Intel and Applied Materials have signed a multi-site, multi-technology process collaboration spanning front-end transistor development through back-end packaging, interconnect and power delivery — with a stated focus on Foveros-based 3D stacking for agentic AI workloads.

The engineering work will run at two locations: Intel's R&D campus in Hillsboro, Oregon, and Applied's EPIC Centre — the Equipment and Process Innovation and Commercialization facility in Silicon Valley, which is slated to open later this year. Intel joins as a founding partner of that facility.

The packaging thrust targets higher interconnect density, improved power delivery and better thermal performance for high-performance compute platforms — the systems that carry agentic AI workloads.

What does the deal cover?

The scope of the collaboration, as described by the two companies:

• Front-end and back-end process technology
• Transistor development
• Advanced packaging, specifically Foveros-based 3D stacking
• Interconnect technology
• Power delivery

Foveros is Intel's 3D face-to-face die-stacking platform, and the emphasis on interconnect density and thermal performance signals that packaging — not just transistor scaling — is now the choke point for AI-class silicon.

Why the R&D model is changing

"With the chipmaking roadmap becoming more complex at each process node, accelerating innovations in semiconductor devices, interconnects and packaging requires deeper collaboration across R&D and manufacturing," said Dr. Prabu Raja, president of the semiconductor products group at Applied Materials. "We look forward to working closely with Intel as a founding partner of the EPIC Centre to shorten development cycles and bring new technologies to market faster."

Naga Chandrasekaran, executive vice president, chief technology and operations officer and general manager of Intel Foundry, framed the deal around AI infrastructure timelines.

"Accelerating our R&D to keep pace with the rapid buildout of AI infrastructure is critical to delivering the technology, manufacturing excellence and foundry services our customers expect from Intel Foundry," Chandrasekaran said. "Applied Materials' new EPIC Centre provides an environment where our teams can collaborate even more closely and at higher velocity to accelerate materials development and process innovations."

The premise is straightforward. Each new node raises the number of materials, integration steps and failure modes. Toolmakers that once delivered equipment against a spec sheet are being pulled earlier into co-development, and Applied's EPIC Centre is its physical bet on that shift — a Silicon Valley R&D hub where chipmakers, suppliers and researchers share infrastructure before technologies transfer to volume manufacturing.

For Intel, the collaboration feeds its foundry pitch. Faster materials and process development at Hillsboro translates into roadmaps it can offer external customers with more confidence.

A pattern of three-way alignment

The Intel–Applied agreement is not an isolated move. Earlier this year, Applied announced a partnership with SK Hynix to co-develop and deploy next-generation DRAM and high-bandwidth memory. In parallel, SK Hynix has held talks with Intel about a memory joint venture at Intel's Ohio fab site.

The three threads connect. Applied deepens its process co-development with a logic maker and a memory maker; SK Hynix secures both HBM technology and potential US capacity; and Intel builds an ecosystem around its foundry and packaging assets. For AI systems, advanced logic, 3D packaging and HBM are increasingly co-designed — and the partnerships now mirror that integration at the supply-chain level.

What to watch

No capacity, revenue or timeline figures accompany the announcement, and neither company committed to specific node targets. The measurable outcomes to track are the EPIC Centre's opening later this year, the pace of joint development cycles cited by Raja, and whether the SK Hynix memory JV in Ohio advances — all of which will determine whether this collaboration shortens Intel's path to competitive foundry and packaging services for AI customers.