Samsung's Foundry Just Got a Rare Taste of Pricing Power: Chip Prices Up to 15%

Samsung’s foundry just got a rare taste of pricing power.

Samsung Electronics raised prices on some advanced contract chipmaking services by up to 15% for new orders in July, according to two people familiar with the matter cited by Reuters. The increases — applied across the 4nm, 5nm, and 8nm process families — are one of the clearest signs yet that the AI chip capacity crunch has spread beyond TSMC, the world’s largest contract chipmaker.

It’s also a turning point for Samsung’s contract manufacturing business, which has run losses for an extended period while chasing the foundry leader. With orders spilling over from TSMC, Samsung finally has the utilization and pricing leverage it has been missing.

What happened

Prices for wafers produced on Samsung’s 4-nanometre SF4 process rose 10% to 15% for U.S. and Chinese customers in July versus the previous month, while Taiwanese customers saw smaller increases of 5% to 10%, according to one source. Wafers on the 5-nanometre SF5 node climbed 10% to 15%, and the older 8-nanometre line rose by nearly 10%.

Samsung declined to comment, citing a policy of not disclosing matters related to individual business operations or clients.

Why now: TSMC is full, and Samsung is filling up too

The root cause is straightforward: surging AI chip demand has pushed TSMC’s advanced-node capacity to saturation. Orders that once flowed almost exclusively to Taiwan are shifting to rivals, and Samsung is the clearest beneficiary. Its SF4 line at the Pyeongtaek complex has reportedly been running at full capacity since late 2025, producing logic chips for customers such as Qualcomm and the base dies used in Samsung’s own high-bandwidth memory (HBM) stacks.

Demand from Chinese customers has been particularly strong, the sources said. But Samsung cannot fill every order: the company must serve U.S. customers and reserve part of its capacity to support its own HBM production.

The overflow is bringing big names to Samsung’s door. The company secured chip production contracts from Tesla and Apple last year and announced an AI chip contract with Broadcom last month. Its 4nm process is also expected to be used for Nvidia’s new AI inference processor, and Reuters reported that Google is in talks with Samsung to manufacture semiconductors on SF4.

Why it matters for business buyers

For enterprise technology buyers, this is the moment a “second source” finally gains leverage. For years, advanced AI silicon meant one foundry: TSMC. That concentration created a bottleneck — and a negotiation problem — for every company trying to secure chips for AI workloads.

Samsung’s price hike is, in a sense, a healthy development: a credible alternative is absorbing overflow demand. But it also means the cost of AI compute is rising across the board, not just at the market leader. Chip designers will likely pass higher foundry costs to hardware vendors and, eventually, to buyers of AI infrastructure.

The increases could also accelerate Samsung’s foundry turnaround. With orders shifting from TSMC, the utilization and pricing power needed to restore profitability are finally in place — provided demand holds and capacity stays scarce.

Material risks to watch

The hikes aren’t guaranteed to stick. Samsung has historically competed for foundry share on price, sometimes undercutting TSMC dramatically at advanced nodes. If AI demand cools or TSMC’s capacity catches up, Samsung may have to trade pricing power back for volume.

Samsung also faces a structural constraint: every wafer dedicated to HBM base dies is a wafer not sold to external foundry customers. And with growth concentrated at the 4nm node, the foundry story is still narrow. The real test is whether Samsung can convert this order wave into lasting share at the cutting edge — 2nm and beyond — rather than a short-term surge.

Takeaway

The report confirms what procurement teams already suspected: the AI chip supply chain is no longer a one-vendor game, and diversification now comes with a price. Companies planning AI infrastructure should treat Samsung as a serious foundry alternative — but they should expect to pay for the privilege and compete for capacity alongside some of the largest names in tech.