Beijing time September 29, according to foreign media reports, the draft IPO prospectus that U.S. generative AI unicorn Anthropic previously submitted to the U.S. Securities and Exchange Commission (SEC) has been exposed. This 261-page prospectus document, for the first time, fully reveals the core financial cards of this frontier AI lab to global investors.
The prospectus shows that Anthropic's target valuation for this IPO exceeds $2 trillion, attempting to set a new record for global tech stock listings. As recently as May this year, the company's own valuation forecast was still at $965 billion. In just a few months, the valuation expectation has more than doubled. Even more astonishing is that in the draft prospectus, Anthropic used 80 pages to warn about the risk factors of its products.
Perhaps the person who dislikes these "risk warnings" the most is Jensen Huang of NVIDIA (NASDAQ: NVDA). His views often run counter to Anthropic founder and CEO Dario Amodei. For example, over the past year, Jensen Huang has been refuting certain of Amodei's views on AI at almost every occasion without naming him. Half a month ago, Jensen Huang was still mocking Amodei, this time on a speech stage, and the reason for the mockery was that he strongly opposed the AI doomsday theory that Amodei is "selling." He previously said bluntly: "I disagree with almost every word he says."
Jensen Huang Gives Amodei No Face
At the recent Dreamforce conference, Amodei and Jensen Huang took the stage one after another. Facing questions from Salesforce founder and CEO Marc Benioff, the two gave completely opposite answers. Amodei had just emphasized that AI needs globally agreed standards. Jensen Huang then said: existing market forces are enough to restrain AI companies, and no new laws or regulation are needed.
In fact, on many public occasions, Jensen Huang does not need to spare Amodei's feelings at all. He has been refuting, without naming him, Amodei's view on almost every occasion that AI will bring massive unemployment. At the 2025 VivaTech conference, Jensen Huang said even more directly: "I disagree with almost every word he says. He thinks AI is so terrifying, yet he thinks only they should be the ones doing this." At the time, an Anthropic spokesperson countered that Amodei had never claimed only Anthropic could build safe and powerful AI. Amodei himself said on the Big Technology podcast that Jensen Huang's accusation was an "absurd lie," and rarely talked about his father's death. He said that was precisely the root of why he was both eager to advance beneficial AI and constantly warning about its risks.
Earlier this year, in an interview with foreign media, Amodei said bluntly: "I want to make this clear and strongly refute this claim. In every interview, I talk about possible ways to address these risks, from tax and microeconomic policy to new jobs." He also complained about the communication mechanism itself: "Those three-second clips on social media make people think this is a marketing gimmick."
Of course, Amodei also does not want to offend Jensen Huang, and the two maintain a subtle relationship. On the one hand, the two publicly refute each other on AI dangers; on the other hand, NVIDIA (NASDAQ: NVDA) has invested in Anthropic and is also its most important computing chip supplier.
"AI Will End Humanity? Complete Nonsense"
Jensen Huang's public stance on AI risk can be broken down into three mutually supporting arguments. First, doomsday theory itself is harmful. On the No Priors podcast in January this year, his wording was quite complete: "Doomsday narratives do not help people, do not help the industry, do not help society, and do not help the government." This summer he told Axios that the claim that AI will lead to the end of humanity is complete nonsense.
Second, AI will not cause mass unemployment. Over the past year, Jensen Huang has said publicly many times that the narrative that AI destroys jobs has it "completely backwards" — AI automates specific tasks, while the total number of jobs the economy needs will actually increase. He believes, "If the world has no new ideas, then productivity gains will translate into unemployment." His advice to workers afraid of AI is: "You are just confusing your job with the tools you use to do your job." Amodei predicted in 2025 that AI might eliminate half of entry-level white-collar jobs, and Jensen Huang's assessment of that was merciless: "Complete nonsense." It should be noted that Amodei later softened his wording and shifted to emphasizing more the boost AI gives to productivity, but still insisted that job losses may be an "inherent" attribute of this technology.
