Anthropic's Valuation Jumped 2.5x in Four Months—Does the $2 Trillion IPO Math Add Up?

Introduction: A Race Between Valuation and Fundamentals

When a company's valuation jumps 2.5x in four months and it plans to launch an IPO targeting $2 trillion, the market's first reaction is usually awe. But after the awe, someone has to open the books and check whether the logic behind that number holds up. Anthropic (mistakenly written as "Anthropeak" in the original, referred to as Anthropic throughout) is under exactly this spotlight. According to cross-verification from multiple sources including a Bilibili creator and the UK's Financial Times, its valuation went from $380 billion in the February funding round to $965 billion after the May Series H round, and now it is touting a $2 trillion IPO target—a pace that turns heads. Yet the latest quarterly data tells a more complicated story.

The Three-Step Valuation Leap: The Logic Chain from $380B to $2T

Funding Cadence and Valuation Anchors

Looking back at Anthropic's funding trajectory: a round in February this year had a post-money valuation of $380 billion, and the May Series H round's post-money valuation surged directly to $965 billion—2.5x in under four months. Now the $2 trillion IPO target is more than double Series H. What supports this target is the company's expectation of reaching $190–200 billion in revenue by 2028. On that basis, the $2 trillion valuation corresponds to roughly 10x forward price-to-sales (P/S). For a high-growth tech company, 10x forward revenue is not absurd—provided the high growth can be sustained.

Forward P/S (price-to-sales) is a company's current market cap (or valuation) divided by expected revenue over the next 12 months or a forecast year, rather than actual revenue over the past 12 months. Unlike the P/E ratio, P/S does not depend on profitability, so it is often used for companies that are not yet profitable but growing revenue rapidly. For a company like Anthropic, a $2 trillion valuation corresponds to expected 2028 revenue of roughly $190–200 billion, or about 10x forward P/S. For reference, mature SaaS companies typically trade at 5–10x, while high-growth phases can exceed 20x, so 10x itself is not extreme—the key is whether the expected revenue materializes.

The Real Curve of Revenue Growth

Lining up revenue over the past year, the trend matters more than the absolute numbers: full-year 2025 revenue was $4.59 billion; Q1 2026 alone was $4.7 billion—one quarter matched the entire previous year; Q2 was $11.6 billion, up 147% quarter over quarter; Q3 was $16.8 billion, with quarter-over-quarter growth down to just 45%. A 45% QoQ growth rate is certainly striking, but for a company sprinting toward a $2 trillion valuation, the key signal is that growth was "cut to a third" within a single quarter. It shows growth is not as fast as it appears on the surface, and whether Q2's high growth rate is entirely real is also questionable.

The Accounting-Basis Debate: The "Makeup" on Revenue Numbers

Gross Basis vs. Net Basis

A more subtle number: the $16.8 billion quarterly revenue is the result of Anthropic recognizing Amazon and Google Cloud distribution revenue on a gross basis, whereas OpenAI only counts amounts after deducting channel revenue share. For the same underlying business, the revenue two companies calculate using different methods can differ by 30% to 40%. According to the Financial Times, OpenAI's annualized revenue is nearly $50 billion, while a figure reported earlier by Ecosios was close to $70 billion—believed to be a conversion investors made themselves for a like-for-like comparison with Anthropic.

Gross basis vs. net basis are two accounting standards for revenue recognition. Under the gross basis, the full amount paid by the customer is recorded as revenue, while the share paid to channel partners is recorded as cost; under the net basis, only the net amount after deducting channel revenue share is recorded. When the distribution chain is long and the revenue-share ratio is high, the two methods can lead to a 30%–40% difference in revenue. When comparing companies' revenue side by side, the basis must be unified; otherwise, the revenue scale of companies using the gross basis will be overstated. Investors' self-conversion of OpenAI's annualized revenue was precisely to align it with Anthropic's gross-basis figure.

The "Electricity-Bill" Method for Gross Margin

Another widely discussed metric is the 85% gross margin, but this is calculated before deducting the cost of training new models. This method is inherently unreasonable, because model training is an ongoing process, not a one-time cost. Excluding training costs to calculate gross margin departs from economic reality. One commentator pointedly remarked: "You're not even counting training costs—is this gross margin after deducting the electricity bill?"

The Real Financial Picture: Losses, Dependence, and Maturity Mismatch

Operating Losses and Channel Dependence

The draft prospectus shows that in 2025 Anthropic had an operating loss of $8.06 billion and a net loss of about $42 billion, of which $34 billion was a non-cash expense from changes in the fair value of convertible financing instruments. Even excluding non-cash items, the operating-level loss is real. More noteworthy is that in 2025, revenue completed through Amazon AWS, Google Cloud, and Vortex AI accounted for about 47% of total revenue. Channel dependence itself is not objectionable, but the problem is that Amazon and Google are not just channels—they are also Anthropic shareholders (Amazon has invested more than $8 billion cumulatively, Google more than $2 billion) and the largest cloud computing providers.

