Falcon Shores now an internal test chip; AI infrastructure system coming with Juaguar Shores
Intel has struggled to compete against AMD and NVIDIA in the AI infrastructure space. While the company could potentially see an uptick in business as emphasis on low-power edge AI inference picks up, right now Intel isn’t competitive in AI datacenters. Its Falcon Shores AI accelerator, meant to combine CPU and GPU functionalities in a unified architecture, was intended to come to market later this year to fill that gap. But on an earnings call yesterday, co-CEO Michelle Johnston Holthaus said Falcon Shores would now serve as an “internal test chip.”
Looking at Intel’s AI strategy, Holthaus said, “This is an attractive market for us over time, but I am not happy with where we are today. On the one hand, we have a leading position as the host CPU for AI servers, and we continue to see a significant opportunity for CPU-based inference on-prem and at the edge as AI-infused applications proliferate. On the other hand, we’re not yet participating in the cloud-based AI datacenter market in a meaningful way.”
Falcon Shores was intended to build on Intel’s Gaudi AI accelerator, an ASIC optimized for AI workloads that’s currently in its third generation. With Falcon Shores shelved, attention turns to Jaguar Shores.
“We plan to leverage Falcon Shores as an internal test chip only without bringing it to market,” Holthaus said. “This will support our efforts to develop a system-level solution at rack-scale with Jaguar Shores to address the AI datacenter. More roadly as I think about our AI opportunity, my focus is on the problems our customers are trying to solve, most notably the need to lower the cost and increase the efficiency of compute.”
She continued: “AI is not a market in the traditional sense. It’s an enabling application that needs to span across the compute continuum from the datacenter to the edge. As such, a one-size-fits-all approach will not work, and I can see clear opportunities to leverage our core assets in new ways to drive the most compelling total cost of ownership across the continuum.”
The point, I think, is that Intel’s AI infrastructure strategy continues to evolve. And that’s not unexpected given former CEO Pat Gelsinger’s abrupt retirement last year. With Jaguar Shores slated for release in 2026, and given the dizzying pace of AI infrastructure and application development, timing could be an issue. That said, in the grand sweep of AI, it’s still early days and a focus on execution against customer feedback could serve Intel well. And, as Holthaus pointed out, Inte’s AI story isn’t limited to enabling datacenters.
“I’m also trying to look at the roadmap to say there’s a lot of IP and assets that we have at Intel…that we can leverage to addres this market. We’ve got great CPUs, GPUs, ASICs, FPGAs, and we need to figure out how we harness those because if we’ve seen anything this week, when there are constraints put on customers, they figure out different ways to deploy technology. And so that’s also a great opportunity and something that I’m looking at and looking at if there’s ways that we can be disruptive there.”
Intel reported Q4 revenue of $14.3 billion, down 7% year-over-year, and full-year revenue of $53.1 billion, down 2% year-over-year. The company is guiding for Q1 2025 revenue of $11.7 billion to $12.7 billion. By business unit, the Client Computing Group (CCG) was down 9% to $8 billion compared to the year-ago quarter, and down 4% to $30.3 billion compared to full-year 2023. Datacetner and AI (DCAI) was down 3% to $3.4 billion compared to the year-ago quarter, and up 1% to $12.8 billion compared to full-year 2023. And the Network and Edge (NEX) business was up 10% to $1.6 billion compared to the year-ago quarter, and up 1% to $5.8 billion compared to full-year 2023.
Intel Foundry reported Q4 revenue of $4.5 billion, down 13% compared to Q4 2023, and full-year 2024 revenue of $17.5 billion, down 7% from full-year 2023.
