While competitors Samsung Electronics (Samsung) and Intel are struggling in their foundry businesses, TSMC is set to continue dominating large orders below 7nm through 2025, securing major clients in the AI chip sector, and its 2nm GAA process is scheduled to debut as planned. This has turned the competitive landscape among the three major foundries into a one-man show.
Market estimates suggest that Samsung Foundry's annual market share may fall below 10%. If Intel cannot successfully spin off its foundry business soon, its overall losses will likely continue to expand.
Prior market reports indicated that due to the US-China AI chip ban and ongoing lackluster demand in smartphone and PC end markets, the recovery in AI chip demand remains limited, suggesting that TSMC's revenue declines in November and December could exceed expectations.
Nevertheless, TSMC's utilization rates for its 3nm and 5nm processes reportedly remain fully loaded, allowing November revenue to still reach NT$270 billion to NT$275 billion (approx. US$8.3 billion to US$8.5 billion), consistent with forecasts, indicating that the impact of the AI chip ban appears manageable for TSMC.
TSMC chairman C.C. Wei stated bluntly that TSMC has no competitors, as nearly all global AI chip customers are under its control. With the timely introduction of the 2nm process, TSMC has left its rivals far behind.
Intel struggles strategically while Samsung struggles technically
On November 9, TSMC founder Morris Chang also pointed out the core reasons why Samsung and Intel's foundry businesses are struggling. He noted that Intel lacks both strategy and leadership, similar to four years ago when the board had no clear vision for future strategies. When selecting a CEO, eloquence alone is insufficient. In contrast, according to Chang, Samsung's issues are primarily technical rather than strategic.
A supply chain insider believes that securing external customer orders is difficult for Samsung and Intel because they operate under the IDM model. Previously, Samsung aimed to leapfrog TSMC with its 3nm GAA process but has yet to secure significant contracts, adversely affecting its own mobile and AI chip developments.
Massive capital expenditures have proven unrecoverable, and with its memory business lagging behind SK Hynix in the HBM arena, a vicious cycle ensues, potentially leading to greater-than-expected losses in its foundry operations, with market share projected to drop below 10% in 2025.
Intel faces even larger challenges, struggling to find a suitable CEO candidate. The sudden change without identifying a replacement first has led to internal morale plummeting. As Chang mentioned, Intel's problems lie in its strategy and management team. While its process technologies and patents remain strong, they are unable to be effectively utilized.
External customers are unlikely to choose a foundry known for inefficiency, high costs, and potential conflicts of interest. After years of stagnation, Intel has yet to discover a viable path to separate manufacturing from design, and prolonged delays will only exacerbate its financial woes.
Experts in the supply chain believe that the current perception in the market is that TSMC's biggest risks stem from expansion efforts by US manufacturers and profitability concerns, along with the stance of the US government. However, given the setbacks faced by Samsung and Intel, TSMC can shift pressure and costs onto its customers since they have no alternative options.
Moreover, TSMC's US clients account for over 70% of its revenue. If TSMC encounters difficulties, major US firms would also suffer, which is undesirable for the US government. Therefore, TSMC holds leverage. What the US government desires is "manufacturing," and TSMC is positioned to deliver that, particularly impacting US companies. By raising the technological and capital investment thresholds for foundry services, TSMC has effectively eliminated competition.
