While TSMC makes promising strides in the 2nm process, slated for mass production in 2025, rivals Samsung and Intel are making headlines with aggressive moves to secure cost-effective deals. This surge in competition for the 2nm process is intensifying.
According to CNA, experts suggest that given the escalating rivalry in the AI chip market, it is paramount to have flawless execution in the process. Despite the recent efforts from Samsung and Intel, TSMC is anticipated to clinch the lion’s share of 2nm orders.
TSMC’s 2nm process is on track for mass production in 2025, with construction underway at its first 2nm fab in the Phase 2 Expansion Area of the Baoshan Site in the Hsinchu Science Park. The tool-in is scheduled for April next year. Simultaneously, TSMC’s Kaohsiung fab is earmarked as a crucial production base for 2nm in the future.
Samsung is gearing up for mass production of its 2nm process in 2025 as well. Reports from the Financial Times indicate that, in a bid to secure orders from industry giants like NVIDIA, Samsung is contemplating discounted offerings to challenge TSMC.
Intel, in its bid to reclaim its place in the semiconductor landscape, has set an big target of advancing 5 nodes in four years. Sanjay Natarajan, Senior Vice President at Intel, revealed in a recent interview with Nikkei Asia that the company aims to commence mass production of 2nm chips in 2024, with a commitment to providing reasonably priced products. Additionally, Intel’s 18A process is poised for trial production in the first quarter of 2024.
Analysts also share the insights of the competitive landscape among the three major semiconductor players, TSMC, Samsung, and Intel. Arisa Liu, the research fellow and director at the Taiwan Industry Economics Services of Taiwan Institute of Economic Research, notes that TSMC’s 2nm is expected to adopt a gate-all-around (GAA) architecture. In contrast, Samsung has taken an early lead by introducing GAA architecture in its 3nm, aiming to outpace TSMC in the 2nm process after 1 or 2 years of adjustments.
Liu highlights the challenges facing Samsung, noting that the 3nm GAA process has exhibited unstable yields. For example, Qualcomm, a major player in mobile chip, has reverted to TSMC for production. Even with Samsung’s plans for bidding at a reduced price, it is anticipated to face difficulties in denting TSMC’s order share in the short term.
Turning attention to Intel, Liu observes that Intel’s current process technology has advanced to Intel 4 and Intel 3, which do not align with the industry’s 4nm and 3nm processes. In reality, Intel’s progress is closer to 7nm or an enhanced version thereof. Notably, Intel’s 3nm products are still estimated to be outsourced to TSMC, signaling a substantial technology gap.
TSMC’s President, C.C. Wei, has previously outlined the company’s plan to initiate mass production of the 3nm N3P process in the latter half of 2024. Notably, its performance metrics, including power, performance, area (PPA), are expected to surpass Intel’s 18A.
Liu further pointed out the news of TSMC’s clients contemplating additional foundry partners has surfaced recently. This move is primarily seen as an attempt to exert pressure on TSMC and gain negotiating leverage. Given the fierce competition in the AI chip market, it is imperative to control precision in the manufacturing process. As such, TSMC’s 2nm is anticipated to secure a significant majority of orders.
(Image: TSMC)
Explore more