According to Woofun AI, as Anthropic prepares to launch its IPO, market skepticism has surged. Independent research firm New Constructs released a report titled “The Most Absurd IPO of 2026,” directly criticizing its valuation logic for fundamental flaws. The core issue lies in the severe mismatch between the elasticity of customer revenue and the rigidity of long-term compute commitments—a structural risk triggering sharp volatility across global tech stocks. As the sell-off in the AI supply chain spreads, Anthropic’s fundraising outcome will not only determine its own R&D and procurement capabilities but also reshape semiconductor investors’ expectations regarding the stability of downstream orders.
Behind the valuation discrepancy lies a fierce debate between financial models and growth expectations. New Constructs’ valuation of just $150 billion is roughly 93% lower than the market’s $2 trillion target, reflecting deep skepticism about revenue sustainability. Anthropic’s annualized recurring revenue (ARR) is projected to reach $85 billion by September; at this level, a $2 trillion valuation implies a price-to-sales ratio of nearly 20x, which appears extremely expensive.
However, Rob Li, Managing Director at Amont Partners, noted that the valuation's reasonableness depends on growth rates: if ARR reaches $100 billion by year-end and doubles to $200 billion next year, the forward P/S ratio would fall to 7-8x, entering a reasonable range.
Notably, Anthropic’s ARR surged from $9 billion in January to $65 billion in July, representing more than a sevenfold increase within the year. According to data compiled by Woofun AI, if New Constructs’ $150 billion valuation is divided by Anthropic’s ARR of $65–80 billion, the P/S ratio is approximately 2–2.3x, below the median P/S of 4.2x for the broader SaaS sector and 4.7x for generative AI-focused companies. For high-growth, unprofitable Anthropic, the P/E ratio is ineffective due to strategically suppressed or highly volatile profits, making P/S a more appropriate valuation metric—but the key question remains whether the growth rate can sustainably justify the high valuation.
Historical cases provide multidimensional reference points for current valuation debates. In March 2000, Cisco (CSCO.US) reached a P/S ratio of 30x; despite revenue growth, its stock price plummeted 86% and failed to recover to its original level for years, illustrating the risks of valuations disconnected from sustainable growth. Conversely, at the end of 1999, Amazon (AMZN.US) had a P/S ratio of about 20x while still posting losses; although its stock fell 95%, it ultimately became a massive long-term winner, demonstrating that high valuations can be justified by sustained long-term growth. Google (GOOGL.US), which went public in 2004 with a P/S ratio of approximately 7x, was considered relatively expensive at the time, but its subsequent high revenue and profit growth made its valuation appear cheap in hindsight. These cases show that success depends not on the absolute level of valuation multiples, but on whether a company can validate its initial valuation through consistent, high growth. While Anthropic has established a strong position in the model space, its challenge lies in the mismatch between revenue and expenses, making its valuation logic more complex than that of traditional SaaS companies and requiring careful analysis of its specific business model.
The business model’s pain point lies in the structural mismatch between rigid computational power expenditures and clients’ elastic demand. Major customers such as Microsoft (MSFT.US) and Meta (META.US), each consuming over $1 billion annually, have publicly stated their intention to reduce reliance on external models and increase use of proprietary models to cut costs. These two giants account for more than a quarter of Anthropic’s revenue and form the core of its income base, yet their collaboration model remains short-term, on-demand procurement with no volume commitments, no price locks, and no long-term contractual obligations.
This means enterprise clients can switch models, reduce usage, or cut budgets at any time, while Anthropic’s contracted compute payments are difficult to reduce proportionally. If newly added capacity is not fully utilized, the company still incurs costs without generating expected revenue, leading to a decline in cash reserves. The mismatch between the timing of new compute investments and contract expiration dates further requires Anthropic to have sufficient funds available at specific points in time, making this imbalance in cash flow a core operational risk.
