TSMC has become the biggest "risk" in the AI supply chain, according to Stratechery, as the company is said to have failed to factor in chip demand early, leading to supply bottlenecks.
TSMC's Reluctance Towards the AI Hype Early On Is Reportedly Costing the AI Industry Right Now
When you look at the AI supply chain today, there is no doubt that TSMC is a major part of it, given the company's extensive foundry services and its role in supporting the hyperscaler buildout we are now witnessing. The growth of HPC-focused orders flowing into the company is so aggressive that it has actually made NVIDIA TSMC's biggest customer, surpassing Apple's long-standing position. However, despite TSMC's recent CapEx increases and expansion ambitions, Stratechery reports that TSMC is a "risk" that the AI industry now faces, not on a geopolitical front, but rather by creating imbalances within the supply chain.
That lack of increased investment earlier this decade is why there is a shortage today, and is why TSMC has been a de facto brake on the AI buildout/bubble.
- Stratechery
The analyst argues that TSMC's skepticism about increased CapEx is one of the major reasons production lines are being bottlenecked, saying that the company under C.C. Wei has shown "mixed emotions" towards the hyperscaler buildout until recently. In a previous report, we discussed how TSMC plans to raise spending to $56 billion this year, even after TSMC's CEO "personally verified" that the hyperscaler buildout is legitimate in terms of financial commitments.
Apart from hyperscalers seeing increased delivery times by the likes of NVIDIA/AMD following supply constraints at TSMC, those who are involved in the custom silicon race, such as Microsoft, Google, Meta, and many others, are unable to place orders that could 'guarantee' adequate delivery times, based on what has been reported. We recently saw Microsoft showcase the Maia 200 AI chip featuring TSMC's N3B process, which is currently facing severe supply constraints. And, Stratechery argues that ultimately, the "risk" shifts to hyperscalers in the form of lost revenue.
Semiconductors aren't the only major bottleneck right now; advanced packaging is also heavily suffering. TSMC's CoWoS and derivatives are the leading choices for chip manufacturers right now, but because its production lines are not as extensive as those of semiconductors, the Taiwan giant faces significant constraints. And, given how advanced packaging is becoming an essential element of the AI supply chain, it is a necessity for TSMC to keep up with demand.
The analyst also explains why diversification within the foundry segment is a 'risky' move for HPC customers and ASIC giants alike, given that TSMC has built a supply chain and trust that are irreplaceable. Despite interest around the likes of Intel Foundry and Samsung, AI chip manufacturers see a much greater risk in outsourcing orders to entities other than TSMC, but in the longer run, companies will have to decide whether sacrificing "billions in revenue" by relying on the Taiwan giant would be the right move.
About the author: Muhammad Zuhair is a hardware and technology reporter for Wccftech, specializing in the semiconductor industry and the complex interplay between technology, manufacturing, and geopolitics. His coverage focuses on the corporate strategies and technological roadmaps of industry giants like TSMC, NVIDIA, Samsung, and Intel. Zuhair's expertise lies in deconstructing complex topics such as fabrication nodes (e.g., 2nm process), the economic impact of policies like the CHIPS Act, and the strategic development of AI infrastructure from NVIDIA, AMD and Intel.
Follow Wccftech on Google to get more of our news coverage in your feeds.
