The plan to build 36 wafer fabs at a cost of 220 trillion yuan was criticized by TSMC executives.
Altman visited multiple countries in 2023, from the United Arab Emirates (UAE) to the United States, and held talks with multiple chip manufacturers in Asia, including TSMC, Samsung and SK Hynix.
According to the New York Times, Altman told TSMC executives that it plans to invest US$7 trillion in the next few years to build 36 wafer fabs and related data centers to support the development of AI technology. However, when he proposed this plan to TSMC executives, TSMC executives expressed high doubts about the feasibility of the plan and even nicknamed Ultraman "Podcast Brother", mocking his overly idealistic thinking and being completely divorced from reality. .
Altman's proposal hopes to combine Asia's manufacturing capabilities, the financial muscle of the Middle East, and the regulatory support of the United States to promote further expansion of AI computing. The amount of the proposal is equivalent to a quarter of the United States' annual GDP. TSMC executives believe that implementing even a small part of it will face extremely high financial risks and technical challenges. In addition, the construction of a wafer fab takes several years, and how to quickly achieve the goal in a short period of time is a big problem. TSMC Chairman Wei Zhejia even bluntly stated at the 2024 Annual Shareholders Meeting that "Ultraman's plan is too radical."
Why is TSMC skeptical of Altman’s proposal?
The scale of Ultraman's plan is too large. Even for a giant like TSMC, building several fabs is already a very challenging task, let alone 36. TSMC executives are worried that such large-scale investments will bring unpredictable risks, especially in a context where the semiconductor industry is already highly competitive and demand is changing rapidly.
Secondly, TSMC has reservations about Altman’s lack of experience in chip manufacturing. Although OpenAI has made remarkable achievements in the field of AI, semiconductor manufacturing requires huge technology investment, supply chain management and manufacturing capabilities. Altman's plan seems to ignore these practical challenges and rely solely on its vision for the future of AI. It seems too naive to realize such a large-scale plan.
Finally, Ultraman’s source of funding has also attracted attention. He has reportedly sought investment from the United Arab Emirates and floated the idea of the country funding billions of dollars to build wafer fabs. However, such a proposal could raise national security concerns. The U.S. government has expressed concerns about this, believing that allowing countries like the United Arab Emirates to lead the construction of AI infrastructure may indirectly allow China to obtain relevant technologies, thereby threatening U.S. technology and national defense security.
OpenAI’s financial challenges and executive turmoil
In addition to facing external doubts about the plan, OpenAI is also facing huge financial pressure and high-level turmoil internally. According to reports, OpenAI’s annual revenue is about $3 billion, but its expenses are as high as $7 billion. This financial situation cannot reassure investors. Although Altman firmly believes that AI will be as ubiquitous as electricity in the future, the current application of AI technology has not yet reached the "killer application" stage, leading to a lot of uncertainty in the market about its prospects.
In addition, OpenAI’s top executives have recently resigned one after another, including Chief Technology Officer Mira Murati, Chief Operating Officer Bob McGrew, and Senior Research Director Barret Zoph. These high-level departures further exacerbate concerns about OpenAI’s future.
[Disclaimer] There are risks in the market, so investment needs to be cautious. This article does not constitute investment advice, and users should consider whether any opinions, views or conclusions contained in this article are appropriate for their particular circumstances. Invest accordingly and do so at your own risk.
This article is reprinted with permission from: "Digital Age"
