turnaround remains supply constrained rather than demand constrained, delaying financial recovery. Tan has also scaled back the ambitious contract manufacturing strategy of predecessor Pat Gelsinger to shore up the balance sheet after capital-intensive expansions hurt margins.

After a more than 60% drop in its share price in 2024, Intel’s stock gained 84% in 2025, outpacing the benchmark semiconductor index’s 42% rise. The company has started shipping its new “Panther Lake” PC chips, the first product made using its critical 18A manufacturing technology, and analysts expected the production ramp-up to hurt margins. Reuters reported that only a small percentage of 18A chips have been good enough for customers, though Intel says yields are improving monthly, with Tan acknowledging that while 18A yields are in line with internal plans they are still below his target. A global shortage of memory chips has driven up prices and made personal computers more expensive, with Zinsner expecting available supply to be at its lowest levels in the first quarter and improve in the second quarter. At the same time, Intel has been consistently losing share in the PC market to rival AMD and chip designer Arm Holdings, underscoring the competitive and operational challenges surrounding its attempt to reassert itself in artificial intelligence and broader computing markets.

56

Impact Score

Fine-tuning embedding models with Unsloth

Unsloth introduces a FastSentenceTransformer based workflow that speeds up fine-tuning of embedding and related models while keeping them fully compatible with popular deployment tools and frameworks.

Impact and challenges of large language models in healthcare

Healthcare organizations are rapidly adopting large language models, but the real differentiator is how well these systems manage clinical context across fragmented data sources. This article outlines the main challenges, a practical implementation framework, and why context-aware Artificial Intelligence architecture is now table stakes for production use.

Contact Us

Got questions? Use the form to contact us.

Contact Form

Clicking next sends a verification code to your email. After verifying, you can enter your message.