【TMGM Financial Recap】Anthropic Sprints For A $2 Trillion IPO! 30 Trillion TAM Surpasses Spacex

The company aims for a valuation of about $2 trillion, plans to raise up to $100 billion, and is expected to go public as early as September or early October. If successful, it will comprehensively surpass SpaceX's IPO record set in June this year in valuation, fundraising scale, and TAM.

Propping up this narrative is Anthropic's rocket-like financial performance. Internal documents show that the company's single-quarter revenue for Q2 2026 reached $11.6 billion, about 14 times the $787 million in the same period last year, and for the first time recorded an adjusted operating profit of $559 million, marking the first time a leading large model company achieved single-quarter operating profitability. Annualized revenue (ARR) surged from $9 billion at the end of 2025 to $65 billion in July, an increase of more than sixfold over eight months.

However, alongside this explosive number of figures, another set of unsettling data has emerged that has unsettled investors. Fintech company Ramp's analysis of consumption data from 70,000 companies shows that Anthropic's flagship model Fable 5, launched in June this year, accounted for only 11.4% of the company's total revenue in just over two months after its launch. The lower-priced older model, Opus 5, overtook enterprise spending on the Fable 5 in less than a month after its release.

How was 30 trillion TAM calculated?

TAM (Potential Market Size) refers to the theoretical annual revenue ceiling a company can achieve if it achieves 100% market share in the relevant market. Anthropic's way of quantifying TAM is not to count existing software or hardware sales, but rather to include the full economic value of "all future global work that can be replaced or completed by AI models" into the calculation scope. This definition directly transforms TAM from a "market share" issue into a "labor substitution" problem — any work AI can do counts as its market.

The impact of this narrative requires a frame of reference to grasp. In its IPO filing in May this year, SpaceX defined TAM as "the largest executable market in human history," with $26.5 trillion coming from AI-related opportunities. Aswath Damodaran, a finance professor at New York University and known as the "Godfather of Valuation," warned at the time that SpaceX's TAM estimate for AI "has reached reasonable boundaries and is beginning to break through." For reference, FactSet data shows that 191 tech companies in the S&P 1500 combined revenue of only $2.4 trillion last year. Anthropic's $30 trillion TAM is equivalent to 12.5 times the combined revenue of these companies, and slightly below the full-year US GDP.

The issue is not whether Anthropic's algorithm is "correct," but that TAM is essentially a narrative tool rather than financial forecasting, but whether the market is willing to pay for narrative. SpaceX's post-IPO trajectory has already provided preliminary answers.

From 9 billion to 65 billion: the underlying support of rocket-like growth

Anthropic's growth rate is truly unprecedented. Annualized revenue will grow from $9 billion at the end of 2025, to $14 billion in February 2026, $30 billion in April, $47 billion in May, and finally surpassing $65 billion in July. Q2 revenue was $11.6 billion, up over 140% quarter-on-quarter and about 14 times year-on-year. During the same period, competitor OpenAI's Q2 revenue was $6.7 billion, up only 18% quarter-on-quarter, and it remained operating at a loss.

The divergence between the two companies stems from their different business model paths. Anthropic is betting on enterprise and programming scenarios, achieving high-value monetization through API calls and tools like Claude CodeEnterprise customers are willing to pay a premium for stability, data security, and automated workflows, because a model that helps engineers write code and fix bugs saves high-paid engineers time. OpenAI, on the other hand, relies more on consumer ChatGPT subscriptions, with low average order value and limited growth potential by its user base. By April 2026, Anthropic's annualized revenue had surpassed OpenAI's, and the gap has continued to widen since.

More importantly, Anthropic's adjusted operating profit of $559 million in Q2 is far more significant than the numbers themselves. Large model companies have long been accused of "burning money for growth," with inference costs, training costs, and computing power procurement eating into every bit of revenue. Anthropic can achieve profitability alongside explosive revenue growth, proving that in certain high-value scenarios, large model revenue can exceed costs. This is the underlying logic supporting a $2 trillion valuation: it's not about telling stories, but about having already established a business model.

