Samsung foundry hits 50% yields on 2 nm GAA node amid push for SF2P

Samsung's first-generation 2 nm GAA process has reportedly reached a 'relatively stable' 50% yield, as the company leans on upcoming Exynos chips and a refined SF2P node to regain foundry momentum.

Samsung’s 2 nm GAA node process, known as SF2, remains under close industry scrutiny as new insider reports attempt to clarify its progress. Late last year, conflicting accounts described the state of Samsung’s cutting-edge 2 nm GAA node, even as the company’s semiconductor division quietly began ‘mass production’ of next generation Exynos 2600 mobile processors. A new ZDNet report now claims that yields for Samsung’s first generation 2 nm wafer product have reached a ‘relatively stable’ 50% milestone, a figure that creates confusion because other internal sources had already suggested a similar yield level was achieved last September.

The reported 50% yield remains below what is generally regarded as suitable for high volume output. The article notes that ideally, closer to 70% yield rates are preferred for mass manufacturing of finalized chip designs, underscoring the work Samsung still needs to do to make SF2 competitive at scale. Throughout 2025, insider reports suggested that development of SF2 has not been smooth, but that company leadership has been determined to push the technology over the finishing line. Industry observers cited in the article believe that the alleged failure of Samsung Foundry’s 3 nm line has already driven some potential return customers to continue relying on rival foundry TSMC, putting more pressure on the 2 nm program to deliver.

The launch of the Exynos 2600 is framed as a crucial ‘proof of concept’ for Samsung’s 2 nm ambitions, offering a chance to lure high profile customers away from competing 2 nm and 3 nm offerings. The ZDNet report also highlights a second generation 2 nm GAA option, labeled SF2P, which is described as a refinement built on ‘additional growth engines’ and is said to have design guidelines distributed to partner firms. When compared to the first iteration, SF2P reportedly does better with: ‘12% improved performance, 25% improved power efficiency, and an 8% smaller footprint.’ An anonymous source stated that SF2P’s basic process design kit was completed mid way through last year, and ZDNet connects this upgraded node with a future ‘Exynos 2700’ system on chip, signaling Samsung’s roadmap beyond its initial 2 nm efforts.

55

Impact Score

Executives see limited Artificial Intelligence productivity gains so far

Corporate enthusiasm around Artificial Intelligence has yet to translate into broad gains in employment or productivity, reviving comparisons to the long lag between early computing breakthroughs and measurable economic impact. Recent surveys and studies show mixed results, with strong expectations for future benefits but little consensus on present gains.

Nvidia skips a new GeForce generation as Artificial Intelligence chips dominate

Nvidia is set to go a year without a new GeForce GPU generation for the first time since the 1990s as memory shortages and higher margins in Artificial Intelligence hardware reshape the market. AMD and Intel are also struggling to capitalize because the same supply constraints are hitting gaming products across the industry.

Where gpu debt starts to break

Stress in gpu-backed infrastructure financing is emerging around deals that lack the structural protections seen in the strongest transactions. Oracle, the Abilene Stargate project, and older CoreWeave debt illustrate different ways residual risk can surface when contracts, collateral, and counterparties fall short.

SK hynix starts mass production of 192 GB SOCAMM2

SK hynix has begun mass production of the 192 GB SOCAMM2, a next-generation memory module standard built on 1cnm LPDDR5X low-power DRAM. The module is positioned as a primary memory solution for next-generation Artificial Intelligence servers.

AMD taps GlobalFoundries for co-packaged optics in Instinct MI500

AMD is preparing a renewed manufacturing link with GlobalFoundries to bring co-packaged optics to its Instinct MI500 Artificial Intelligence accelerators. The move is aimed at improving bandwidth and power efficiency in data center systems by moving beyond copper-based interconnects.

Contact Us

Got questions? Use the form to contact us.

Contact Form

Clicking next sends a verification code to your email. After verifying, you can enter your message.