Anthropic's $2 Trillion IPO Faces Doubts: Risks of Mismatched Computing Commitments and Revenue

According to Woofun AI, as Anthropic is about to initiate its IPO process, the market’s doubts have surged abruptly. Independent research firm New Constructs has released a report titled “The Most Absurd IPO of 2026,” directly criticizing the fundamental flaws in its valuation logic. The core contradiction centers on the severe mismatch between the flexibility of client revenue and the rigidity of long-term compute commitments—a structural risk that is triggering sharp volatility across the global technology equity markets. As the sell-off spreads throughout the AI supply chain, Anthropic’s fundraising outcome will not only determine its own R&D and procurement capabilities but will also reshape semiconductor investors’ expectations of downstream order stability.

Behind the valuation disputes lies an intense contest between financial models and growth expectations. New Constructs offers a valuation of only $150 billion, approximately 93% lower than the market's target of $2 trillion—a massive gap stemming from deep skepticism about revenue sustainability. Anthropic’s annualized recurring revenue (ARR) is projected to reach $85 billion in September. Based on this figure, the $2 trillion valuation translates to a nearly 20x price-to-sales ratio (P/S), which appears extremely expensive.

However, Amont Partners Managing Director Rob Li points out that valuation rationality depends on growth rates: If ARR reaches $100 billion by year-end and doubles to $200 billion next year, the forward P/S would drop to 7–8x, entering a reasonable range.

Notably, Anthropic’s ARR has skyrocketed from $9 billion in January to $65 billion in July, growing more than sevenfold within the year. Woofun AI’s compiled data shows that, using New Constructs’ $150 billion valuation divided by ARR of $65–80 billion, the P/S is about 2–2.3x—lower than the SaaS sector median of 4.2x and generative AI-focused companies’ 4.7x median. For high-growth, unprofitable companies like Anthropic, the price-to-earnings ratio (P/E) is rendered ineffective due to strategically suppressed or volatile profits, making P/S a more suitable metric. However, the key lies in whether growth rates can continually absorb the high valuation.

Historical cases provide multidimensional references for today’s valuation debates. In March 2000, Cisco’s (CSCO.US) P/S soared to 30x. Despite revenue growth, its share price crashed 86% and has never fully recovered, warning against the risk of high multiples unmatched by growth. In contrast, at the end of 1999 Amazon (AMZN.US) had a P/S around 20x and was losing money, but while its share price fell 95%, it became a long-term winner, proving the validity of high valuation under sustained growth. Google’s (GOOGL.US) 2004 IPO reflected a P/S of about 7x—then considered high, but subsequent surges in revenue and profit made the valuation appear cheap in hindsight. These examples show that success hinges on a company’s ability to validate initial valuation with sustained high growth, rather than purely on multiples. Although Anthropic holds a solid position in the model field, the challenges it faces relate to mismatches between expenses and revenues, making its valuation logic more complex than traditional SaaS companies and requiring analysis tailored to its specific business model.

The critical pain point in its business model is the structural mismatch between rigid compute expenditures and flexible customer demand. Major clients like Microsoft (MSFT.US) and Meta (META.US), each with annual spending over $1 billion, have publicly stated intentions to reduce external model usage and increase adoption of their own models to cut costs. These two giants contribute more than a quarter of Anthropic’s revenues, forming the core income base, but collaboration remains on a short-term, on-demand purchasing basis—with no volume, price, or long-term performance commitments.

This means enterprise clients can switch models, reduce usage, or scale back budgets at any time, while Anthropic’s agreed-upon compute payments are difficult to scale back proportionally. If new capacity is not fully utilized, the company must still pay for resources without realizing the anticipated income, depleting cash reserves. The mismatch between the timing of compute investment and contract expiration further dictates that Anthropic must prepare sufficient funds at specific points, and this mismatch in cash flow rhythms constitutes a core operational risk.