Flagship sales fail: Fable 5 cools off

But a successful business model doesn't mean there are no cracks. Fable 5 is Anthropic's flagship model released in June this year, with input and output prices of $10 and $50 per million tokens, respectively. Ramp's benchmark shows the Fable 5 scored 70.5 points, with a single task cost of about $17.32; while the older Opus 5, priced at only half the price, scored 70.0 points and cost only $8.23 per task. A 0.5-point difference doesn't translate into double the price, which is exactly why corporate customers vote with their feet.

In contrast, OpenAI's flagship model GPT-5.6 (Sol) now accounts for 23% of total revenue and is priced significantly below Fable 5. Fable 5 was also forced to be taken down for several weeks due to U.S. government export orders, and the 30-day data retention policy has also deterred some compliance-sensitive customers.

The deeper meaning of this phenomenon is more important than the success or failure of individual models: AI commercialization is entering its second phase. Mccel partner Miles Clements points out that the era when customers only chased the most cutting-edge models is no longer sustainable. Enterprise clients are shifting from "chasing the strongest model" to "just having enough is enough"; as long as the cost-performance ratio is right, older or even low-priced models are sufficient. If this trend continues, Anthropic will face greater challenges in achieving its $190 billion to $200 billion revenue target by 2028, as its growth engine may need to shift from "selling flagships" to "selling volume," with the latter having far lower gross margins.

SpaceX's Lesson: How the Market Views After the Honeymoon Period

Anthropic's IPO narrative is strikingly similar to SpaceX's: massive TAMs, the largest valuation ever, and rocket-like growth. SpaceX's post-IPO performance has provided Anthropic with a mirror.

SpaceX went public on June 12 this year, raising $86 billion at $135 per share, with a valuation of $1.77 trillion. On its first day of listing, the stock price surged to a high of $225.64, with a market value approaching $3 trillion at one point. But the honeymoon lasted only four days. By early August, the stock price had dropped to $104.83, a retracement of more than 53% from its peak. As of the close on August 25, it was $137.95, barely returning near the issue price.

SpaceX's turning point came in its first earnings report. Q2 revenue was $7.814 billion, up 91.9% year-on-year—impressive figures, but the market is focused on another figure: capital expenditure was $18.369 billion, up 556% year-on-year, with 86% of that allocated to AI infrastructure. How much does it cost to achieve growth? This is the question the market began to ask after the honeymoon period ended. Once the narrative switches from "storytelling" to "settling accounts," the narrative's margin for error narrows dramatically.

The scrutiny Anthropic faces will only get harsher. SpaceX was at least profitable at the time of its IPO, while Anthropic had just achieved quarterly profitability. Its sustainability depends on whether enterprise clients continue to pay for high-priced modelsThis is precisely the premise that Fable 5's lukewarm performance is now shaking. What is even more concerning is that investment banks are using an "advanced valuation" strategy for Anthropic: bypassing current performance and pricing it directly based on forward revenue forecasts of $190 billion to $200 billion in 2028. This means that any revision to 2028 revenue expectations will be amplified by leverage into valuation shocks.

There has been a clear divergence in market pricing for Anthropic. The financing valuation completed in May reached $965 billion, surpassing OpenAI for the first time. The secondary market transaction valuation has soared to $1.5 trillion, and "almost no one is willing to sell." Optimists believe that based on $200 billion in 2028 revenue, a $2 trillion valuation corresponds to a price-to-sales ratio of only 10 times, which is not expensive for a company with growth exceeding 14 times. Nomura warned that Hong Kong hedge funds are optimistic about AI but hesitant to increase their holdings, and that Anthropic's IPO could trigger a wave of AI selling in October, raising concerns about trading crowding risks.