Anthropic’s Financials Revealed — The Losses Are Stunning

SUMMARY

Ed Elson hosts a discussion with Paul Kedrosky and Jay Ritter, analyzing Anthropic’s leaked S-1 financials, the risks of its upcoming IPO, and the broader implications for the IPO market. The episode also covers Oura’s postponed IPO and the Manchester City financial fraud scandal, highlighting concerns about transparency and trust in both markets and sports.

MAIN POINTS

• Anthropic's leaked S-1 reveals $4.6 billion in revenue, $8 billion in operating losses, and a net loss of $42 billion, with significant customer concentration risks. [1:02]
• Debate centers on whether training costs should be excluded from operating losses, with concerns about accounting practices and the sustainability of Anthropic's business model. [5:00]
• Anthropic faces a strategic dilemma: continuing high training costs to maintain a competitive moat or risking being undercut by industrial-scale competitors like China if they stop. [11:01]
• Oura postpones its IPO despite profitability and strong demand, citing market uncertainty and raising questions about investor appetite for new listings. [17:00]
• Jay Ritter explains that most companies delaying IPOs never go public, often due to investor skepticism about growth prospects and market size. [23:01]
• Manchester City is found guilty of financial fraud spanning 15 years, casting doubt on the legitimacy of their Premier League titles and raising broader concerns about systemic corruption. [27:02]
• The Manchester City scandal is used as a metaphor for widespread financial corruption, fueling public distrust in institutions and the economic system. [29:02]

DETAILED ANALYSIS

The leaked S-1 financials for Anthropic, a leading artificial intelligence company, have brought to light extraordinary figures that have sparked widespread debate among investors and analysts. The company reportedly generated $4.6 billion in revenue last year, a dramatic increase from the previous year, but this was overshadowed by an $8 billion operating loss and a staggering $42 billion net loss. A significant portion of the net loss—approximately $34 billion—was attributed to a non-cash charge related to the revaluation of financing instruments, but the operating loss remains a critical concern.

Notably, Anthropic’s revenue is highly concentrated, with two customers accounting for nearly a quarter of total revenue and six customers making up about 60%. This level of customer concentration is atypical for a company of Anthropic’s size and raises questions about the sustainability and scalability of its business model.

Paul Kedrosky, managing partner at SK Ventures, emphasized that such customer concentration is expected in smaller firms but is alarming at Anthropic’s scale. The risk is twofold: not only does the company rely heavily on a small group of early adopters, but these customers may not be representative of the broader market, casting doubt on Anthropic’s ability to diversify its client base. Furthermore, the discussion delved into the contentious issue of how Anthropic and similar AI firms present their financials.

There is a growing trend to exclude training costs from operating losses, arguing that these expenses are extraordinary or non-recurring. However, Kedrosky argued that training costs are fundamental to the ongoing operation of AI businesses, as new models require frequent retraining. He likened the practice of excluding these costs to the kind of “earnings before bad things” accounting seen in notorious cases like WeWork, which ultimately proved disastrous for investors.

The debate extends to the broader implications for Anthropic’s upcoming IPO, which is expected to value the company at $2 trillion—the highest ever for an initial public offering. Kedrosky warned that this valuation is disconnected from the company’s underlying financial health and suggested that the IPO is less a financing event and more an opportunity for insiders and early investors to exit their positions. He further noted that if Anthropic were to halt its costly model training and focus solely on inference (the application of trained models), it would lose its competitive edge and be vulnerable to industrial-scale competitors, particularly from China, who can produce AI tokens more cheaply.

This strategic bind means Anthropic and its peers may be forced into a niche, high-performance segment of the market, analogous to luxury car manufacturers, rather than achieving mass-market dominance.

The episode also examined the state of the broader IPO market through the lens of Oura, the smart ring maker, which recently postponed its own public offering despite being profitable and experiencing strong investor demand. Jay Ritter, director of the IPO Initiative at the University of Florida, explained that while stock markets are near all-time highs, investor skepticism about valuations and growth prospects persists. He pointed out that many companies that delay IPOs never end up going public, often due to concerns about market saturation and competition.

Ritter cited examples such as GoPro, Peloton, and SoulCycle, where initial rapid growth was followed by stagnation or acquisition, reinforcing the notion that not all promising companies are suitable for public markets.

Finally, the discussion shifted to the sports world, where Manchester City was found guilty of extensive financial fraud over a 15-year period, involving misrepresentation of financial statements and the use of sham contracts to circumvent spending regulations. This revelation casts a shadow over the legitimacy of the club’s achievements, including eight Premier League titles, and serves as a broader metaphor for systemic corruption in both sports and financial markets. The episode concluded by highlighting the erosion of public trust in institutions, as repeated instances of high-profile fraud contribute to growing disillusionment with capitalism and established systems.

LINKS

• Vote for Prof G Markets at the Signal Awards
• Prof G Markets YouTube Channel
• Prof G Markets newsletter subscription
• Order 'Notes On Being A Man' book
• Subscribe to No Mercy / No Malice
• Prof G Markets Instagram
• Scott Galloway Instagram
• Ed Elson Instagram
• Ed Elson X (Twitter)
• Ed Elson Substack
• Zero business finance platform

KEYWORDS