Changes in the fair value of convertible financing instruments refers to instruments such as convertible bonds issued by a company, whose fair value increases as the company's valuation rises, requiring recognition of a corresponding non-cash expense under accounting rules (a loss for the issuer). Of Anthropic's $42 billion net loss in 2025, about $34 billion is this type of non-cash expense and does not involve actual cash outflow. After excluding it, the $8.06 billion operating loss better reflects the true cash burn at the operating level. Understanding this helps distinguish "accounting losses" from "cash losses" and avoids misreading accounting fluctuations as operational deterioration.

Mismatch Between Rigid Spending and Elastic Revenue

This multiple identity has two consequences. The first is a maturity mismatch between costs and revenue: the prospectus shows that over roughly the next 10 years, Anthropic has committed at least $518.1 billion in cloud computing and infrastructure agreements with six partners, about 80% of which is irrevocable. That money must be paid when due, regardless of whether there is revenue. Revenue, meanwhile, is elastic—in 2025, usage-based revenue was about $3.8 billion and subscription revenue only $789 million, and customers have no obligation to renew. Rigid spending against elastic revenue: this mismatch is invisible in a boom cycle but fatal in a downturn.

The Reflexive Loop: How Equity Gains Distort the Index

The second consequence shows up on shareholders' books. Take Amazon: the change in fair value of its Anthropic stake was recognized as a $16.8 billion pre-tax gain in Q1 this year, and Q2 recorded a gain of as much as $53.4 billion. These paper profits directly boosted Amazon's earnings per share and the overall earnings of the S&P 500. According to analysis, the S&P 500's unadjusted earnings growth in Q2 was 50%, but excluding the equity valuation gains from Alphabet's and Amazon's AI investments, it was only 32%. The reflexivity of this loop has reached the point of distorting the entire index.

Reflexivity was proposed by George Soros and refers to the mutual influence and feedback loop between market participants' perceptions and reality. In AI investment, tech giants' equity investments in startups generate paper gains as the latter's valuations rise, boosting the former's profits and share prices; rising share prices in turn enhance their investment capacity, further pushing up startup valuations. This loop amplifies prosperity in upcycles and may accelerate reversals in downcycles. The S&P 500's earnings growth being inflated by AI equity valuation gains from Alphabet and Amazon is a typical example of reflexivity distorting the index.

Consensus and Divergence Under Multi-Source Cross-Verification

Combining information from multiple sources including the Bilibili creator, the Financial Times, and Ecosios, several points of consensus emerge: Anthropic's valuation growth far outpaces its revenue growth; the difference between gross and net bases significantly affects like-for-like comparisons; and the shareholder status of channel partners creates complex financial affiliations. Divergence lies in judgments about the authenticity of growth—some argue there were one-off factors in Q2's high growth, while others believe 45% QoQ growth is still healthy. Unique insights from single sources include: Anthropic told several investors this summer that annualized revenue would exceed $100 billion, and before the IPO it disclosed that quarterly revenue grew nearly 50% to $16.8 billion.

Outcome Scenarios: $2 Trillion or $1.4 Trillion?

Overall, the most likely outcome for Anthropic is a valuation cut from $2 trillion to $1.4–1.7 trillion, a discount of 15% to 30%. Even so, that price still represents about a 50% premium for Series H investors, and underwriters can still make it work. A discount does not mean Anthropic is not a good company—a good company and a good price are two different questions. The real test is that with rigid spending already locked in and revenue still highly elastic, the $2 trillion math requires more than a growth story; it requires financial discipline that can withstand the cycle.

One-line Summary

Anthropic's valuation jumped 2.5x in four months to a $2 trillion IPO target, but slowing growth, accounting-basis differences, and rigid spending reveal financial concerns.

Paragraph Summary

Anthropic's valuation soared from $380 billion in February to $965 billion in May, with a $2 trillion IPO target corresponding to about 10x forward P/S on 2028 revenue expectations. However, QoQ revenue growth plunged from 147% to 45%, a clear slowdown. Gross vs. net accounting can create a 30%–40% revenue gap, and the 85% gross margin excludes ongoing training costs. Amazon and Google are both channels and shareholders, and equity gains distort S&P 500 earnings growth. Rigid cloud commitments of about $518.1 billion, 80% irrevocable, create a maturity mismatch with elastic revenue. The most likely outcome is a 15%–30% valuation cut to $1.4–1.7 trillion.