Samsung’s 2nm Push Faces a 66.6-Point Foundry Gap With TSMC

By Rebecca Stone

Samsung Electronics’ July 2024 contract manufacturing agreement with Tesla is worth about 2.28 trillion won ($16.5 billion), with Tesla’s next-generation AI6 chip slated for Samsung’s 2nm process at the Taylor, Texas, plant.

Can Samsung close the foundry gap?

At the July 2025 Sun Valley Conference in Idaho, Lee Jae-yong attended with Han Jin-man, the foundry division’s head, rather than the global marketing chief who accompanied him the previous year. The choice made the commercial priority visible: winning foundry accounts.

Four years into his chairmanship, Lee has made Samsung’s foundry business the company’s top goal for four consecutive years, according to company insiders and outside specialists. His backing has continued despite chronic losses and sluggish operations. The Tesla contract, signed after Lee’s legal risk was resolved, gives that strategy a major customer test.

A senior Samsung official said: “Whether in public or private settings, Lee has shown a strong vision for the foundry business.” The official added that the philosophy has continued since Lee’s vice-chair days.

Samsung announced a 133 trillion won investment in system semiconductors in 2019 and raised the plan to 171 trillion won in 2021. Even after the larger pledge, foundry operations remained mired in chronic losses and sluggish performance. Samsung has not abandoned the turnaround, instead reinvesting operating profit earned from memory chips into foundry fundamentals and growth.

What do the new orders show?

The order book now extends beyond Tesla. In July 2025, Samsung reached an agreement with Broadcom to apply sub-2nm process technology. The language processing unit (LPU) for Nvidia’s Grok 3 LPX will be produced on Samsung’s 4nm node. Samsung’s DRAM base die for sixth-generation high-bandwidth memory, HBM4, entered 4nm mass production.

Elon Musk said he discussed cooperation with Lee by video call as Samsung signed the Tesla agreement.

At its January 2025 earnings conference call, Samsung said it expected 2nm order backlogs to expand by more than 130% from the prior year. That is a company forecast, not a reported order total; production yield, customer wins and fab utilization will determine how much revenue follows.

Analyst data shows the commercial distance still to travel. TrendForce estimated last month that TSMC’s foundry market share reached a record 72.5% in Q2 2025, while Samsung slipped to 5.9% from 6.5% in the previous quarter.

In Q2 2022, before Lee became chairman, TSMC held 53.4% and Samsung 16.5%, a gap of 36.9 percentage points. Four years later, the gap had widened to 66.6 points.

Can restructuring restore profitability?

The foundry business is estimated to have posted an operating loss of around 1 trillion won in Q3 2025. Production-yield problems, weak customer wins, low utilization and preparation costs for the Taylor plant’s year-end startup have compounded the burden.

Management has reportedly announced a plan to phase out the currently profitable 8-inch wafer business and focus on advanced nodes as part of structural reforms.

Samsung was expected to disclose progress on narrowing losses, production yields and utilization at its Q3 earnings call on October 29. Backlog volume does not become revenue if yields stay weak or equipment utilization stays low.

Over the past two to three years, Samsung has paired those investments with personnel moves. In the November 2024 executive reshuffle, the company put Han Jin-man in charge of the foundry unit alongside President Nam Seok-woo, creating a two-president structure.

Sales came first.

Han, a sales veteran who led Device Solutions America in Silicon Valley, handles major technology customers; Nam, the chief technology officer, oversees technology.

Samsung had also released photographs of Lee’s May 2023 meeting with Tesla CEO Elon Musk, with Han present as an executive vice president heading DSA. Those appearances show how the chairman’s outreach has increasingly included the person responsible for foundry sales.

At a June 2025 business update, Han said Samsung’s foundry business could swing to profit in 2026 under the existing bonus framework. Losses could continue, however, if special management performance bonuses were reflected in results. His longer-range assessment was blunt: “Ultimately, management is responsible for the losses,” he said, adding, “There is a high possibility of achieving profitability in 2028.”

One source familiar with Samsung argued that lifting foundry share from the high teens to above 20% would raise the quality and scale of Samsung’s semiconductor business and create a virtuous cycle for South Korea’s semiconductor ecosystem. That is why Lee and other senior executives keep reinvesting in foundry fundamentals and growth despite large losses.

The next checkpoint is more concrete: Samsung must convert its forecast of more than 130% 2nm backlog growth into higher utilization and better yields as Taylor starts. Otherwise, TSMC’s 66.6-point lead will widen before the possible 2026 profit arrives.