Intel Cancels Falcon Shores AI Chip, Shifts Focus to Jaguar Shores for AI Data Centers

Intel is pulling the plug on Falcon Shores, its next-gen GPU designed for high-performance computing and AI workloads, marking a significant shift in strategy for the company. This decision comes as Intel aims to recalibrate after a series of disappointing product releases and major financial setbacks, all while competitors like AMD and Nvidia continue to make significant headway.

The Change in Focus

During Intel's fourth-quarter earnings call, co-CEO Michelle Johnston Holthaus announced that the company would be using Falcon Shores as an internal test chip, effectively scrapping any plans to release it commercially.

“AI data centers are an attractive market for us,” Holthaus explained, “but we’re not happy with where we are today. We’re not yet making a meaningful impact in the cloud-based AI data center market.”

This is why Intel looks to simplify its roadmap because it sees Jaguaring Shores a more strategic move. “Our shot here is to build at system level and at rack scale,” she said, as an attempt to reach out to the bigger market for AI data centers.

A Shift in Approach

Last month, Holthaus had already tempered expectations for Falcon Shores as it’s a step up from Intel’s last attempt at an AI chip, Gaudi 3. "We learned from Gaudi one of the things we deliver Silicon isn’t enough," she says. “Falcon Shores will help with the system, the networking, and the memory, but what customers really want is a full scale rack.” Jaguar Shores is there.”

The new Intel AI chip, Gaudi 3, has been tarnished as it has fallen short of the company’s hopes that the chip would sell $500m worth, and that is due to software issues. In fact, IBM and some other service providers are the only ones using the chip as of now.

An Uphill Battle

Intel is facing an uphill battle in the AI chip space. AMD is projected to generate around $7 billion in AI chip revenue by 2025, while Nvidia, the dominant player in the market, is expected to hit a staggering $195 billion in revenue by fiscal 2026, according to analysts.

Holthaus remains focused on solving AI challenges, particularly around cost reduction and compute efficiency. She noted that a one-size-fits-all approach wouldn't be effective, and emphasized Intel’s ability to leverage its core assets to deliver compelling total cost of ownership solutions across the AI continuum.