Anthropic filed its S-1 on October 26, 2026. The filing immediately became the center of intense discussion because it is the first detailed public look at the financials of a leading frontier AI company. The document is long and dense. Both optimists and skeptics are already selectively quoting it. A more useful approach is to separate the hard numbers from the narratives and then test one coherent valuation story against those numbers.
Key Numbers from the Filing
Revenue growth
First-half 2026 revenue reached $16.3 billion — roughly fourteen times the level of the same period a year earlier. Growth remains rapid, yet the rate of acceleration slowed in the third quarter amid rising competition, temporary compute constraints, and some enterprise migration toward lower-cost open-source alternatives.
Customer concentration is still material: the top five customers account for about 25% of revenue. At the same time, the base of larger enterprise users has expanded sharply — roughly 1,900 customers now spend more than $1 million a year.
Profitability path
The reported net loss of $80.7 billion for the first half includes approximately $72 billion in non-cash accounting charges related to convertible notes. Adjusted operating losses remain significant, but the direction is clearly improving. Compute costs as a percentage of revenue have declined substantially (from well above 200% toward the mid-40s on a recent quarterly basis). The company also reported an operating profit in its most recent quarter — an important milestone, even if it is still early.
Compute commitments
Anthropic has locked in hundreds of billions of dollars in future compute obligations. These contracts secure the capacity needed for continued model development and scaling, but they also create substantial fixed-cost leverage and ongoing reinvestment requirements. Chip prices may fall over time; data-center construction and power costs are less certain to decline.
Balance-sheet buffer
Cash and short-term investments already exceed $100 billion before IPO proceeds. A successful offering would further strengthen that buffer, giving the company several years of runway to meet contractual commitments even if growth moderates.
One Explicit Valuation Framework
Valuing a company at this stage requires an explicit set of assumptions about the future rather than simple extrapolation of recent results. One internally consistent base case rests on the following story:
The overall market for AI products and services expands into the multi-trillion-dollar range, with Anthropic capturing a meaningful share of the higher-value, frontier-model segment.
Revenue approaches $1 trillion by 2036.
Long-term pre-tax operating margin settles around 27.5% after a moderate regulatory impact (starting from a higher structural margin of roughly 32.5%).
Reinvestment intensity averages about $1 of capital for every $3 of incremental revenue.
Cost of capital is set at 9.74%.
Probability of severe distress is estimated at 10%, with a 50% loss of value in that scenario.
Under these inputs, equity value is approximately $1.475 trillion. That figure sits at the lower end of the $1.5–2 trillion range that has been widely discussed around the offering.
The range of plausible outcomes is wide. A materially smaller end market, thinner margins, or heavier ongoing capital needs can produce much lower values — in extreme cases, values that imply the company does not survive as a going concern. More optimistic assumptions on market size, margin trajectory, and capital efficiency can push the equity value well above $2 trillion. The sensitivity itself is the main point: small changes in the core story produce large differences in value.
Practical Implications
The S-1 confirms three things clearly:
Recent growth has been extraordinary.
Unit economics are improving.
Future compute obligations are very large.
It does not resolve the bigger uncertainties — how large the premium AI market will ultimately become, how sticky enterprise demand proves through an economic cycle, or how expensive it will remain to stay at the frontier of model performance.
For readers evaluating the IPO, the useful discipline is to state an explicit story about market size, margins, reinvestment, and risk, then check whether the resulting value is consistent with the price being asked. Treating the filing as definitive proof of either inevitable dominance or inevitable collapse overstates what the document can show. It is a detailed snapshot of a still-young company, not a finished forecast.
The numbers are now public. The harder work of deciding what they imply remains with each investor.
No letters yet — pray write the first.