Macroeconomic headwinds and market sentiment disruptions intensify short-term volatility. Ongoing Middle East conflicts, this week’s US CPI data vastly surpassing expectations, and surging rate hike expectations have clearly pressured rate-sensitive technology stocks. Rob Li notes these challenges typically serve as market bearish arguments and are difficult to disprove in the short term. Momentum-based trading by hedge funds amplifies AI supply chain volatility, and any monthly revenue data falling slightly short can trigger panic. According to Fidelity Fund’s Head of Investments, Nie Yixiang, in a high-interest rate environment, growth stocks—akin to “long-duration bonds”—are suppressed, and bear markets prioritize cash flow and balance sheet strength. Early investors' selling pressure cannot be ignored: in 2023, Anthropic’s valuation was only several billion dollars, but today it stands at $2 trillion. These investors are highly motivated to sell shares after the lock-up period ends—a structural reality rather than a commentary on company quality. This phenomenon can also be observed in SpaceX’s (SPCX.US) share price volatility.

Future growth verification points will focus on enterprise adoption and new industry expansion. Rob Li believes key points to monitor include existing customer upsell, new customer acquisition, and expansion into new sectors like healthcare and insurance. Enterprise implementation is the core of sustainability—whether end clients like Starbucks (SBUX.US) and Walmart (WMT.US) are willing to consistently pay and achieve positive returns is critical. Currently, most enterprises are in early-stage experimentation, and financial institutions have yet to adopt at scale due to security concerns. Investors should observe actual usage and payment willingness through intermediary service providers. Taking Salesforce (CRM.US) as an example, enterprise spending growth has accelerated from 20% at the beginning of the year to an expected 40% by year-end, indicating significant growth.

Additionally, cloud providers and AI models have a "symbiotic relationship." Microsoft (MSFT.US), Amazon (AMZN.US), and Google (GOOGL.US) profit by providing compute infrastructure, and AI application booms directly drive cloud service revenue growth. If AI demand is refuted, cloud businesses will be hit hard, and vice versa.

While cloud providers may develop their own models to reduce costs, their overall interests are tied to AI ecosystem expansion, and their ability to apply proprietary models remains in doubt—providing a buffer for model companies like Anthropic.

Within the industry’s hardware segment, storage and optical modules face diverging risk profiles. Currently, GPUs, semiconductors, and optical modules are all in "oversold" or shortage states, but investors prefer to "look ahead." Storage is the tightest link, but SK Hynix (SKHY.US) and Samsung’s share prices have plummeted nearly 30%. Rob Li points out that storage has the highest safety margin, but market hesitation stems from its strong cyclical nature and the historical risk of customers suddenly slashing orders.

However, AI has partially smoothed out memory cycles, and the memory industry now has more mature safeguard mechanisms, such as requiring customers to pay 20%–30% upfront or as deposits. Even in case of default, suppliers can retain part of the funds, reducing tail risks. In contrast, though US optical module sectors are in short supply, their valuations are extremely high, with insufficient cash reserves for protection—making downside risks more acute if the bubble bursts. Morgan Stanley (MS.US) notes that it’s hard to predict inflection points for storage companies; cyclical turning points in 2000, 2016, 2019, and 2022 were all caused by supply catching up and cloud vendors pausing purchases. Considering the time required for capacity expansion, it is unlikely that similar events will occur during most of 2026 and 2027, while other AI hardware segments lack prepayment mechanisms. Investors’ memories of historical "supply cuts" result in continued doubts about sustainability, possibly explaining the recent sharp declines in the AI supply chain.

To summarize the market cycle and investment outlook: despite the AI supply chain’s strong fundamentals, investors remain cautious amid unresolved macro headwinds. In each cycle, theoretical justifications drive surges, but during downturns, any slight disturbance becomes a trigger, as the market spontaneously searches for reasons. Debates on Anthropic’s IPO are not only about valuation but also serve as a stress test for the sustainability of its business model. As matters become clearer, enterprise adoption progress, macroeconomic changes, and the pace of early investor sell-offs will become key variables.