Macroeconomic headwinds and market sentiment disruptions have intensified short-term volatility. The ongoing conflict in the Middle East remains unresolved, and this week’s U.S. CPI data significantly exceeded expectations, causing a sharp rise in rate hike expectations and putting clear pressure on interest-rate-sensitive tech stocks. Rob Li noted that these challenges often serve as bearish justifications for the market and are difficult to disprove in the short term; hedge funds employing momentum-based trading strategies have amplified volatility across the AI value chain, where even minor misses on monthly revenue figures can trigger panic. Nie Yixiang, Investment Head at Fidelity Funds, stated that in a high-interest-rate environment, growth stocks—seen as "long-duration bonds"—are under pressure, and bear markets place greater emphasis on cash flow and balance sheets. Early investor selling pressure is also significant: when the company was valued at just tens of billions of dollars in 2023, it has now reached $2 trillion. These investors have strong incentives to sell their shares once lock-up periods expire—a structural reality, not a reflection of the company’s quality. This dynamic is clearly evident in the price volatility of SpaceX (SPCX.US).
Future growth validation points focus on enterprise adoption and expansion into new industries. Rob Li believes key indicators include upselling to existing customers, acquiring new customers, and expanding into new sectors such as healthcare and insurance. Enterprise adoption is central to sustainability, and it is critical whether end customers like Starbucks (SBUX.US) and Walmart (WMT.US) are willing to continue paying and achieve positive returns. Currently, most enterprises are still in early pilot stages, and financial institutions have not yet adopted at scale due to security concerns. Investors should monitor actual consumption and willingness to pay through intermediate service providers; for example, enterprise spending growth for Salesforce (CRM.US) rose from 20% at the beginning of the year to an expected 40% by year-end, indicating significant growth.
Moreover, cloud providers and AI models share a symbiotic relationship: companies like Microsoft (MSFT.US), Amazon (AMZN.US), and Google (GOOGL.US) profit by providing computing infrastructure, and the growth of AI applications directly drives revenue growth in cloud services. If AI demand proves unfounded, the cloud business will suffer significantly, and vice versa.
Although cloud providers may develop their own models to reduce costs, their overall interests are tied to the expansion of the AI ecosystem, and their ability to effectively apply proprietary models remains uncertain, creating room for model companies like Anthropic to thrive.
In the hardware segment of the industrial chain, storage and optical modules face different risk profiles. Currently, GPU, semiconductor, and optical module segments are all in a state of overwhelming demand or shortage, yet investors tend to look ahead. Storage is the most constrained segment, yet SK Hynix (SKHY.US) and Samsung shares have declined by nearly 30%. Rob Li notes that storage offers the highest safety margin, but market hesitation stems from its cyclical nature and historical precedent of customers being able to immediately reduce order volumes.
However, AI has partially smoothed out memory cycles; today, the memory industry has established more mature safeguards, such as requiring customers to pay 20%-30% upfront or deposit payments, ensuring suppliers retain part of the funds even in the event of default, thereby reducing tail risk. In contrast, sectors like U.S. optical modules, though scarce, are highly valued and lack comparable cash reserve protections; should a bubble burst, the downside risk would be greater. Morgan Stanley (MS.US) noted that the turning point for memory companies is difficult to predict—past cycle endings occurred in 2000, 2016, 2019, and 2022, all triggered by supply catching up and cloud providers pausing purchases. Given the time required for capacity expansion, similar conditions are unlikely to occur for most of 2026 and 2027. Meanwhile, other AI hardware segments lack upfront payment mechanisms, and investors, still wary from past experiences of supply disruptions, remain skeptical about sustainability—this may explain the recent sharp decline in the AI supply chain.
The market cycle summary and investment outlook indicate that, despite strong fundamentals in the AI industry chain, investors will remain on the sidelines until macroeconomic headwinds subside. In every cycle, rallies are justified by theoretical rationales, while sell-offs turn even minor developments into catalysts, as the market naturally seeks explanations. The controversy surrounding Anthropic’s IPO is not merely a valuation issue but also a stress test of its business model’s sustainability. As the situation becomes clearer, key variables will include enterprise adoption progress, shifts in the macroeconomic environment, and the pace at which early investors sell their holdings.