Third, safety is an engineering problem, not a legal problem. This is the core argument for his opposition to regulation. In Jensen Huang's view, existing U.S. product liability law and laws related to unauthorized access are already sufficient. He also added that he agrees with President Trump's view — the AI industry does not need additional regulatory guardrails. Jensen Huang did not deny that accidents involving AI going out of control exist: "Some accidents have indeed happened, and fortunately those accidents did not cause harm."
Jensen Huang Has Become the Most Active Promoter of the U.S. Open-Source AI Camp
To understand Jensen Huang's position, the most effective way is not to analyze his arguments, but to look at his balance sheet. What does the AI wave mean for NVIDIA (NASDAQ: NVDA)? Its most recent full fiscal year revenue was $215 billion, while in fiscal 2021 it was only $17 billion, an increase of nearly 13 times in five years. NVIDIA (NASDAQ: NVDA)'s market value has soared even more, exceeding $5.2 trillion, driving Jensen Huang's personal fortune to nearly $200 billion and putting him among the top eight global super-rich.
Look at NVIDIA (NASDAQ: NVDA)'s latest quarterly earnings: fiscal 2027 second-quarter revenue of $96.22 billion, up 106% year over year; net profit of $59.7 billion, up 126% year over year; data center business revenue of $89 billion, surging 117% year over year. Apple, widely recognized as a money-making machine, had quarterly revenue of $109.4 billion in the same period, $13.2 billion more than NVIDIA (NASDAQ: NVDA); but Apple's net profit was only $29.8 billion, just half of NVIDIA (NASDAQ: NVDA)'s.
At present, NVIDIA (NASDAQ: NVDA) is almost tied to the entire AI industry. As of April this year, NVIDIA (NASDAQ: NVDA) had invested more than $30 billion in OpenAI, with a commitment of up to $10 billion to Anthropic; the total value of private company holdings recorded as "non-marketable securities" on its balance sheet reached $43 billion, double the previous quarter; there were also about $30 billion in marketable securities, including Intel stock bought for $5 billion in the fall of 2025 and now worth about $22 billion.
That is not all. NVIDIA (NASDAQ: NVDA) provided up to $105 billion in financing guarantees for OpenAI's data center in Ohio; together with Apollo, BlackRock, Blackstone, Brookfield and Goldman Sachs, it built a computing financing platform with a total scale of $500 billion; it also personally leased a Hut 8 data center in Texas under a 15-year lease with a base contract value of $19.6 billion, then transferred it to its invested cloud service provider Lambda to take on Anthropic's $35 billion order.
This business logic also explains why he is the most active promoter of the U.S. open-source AI camp. Because open source disperses computing demand to thousands of enterprises and developers, rather than concentrating it among a few super-large customers. In August, NVIDIA (NASDAQ: NVDA) acquired Hugging Face, the default distribution platform for open-source models, for about $13 billion; in addition, its previous licensing deal worth about $20 billion with inference chip designer Groq also triggered a regulatory investigation and was seen as a disguised acquisition to evade antitrust review.
In other words: the faster the AI industry runs, no matter who wins, the money flows to the same place — NVIDIA (NASDAQ: NVDA). And any form of slowdown — whether voluntary coordination or statutory regulation — will directly slow the speed of NVIDIA (NASDAQ: NVDA)'s money-absorbing machine. Of course, this does not mean Jensen Huang must be wrong in refuting AI doomsday theory. But it does mean that among all those participating in this debate over AI risk and slowdown, he is the one with the highest degree of vested interest.
However, Jensen Huang does not need to win this debate at the level of volume or logic. He not only controls the computing heart of the AI industry, but also influences the regulatory policy of U.S. authorities. U.S. Treasury Secretary Bessent said at a House hearing: on the issue of AI risk and regulation, "the president and NVIDIA (NASDAQ: NVDA) CEO Jensen Huang are completely aligned." The implication of this statement is that Jensen Huang may be the AI industry giant President Trump trusts most, even more than most entrepreneurs.