- Fiscal.ai
- Posts
- đź—ž This Is the Fastest Growing Company in History
đź—ž This Is the Fastest Growing Company in History
Gross Profit: $0 → $40 billion in 5 years
Happy Sunday! đź‘‹
This week, we got a glimpse at arguably the most anticipated IPO of all-time: Anthropic.
Let’s dive in!
The Fastest Growing Company of All-Time
According to a quick Fiscal.ai screener, there are 79 companies in the world that currently earn more than $40 billion in annual gross profits.
With the exception of Meta and Alphabet, it took every other company on that list 20+ years to achieve $40B in gross profits.
For Anthropic, it took 5 years.
Anthropic filed their S-1 with the SEC this week, officially declaring their intent to go public.
While most people probably have at least some vague idea about what Anthropic does, this is the first time the public has been given a clear look at the underlying economics of the frontier models, so it warrants some dissection.
Business Model
Anthropic splits its revenue into two sources: Subscriptions (20% of revenue) and Consumption (80% of revenue).
Subscriptions are relatively straightforward. Customers pay a flat-rate for Claude Pro or Max, and in exchange, they are given dynamic usage limits. The “dynamic” part has become a bit of a controversy, but we can sidebar that for another time.
Consumption-based pricing, on the other hand, operates like a metered utility. It’s essentially an open-ended bill predicated on a customer’s overall token usage.
Importantly, consumption-based pricing does not operate on a traditional cost-plus model. Instead, Anthropic sets a fixed price per token for each model and as the cost of compute comes down over time (through hardware efficiencies and inference optimizations), Anthropic’s per-token profits rise gross margins rise.
This diminishing cost of compute is the primary reason why Anthropic’s gross margin has gone from below zero to 80%+ in less than two years.
Prior to the S-1, there were lots of rumors about just how unprofitable the frontier labs were. Suffice to say, this filing put most of those rumors to rest.
While, yes, technically Anthropic is extremely unprofitable on a GAAP accounting basis ($81B net loss in the first half of 2026), virtually the entirety of that loss is attributable to non-cash charges.
To sum up the somewhat tricky accounting, over the years, Anthropic chose to raise money with instruments that could result in future dilution instead of selling shares up front. They primarily did this through convertible bonds, and every time Anthropic’s valuation rises, so too does the value of those convertible bonds. Under GAAP accounting, they are required to mark the change in fair value of those securities as an expense.
**Sidenote** Amazon owns $6.2 billion worth of Anthropic’s convertible debt, which is equivalent to 10%-12% ownership.
If you strip out these non-cash charges, Anthropic reported $974 million in operating income for the 2nd quarter.
Insane Stats
Before diving into some of the interesting governance items from this filing, it’s worth taking a sec to simply admire the outrageous growth Anthropic has generated.
For starters, Anthropic has generated outrageous enterprise adoption. Here’s a timeline of the number of companies that spend more than $100,000 annually with Anthropic:
Dec 2023: ~20 customers
Dec 2024: ~200 customers
Dec 2025: ~1,500 customers
Sep 2026: 11,900+ customers
And those customers spend more with Anthropic every year…
“We believe that enterprise customers achieve a compelling return on investment from Claude, as agentic workflows can drive both top-line growth and efficiency. This enables us to expand quickly with customers, as demonstrated by our Net Revenue Retention of approximately 900% as of September 30, 2026.”
Not So Fast
Before investors go plowing into this IPO, it’s worth raising a few points of caution.
Number one, Anthropic is technically organized as a Public Benefit Corporation (PBC). A PBC is legally chartered to balance two separate mandates: 1) satisfying the interests of their shareholders and 2) pursuing a specific public benefit/mission.
In Anthropic’s case, their public mission is “the responsible development and maintenance of advanced AI for the long-term benefit of humanity”. What this means in practice is that they may make decisions that are not intended to maximize shareholder value (delaying a model launch for example) , if they feel it does not benefit the public good. Additionally, Anthropic is also giving 5% of its equity (with the option for more later on) to the “Anthropic Public Benefit Fund”.
Another item worth mentioning is that in conjunction with this IPO, Anthropic will be issuing a single share of Class F common stock to the founders (Dario Amodei and Daniela Amodei) that gives them 50.1% of the total voting power. In other words, the founders have complete control.
Lastly, and this will perhaps be the biggest near-term risk, Anthropic is developing its own data center infrastructure. From page 74:
“To help reduce our dependence on CSPs (Cloud Service Providers), address constraints on available compute, and secure access to compute at the scale necessary for frontier model development, we are pursuing initiatives to develop our own dedicated on-premise data center infrastructure.”
One nice benefit of Anthropic’s business model thus far is that they’ve been able to buy compute capacity from other data center providers (AWS, Google Cloud, Azure, SpaceX AI) instead of developing their own. This has been a big boost to their cash flow.
However, since they pretty much compete with all of their major cloud vendors on the AI front, they are recognizing the need to build their own infrastructure. As we’ve seen with the other big tech companies, data center buildouts are not cheap, and this could end up being pretty consequential for Anthropic’s cash flow in the near-term.
That’s all for this week.
If you have any questions about Fiscal.ai or this newsletter, simply reply to this email!



