AI Laboratory Anthropic has launched its long-awaited initial public offering roadshow. The company announced an issue price range of $70 to $78 per share, which will value this AI Laboratory at up to $1.4 trillion on a fully diluted basis, although it is still significantly lower than the $2 trillion reported by some news agencies a few weeks ago.
Anthropic secretly submitted its listing application in June and made its prospectus public on October 26. SpaceX sought a valuation of approximately $1.75 trillion when it went public in June. Anthropic shares may start trading as early as next Wednesday or Thursday.
Anthropic is issuing 825 million shares on NASDAQ under the code “ANT”, with the potential to raise up to $64.4 billion based on the upper end of the range. Seven strategic partners, including Google and NVIDIA, have also agreed to subscribe for $26 billion in non-voting shares at the price of IPO.
The updated files from Tuesday also show that as of the end of October, the annualized revenue operating rate for Anthropic was approximately 81 billion US dollars, which is higher than the around 71 billion US dollars in late September.
This time, the broader AI industry is going through a tough period. With the Democratic Party achieving a significant victory in the midterms, the AI sector has seen heavy selling due to concerns about stricter regulation and more stringent scrutiny, especially regarding data center construction, which has caused unease among investors. Nasdaq fell 5.5% last week, with its market value evaporating by nearly $2.5 trillion in a single day. Some analysts warn that the market is approaching a correction zone.
The turmoil at NVIDIA, one of the key strategic partners of Anthropic, has also intensified market tensions. The chip manufacturer's shares fell by more than 4% on Monday following the announcement by its board of directors to initiate an independent investigation after reports that CEO Jensen Huang had become a target of Chinese intelligence gathering activities; these reports also claimed he was in a romantic relationship with a woman named Alice Huang. She previously held a senior position at the Singapore-based cloud computing and data center company Megaspeed, and U.S. intelligence officials believe she is an asset of a major Chinese intelligence agency. Beijing has denied these allegations as unfounded, while NVIDIA stated that Jensen Huang fully supports this investigation.
Concerns regarding the mid-term elections and NVIDIA's future market prospects are compounded by another issue that Anthropic has previously made public: the security of the AI system they develop, as well as potential risks to its viability.
The prospectus for Anthropic also includes a new warning regarding the company's progress in recursive self-improvement ( RSI ). RSI refers to the stage where the system has become powerful enough to significantly build its own successor systems, and each generation of the system in turn accelerates the development of the next generation. Since this feedback loop theoretically could allow the improvement speed of AI to exceed human abilities to test, understand, or control it, RSI has always been a long-term security concern. The document for Anthropic acknowledges this risk and warns that using AI to build more powerful AI "may also bring considerable security risks, especially when it significantly accelerates the pace of development."
In a new revision document, Anthropic indicates that internal estimates for future models show that the company “may soon meet” the definition of significant acceleration in its responsible expansion policy, which can be attributed to “AI.” This policy sets this threshold as roughly doubling the progress rate of AI, and this acceleration is largely driven by AI's automation research institute. Anthropic states that currently, no existing model has crossed this threshold, and mentions that the company is considering modifying the policy “to maintain human control over the development of cutting-edge models.”
