AI · 2026-11-20
Sam Altman Says OpenAI Could Go Public Someday, While Warning AI Chip Demand May Be Overestimated
Sam Altman, CEO of OpenAI, said investors are generally optimistic about the AI industry and that OpenAI will likely go public at some point. He also cautioned that semiconductor demand forecasts could be inflated if multiple chipmakers are counting the same customer demand in their own outlooks.

As expectations for the AI industry continue to rise, Sam Altman, CEO of OpenAI, pointed to both the upbeat mood among investors and the uncertainty surrounding semiconductor demand forecasts. In an interview with Semafor, he said OpenAI would “go public someday,” though he did not give any specific timeline or plan.
Altman said skepticism around AI seems more noticeable among journalists than among investors. Still, he did not paint an entirely one-sidedly optimistic picture of the industry outlook. He explained that AI investment could see cycles of overbuilding and shortages, and that there would be periods when things go well and periods when they do not. Even so, he added that, overall, major value is being created while existing value is also disappearing, and investors understand that shift.
On the question of an IPO, he acknowledged that there are things that are easier to do while remaining a private company, but said it is a positive direction for more people to have the chance to invest in AI. In that sense, he was signaling an intention to eventually go public, though the interview itself does not reveal any timing or specific preparations.
Regarding semiconductor demand, he raised concerns that industry forecasts may be running higher than reality. Altman pointed to a situation in which several chipmakers expect to sell their output to customers such as OpenAI and Anthropic, noting the possibility that the same customer demand is being reflected separately in multiple suppliers’ forecasts. If those estimates overlap in this way, he explained, the industry-wide demand outlook could be inflated beyond what is actually needed.
That does not mean he believes demand for AI chips is weak. Altman said demand and data center construction are solid, while also noting that OpenAI cannot buy every product from every semiconductor company, nor does it need to. He said he expects massive computing demand to continue over the long term, but added that it is difficult to forecast supply and demand precisely on a quarterly or yearly basis.
The main point of his remarks is that expectations for AI growth and the uncertainty around investment and supply chains coexist. Expansion in AI services and data centers may support long-term semiconductor demand, but whether each company’s forecast turns into actual purchases is another matter. So when looking at industry outlooks, it is worth checking not only the total demand figure, but also whether demand has been double-counted across customers and suppliers. (Source: Yonhap Infomax, Semafor interview coverage)
Altman said skepticism around AI seems more noticeable among journalists than among investors. Still, he did not paint an entirely one-sidedly optimistic picture of the industry outlook. He explained that AI investment could see cycles of overbuilding and shortages, and that there would be periods when things go well and periods when they do not. Even so, he added that, overall, major value is being created while existing value is also disappearing, and investors understand that shift.
On the question of an IPO, he acknowledged that there are things that are easier to do while remaining a private company, but said it is a positive direction for more people to have the chance to invest in AI. In that sense, he was signaling an intention to eventually go public, though the interview itself does not reveal any timing or specific preparations.
Regarding semiconductor demand, he raised concerns that industry forecasts may be running higher than reality. Altman pointed to a situation in which several chipmakers expect to sell their output to customers such as OpenAI and Anthropic, noting the possibility that the same customer demand is being reflected separately in multiple suppliers’ forecasts. If those estimates overlap in this way, he explained, the industry-wide demand outlook could be inflated beyond what is actually needed.
That does not mean he believes demand for AI chips is weak. Altman said demand and data center construction are solid, while also noting that OpenAI cannot buy every product from every semiconductor company, nor does it need to. He said he expects massive computing demand to continue over the long term, but added that it is difficult to forecast supply and demand precisely on a quarterly or yearly basis.
The main point of his remarks is that expectations for AI growth and the uncertainty around investment and supply chains coexist. Expansion in AI services and data centers may support long-term semiconductor demand, but whether each company’s forecast turns into actual purchases is another matter. So when looking at industry outlooks, it is worth checking not only the total demand figure, but also whether demand has been double-counted across customers and suppliers. (Source: Yonhap Infomax, Semafor interview coverage)
Related search terms
#OpenAI IPO#Sam Altman interview#AI semiconductor demand#AI chip demand forecast#data center construction#Semafor interview#AI industry outlook