Key Points
- OpenAI CEO Sam Altman expressed deep personal empathy for former Nvidia CEO Jensen Huang, who recently resigned amid a scandal, calling him a longtime hero while noting his own ouster three years ago.
- The story is notable in South Korea because Altman recently visited Yongin to brief Samsung executives on his view that AI chip demand is being double-counted as competitors like AMD, Google, and Cere
- Altman also addressed OpenAI's security after a wiretapping incident, pushed back against fears of unchecked recursive self-improvement (RSI), and confirmed OpenAI plans to eventually go public.
Summary OpenAI CEO Sam Altman has expressed deep personal empathy for former Nvidia CEO Jensen Huang, who recently stepped down amid a scandal, calling Huang a longtime hero while acknowledging that "even heroes have flaws."
Speaking at Semafor's "Silicon Valley and the World" event on November 16 (local time), Altman reflected on Huang's pivotal role in the artificial intelligence boom. "It's hard to imagine the AI revolution without Jensen and the company he built," Altman said, noting that Nvidia placed an early bet on OpenAI just as OpenAI bet on Nvidia. When asked whether Huang should return to lead Nvidia, Altman demurred, saying he did not know enough to answer. However, he added that he felt a strong personal connection to the situation, referencing his own dramatic ouster by OpenAI's board of directors three years ago.
The remarks drew particular attention in South Korea, where the semiconductor industry closely watches both leaders. Altman recently visited Korea and shared his market analysis at an executive meeting with Samsung in Yongin, Gyeonggi Province—a hub of Korea's chipmaking ecosystem. During the Semafor discussion, Altman warned that demand for AI semiconductors may be double-counted across the industry. In the past, when Nvidia's graphics processing units (GPUs) held a virtual monopoly, chipmakers could estimate wafer demand simply by asking companies like OpenAI about their future needs. Now, with competitors like AMD, Google, and Cerebras entering the market, multiple suppliers assume that OpenAI or Anthropic will buy most of their output, creating an illusion of inflated demand—even though real demand for data center construction remains solid.
Altman also addressed recent espionage-related incidents at OpenAI, including the discovery of a wiretapping device in a conference room. While admitting that OpenAI's security is not perfect and that the company has occasionally been fooled, he said OpenAI has already lived through the reality that others are only now suddenly confronting.
On broader societal concerns, Altman pointed to public backlash against AI seen in the recent U.S. midterm elections, stressing that humanity must not hand over decision-making power about its future to AI. Addressing fears around recursive self-improvement (RSI)—where AI improves its own performance without human control—he clarified that while OpenAI uses AI to develop new models faster, it is not simply pressing a button and leaving AI to improve itself unchecked. Finally, on the prospect of going public, Altman said OpenAI will eventually hold an IPO despite the drawbacks, dismissing concerns that an earlier IPO by rival Anthropic would exhaust investor appetite.
Sponsored · Ad