Inside Samsung Foundry's Q3 2026 Profit Turnaround: 2nm Yields, Tesla's $16.5B Deal, and the Taylor, Texas Ramp-Up

2nm Yields and the Advanced-Node Push

At the core of the turnaround is Samsung’s second-generation 2nm Gate-All-Around (GAA) process. Yields on the 2nm node have climbed to an estimated 55–60% as of early 2026, according to Chosun Daily and other sources . While that figure remains short of the 70% threshold often associated with mass-production economics, analysts judge it sufficient to begin initial commercial output and attract new customer orders. Samsung itself confirmed in its Q1 2026 earnings presentation that utilization of its advanced-process lines had reached a peak level .

The 2nm ramp is central to winning high-value orders from companies designing AI accelerators, high-performance computing chips, and automotive processors. Samsung expects 2nm-related orders to grow more than 30% year-on-year in 2026, driven partly by the Tesla contract and by ongoing talks with other major customers in the U.S. and China .

The Tesla Mega-Deal and the Taylor, Texas Facility

The single biggest validation of Samsung’s foundry strategy arrived in mid-2025 when Tesla signed a wafer fabrication agreement worth approximately $16.5 billion . The contract, which runs through 2033, covers Tesla’s upcoming AI5 and AI6 autonomous driving chips — processors designed to run the company’s full self-driving software and its Optimus humanoid robot .

Production for Tesla will center on Samsung’s $37 billion Taylor, Texas foundry. The plant reached a key operational milestone on April 24, 2026, when Samsung held an equipment move-in ceremony, formally installing lithography and process tools for 2nm GAA production . Days later, ENF Technology became the first Korean materials company to begin shipping semiconductor process chemicals to the Taylor site from its nearby Kyle, Texas plant, a sign that the supply chain is now operational .

The production timeline for Tesla’s chips carries some nuance. Mass production at Taylor is widely targeted for the second half of 2026, but full-scale output is expected to ramp up in earnest in early 2027 . Some reports indicate that pilot production readiness will be achieved by the end of 2026, with Tesla’s AI5 chips starting in the second half of 2027, though others have flagged the possibility of partial early production in the second half of 2026 .

How the Foundry Turnaround Fits Into Samsung’s Semiconductor Surge

The foundry’s path back to profit is not happening in isolation — it’s unfolding during a historic upcycle in Samsung’s memory-chip business. In Q1 2026, the Device Solutions (DS) division posted its highest-ever quarterly operating profit, driven by surging demand for AI-linked memory such as HBM4 and DDR5 . Average memory selling prices surged roughly 146% compared to the 2025 full-year average, reflecting tight supply and explosive demand for high-bandwidth memory in data centers .

Samsung began mass-producing its industry-first HBM4 memory and used NVIDIA GTC 2026 to showcase HBM4E, its next-generation successor . That AI-memory boom is generating massive profits for the DS division, effectively buying time and creating financial cover for the foundry’s continued investment needs. KB Securities has forecast that Samsung’s DS division operating profit could reach KRW 64.2 trillion in 2026, an increase of roughly 79% year-on-year, with contributions from both memory pricing strength and foundry utilization improvements .

Within this broader surge, the foundry’s role is shifting from a persistent drain on earnings to a growth driver with real momentum. The combination of improved 2nm yields, record-high utilization, HBM base die production, and the launch of Tesla’s chip program at the Taylor site is what analysts point to when they describe the “structural improvement” now underpinning Samsung’s foundry trajectory .

Caveats and Risks

The Q3 2026 profit inflection, while broadly expected, is not guaranteed. Some reports from early 2026 noted delays at the Taylor fab, indicating that full mass production could slip into early 2027 . Depreciation expenses tied to the enormous Taylor investment — estimated to require more than KRW 5 trillion for 2nm equipment alone — will weigh on margins even as revenue grows . And yield improvement on the 2nm process must continue; the current 55–60% range is workable but leaves Samsung with less margin for error against rivals, most notably TSMC . What has changed is the evidence that Samsung’s operational levers — utilization, yield, and large orders — are all pointing in the right direction at the same time.