Samsung has secured a chip supply agreement with Broadcom worth more than $200 billion running through 2030, in what multiple outlets are describing as one of the largest contracts the semiconductor industry has seen.
According to CNBC, the partnership is a significant boost to Samsung's foundry business — the division that manufactures chips designed by other companies. HotHardware reports the contract covers 2nm AI chips, referring to Samsung's most advanced manufacturing process. TechSpot describes the arrangement as built around closer integration between Broadcom's ASIC designs and Samsung's manufacturing stack, with Broadcom's next-generation communications chips produced by Samsung.
The distinction matters. Broadcom designs custom AI accelerators — application-specific integrated circuits, or ASICs — for large technology customers building their own AI hardware rather than buying off-the-shelf parts. Broadcom does not own factories, so it needs a manufacturing partner to turn those designs into silicon.
That partner has overwhelmingly been TSMC, the Taiwanese contract manufacturer that dominates advanced chipmaking. Benzinga frames the Samsung agreement as intensifying the foundry race with TSMC, and TelecomLead and finance.biggo.com both cast it as a direct challenge to TSMC's leadership in the market. Fortune reported the deal as a $200 billion order for Samsung to supply chips to Broadcom.
Investors reacted quickly. TradingView and Yahoo Finance UK both reported that Broadcom's stock jumped on the news.
Samsung has spent years struggling to win marquee customers for its foundry operation, with questions persistently raised about its manufacturing yields at leading-edge nodes. A commitment of this size from a customer as demanding as Broadcom is a vote of confidence that money alone cannot buy.
It matters because the AI boom has been bottlenecked by a single dominant chip manufacturer, and a credible second source could ease supply constraints, spread geopolitical risk, and put competitive pressure on the prices everyone pays for AI computing.