GlobalFoundries will manufacture a key component used in AI chips for TSMC, according to a report published by MarketScreener. The deal links the world's two largest contract semiconductor manufacturers — one dominant in advanced AI processors, the other long focused on mature and specialty process nodes.

The announcement marks a notable shift in strategy for both companies. TSMC currently produces the overwhelming majority of advanced AI accelerators for customers such as Nvidia, AMD and major cloud providers, while GlobalFoundries exited the race for leading-edge lithography nodes in 2018 and has concentrated since on trailing-edge, specialty and differentiated processes.

What does the agreement cover?

The arrangement assigns GlobalFoundries a production role for a component that MarketScreener describes as critical to AI chips. The report did not initially specify:

• which component or process node is involved;
• the financial terms or volume commitments;
• a start date for production or the customer end-points in the chain.

Under its CEO Thomas Caulfield, GlobalFoundries has repeatedly stated that the company sees demand growth in AI-adjacent markets — power delivery, packaging-adjacent components, silicon photonics and radio-frequency parts — rather than in the GPU logic itself. The new agreement appears consistent with that positioning: supplying pieces of the AI hardware stack without competing at the leading edge.

Why would TSMC outsource part of an AI chip?

Capacity pressure offers the most direct explanation. Demand for AI accelerators has strained TSMC's advanced packaging and wafer capacity, and analysts have tracked allocation tightness across CoWoS and related lines since 2023. Moving a non-leading-edge component to an external partner frees internal capacity for the highest-value silicon.

For GlobalFoundries, the deal provides volume anchored to the fastest-growing segment of the semiconductor market without requiring the capital expenditure of a return to advanced nodes.

What is the market impact?

The cooperation signals that AI supply chains are becoming more distributed. Reports of the deal follow other examples of large chipmakers dividing production across specialty foundries, as system complexity pushes component sourcing beyond a single fab network.

Neither company has publicly commented beyond the initial report. MarketScreener listed the story without disclosed contract value, timeline or technical specifications.

"Die Signal" will update this article as companies confirm details of the agreement.