Jim Keller on Intel’s Foundry Future: Promising 18A Chips, But Work Remains

Jim Keller on Intel’s Foundry Future: Promising 18A Chips, But Work Remains

https://i.ytimg.com/vi/_MrGNlXRi9M/hq720.jpg?sqp=-oaymwE7CK4FEIIDSFryq4qpAy0IARUAAAAAGAElAADIQj0AgKJD8AEB-AH-CYAC0AWKAgwIABABGF4gXiheMA8=&rs=AOn4CLBvVjDOGaBNpU4X3jNVdRHeTeyMVQ

Intel’s push to become a leading foundry player just got a notable nod from one of the industry’s most respected figures: chip architect Jim Keller. In a recent interview, the legendary designer behind some of AMD, Apple, Intel, and Tesla’s biggest silicon successes shared insights on Intel’s cutting-edge 18A process. While optimistic about its potential, Keller emphasized that Intel has significant hurdles to clear to deliver a competitive product for external clients like his AI startup, Tenstorrent. Let’s dive into what this means for Intel’s foundry ambitions and the broader chip landscape.

Keller’s Take: Intel’s 18A in the Mix, But Not There Yet

Jim Keller, now leading Tenstorrent’s charge into AI chips built on RISC-V architecture, revealed that his company is exploring Intel’s 18A node alongside other heavyweights like TSMC, Samsung, and Rapidus for 2nm technologies. In an interview with Nikkei Asia, Keller expressed cautious optimism about Intel Foundry, stating, “We would consider Intel for chip production, though they still have a lot of work to do to deliver a really solid technology roadmap.” This is a big deal—Keller’s endorsement, even with caveats, signals that Intel’s 18A process is on the radar of cutting-edge AI firms. Curious about Intel’s foundry pivot?

Why Intel’s Foundry Push Matters

Intel’s 18A node, set to power its Panther Lake Core Ultra 300 CPUs in late 2025, is a cornerstone of its plan to rival TSMC and Samsung as a global foundry leader. Unlike its historical focus on internal chip production, Intel is now courting external partners to adopt its advanced nodes, including 18A and the upcoming 14A. Keller’s comments suggest that Intel is making strides, but reliability and scalability remain unproven. For a company like Tenstorrent, which is designing AI chips to compete in markets underserved by NVIDIA, a dependable foundry partner is critical.

FoundryProcess NodeKey Clients (Known/Prospeculative)
Intel18A (1.8nm)Internal (Panther Lake), Tenstorrent (potential)
TSMC2nmApple, NVIDIA, AMD
Samsung2nmQualcomm, Samsung Exynos
Rapidus2nmTenstorrent (confirmed interest)

Jim Keller’s Influence and Tenstorrent’s AI Ambitions

For those unfamiliar, Jim Keller is a titan in chip design. His resume includes architecting AMD’s Zen CPUs, Apple’s A-series chips, Intel’s Skylake processors, and Tesla’s self-driving silicon. Now at Tenstorrent, he’s building AI chips using the open-source RISC-V architecture, targeting niches where NVIDIA’s dominance is less entrenched, such as edge AI and specialized workloads. Tenstorrent’s interest in Intel’s 18A, alongside TSMC, Samsung, and Rapidus’ 2nm nodes, underscores Keller’s strategy to leverage the best manufacturing tech available. His comments carry weight, making Intel’s inclusion in his shortlist a vote of confidence, albeit with a call for refinement.

Intel vs. the Foundry Giants: The Competitive Landscape

Intel’s foundry journey is a high-stakes pivot. TSMC dominates with clients like Apple and NVIDIA, while Samsung powers Qualcomm and its own Exynos chips. Rapidus, a Japanese startup, is gaining traction with 2nm plans backed by Tenstorrent’s interest. Intel’s 18A, equivalent to a 1.8nm process, promises high transistor density and power efficiency, ideal for AI and mobile chips like Panther Lake. However, Keller’s caveat about Intel’s “work to do” points to challenges in yield rates, cost competitiveness, and a robust roadmap to rival TSMC’s proven track record. Intel’s success hinges on delivering reliable 18A chips to external clients by 2026, when Panther Lake-powered devices hit shelves.

What’s Next for Intel and Tenstorrent?

Intel’s 18A process is already slated for internal use in Panther Lake and Clearwater Forest CPUs, with production ramping up in 2025. Securing external partners like Tenstorrent could validate Intel’s foundry credentials and boost its U.S.-based manufacturing push, especially amid geopolitical shifts favoring domestic production. For Tenstorrent, tapping Intel’s 18A could diversify its supply chain, reducing reliance on TSMC or Samsung. Keller’s multi-foundry approach—engaging TSMC, Samsung, Rapidus, and Intel—shows a pragmatic strategy to ensure Tenstorrent’s AI chips hit the market with cutting-edge performance. Want to know more about AI chip trends?

Why This Matters for PC Enthusiasts

For PC enthusiasts, Keller’s comments spotlight Intel’s potential to disrupt the foundry space, which could lead to more competitive chips for laptops and desktops. If Intel nails 18A, we could see third-party AI chips from firms like Tenstorrent powering next-gen PCs, alongside Intel’s own Core Ultra 300 series. However, Keller’s caution suggests Intel must prove its foundry can match TSMC’s reliability before it wins major clients. This is a space to watch as Intel’s 2025-2026 rollout approaches, with implications for pricing, performance, and innovation in the PC market.

Can Intel’s 18A compete with TSMC and Samsung? Will Tenstorrent’s AI chips shake up the market? Share your thoughts in the comments!

Share