🧠 Takeaways:
Anthropic just pulled off the biggest IPO on record. What just happened?
They made it: gross margin went from -95% to 85.5%, and FCF of $5.6B in 9M 2026
They'll have to raise prices: 20% to 50% a year for the next 2 to 3 years
Too big to fail: Amazon and Google need these labs to win
Let's Examine This Biz
Note: As always, none of what follows is legal, tax, investing, financial, or any other sort of advice. And I was never here 😉.
Anthropic (the LLM with a 10% chance of ending humanity, but is really a pretty good autocomplete) made it. We can debate the AI doomerism, the value of software and the rest of the AI circus.
Anthropic, scaling faster than any biz in recorded history, just went public at $1.6T fully diluted and stuck the landing in the first week of the biggest IPO ever.
Today we're making the case for why Anthropic is too big to fail.
Financial Summary

Stock price: $88.91 (+7% vs IPO)
Market Cap: ~$1.6T (fully diluted)
L5 Performance: N/A
P/E Ratio: N/A

FY 2025 Financial Statements (YoY Comparison)
Rev: $4.6B (+1,088%) 🤯🤯
Gross Profit: $1.8B (vs a $366M loss) 🤩
OPEX: $9.9B (+278%) 😬
Net Income*: -$7.7B (-166%) 😰
FCF: -$2.8B (+52%) 🙂
*Adjusted, one time only. Excludes $34.3B of non-cash marks on convertible notes that go up when Anthropic's valuation goes up. GAAP net loss was $42.0B (vs $8.3B in FY24). The notes converted at the IPO.

Competitive Benchmarking
(GAAP, FY2025 each. Segment figures are segment operating income.)
Metric | Anthropic | OpenAI (1) | SpaceX AI | Google Cloud |
|---|---|---|---|---|
Rev | $4.6B | $13.1B | $3.2B | $58.7B |
Rev growth | +1,088% | +253% | +22% | +36% |
Gross margin | 39.5% | 42.6% | n/d | n/d |
OPEX % of rev | 215% | ~203% | n/d | n/d |
Operating margin | -176% | -160% | -199% | +24% |
FCF margin | -60% | ~-61% | n/a | n/a |
Model | Tokens + subs | Tokens + subs | Compute + Grok | Cloud + Gemini |
(1) OpenAI figures are leaked, unaudited FY2025 financials (Quartz, Jul 3, 2026; Techstrong.ai, Jun 17, 2026). Sources: Anthropic S-1; SpaceX 424B4 and Q3 2026 10-Q; Alphabet FY2025 10-K.
OpenAI is the only real peer and its numbers are leaked, not audited. Google Cloud is what profit looks like at scale.

TLDR Analysis: The Ferrari actually runs
Rev 11x to $4.6B. Then $33.2B in 9 months of 2026
Gross margin 85.5% in 9M 2026
OPEX still 215% of rev in FY25
Real loss of $7.7B
FCF of $5.6B in 9M 2026
For a company losing so much money, they print cash now. The aggressive revenue ramp is scaling fast enough to outpace their costs. Flipping a deeply unprofitable biz into a barely profitable/unprofitable one.


Let's TLDR This Biz
Founded:
2021, when Dario + Daniela Amodei and 5 other co-founders left OpenAI
A research lab first. Launched Claude to the public in Q1 2023
Aha Moment:
2024. The Claude 3 family (Opus, Sonnet, Haiku) opened up real enterprise use cases
Feb 2025. Claude Code launches as a research preview. It plans, writes, tests and debugs code
Insight: companies pay for work done, not for chat
Growth:
Enterprise API sold through AWS, Google Cloud and Azure. ~47% of FY25 rev ran through cloud marketplaces
Claude Code as the wedge. Menlo puts Anthropic at 40% of enterprise LLM API spend, 54% in coding
Claude Cowork (Jan 2026) takes the same agent to every knowledge worker
Model:
83% consumption (tokens), 17% subscriptions in FY25
Rents almost all its compute
Its landlords (Google, Amazon) are also its investors, resellers and competitors
Where We Are Now:
IPO Nov 18, 2026 at $83. Raised $81.3B
Opened at $111.50 on day 1. Now $88.91, +7% vs IPO
Let's Figure out $1.6T
Quick nerd note: the headlines say $1.2T. That's "basic," which isn't completely accurate. It only counts the 13.2B shares that trade today.
"Fully diluted" adds every stock option and RSU employees and early investors are sitting on (another ~4.3B shares) that turns into real stock soon. If you're buying the whole biz, you're paying for those too. $88.91 x 17.6B shares = $1.56T.
The overhang from shares that haven't vested yet is where the difference comes from.
Step | Number | How |
|---|---|---|
Fully diluted market cap | $1.56T | $88.91 x ~17.56B shares |
Less cash after the IPO | -$232B | pro forma, 424B4 |
What you pay for the biz | $1.33T | market cap minus cash |
Worth in 2031 at 11.8%/yr | $2.32T | our hurdle rate, 5 yrs |
2031 net income needed | $93B | at 25x earnings |
2031 revenue needed | $310B | at a 30% net margin |
vs Q3 2026 run rate | $68B | ~36%/yr for 5 yrs |
vs our base case 2031 | $264B rev / ~$46B NI | half the profit needed |
To be worth $1.6T, Anthropic needs ~$310B of rev by 2031. That's 5x Google Cloud today.
And the share count keeps growing. Anthropic already gave 5% of its fully diluted shares to its own Public Benefit Fund. The board can hand it up to 2% more every time the stock rises 50% from the last milestone.

Let's Understand This IPO
Here are the 3 sales pitches you need to believe.
1) They made it
Despite everyone's worry, the doom and gloom, the AI doomerism, Anthropic made it.
The revenue showed up. The profit showed up. They figured out a model for AI.
They're making insane investments in the future. But right now they're hitting + surpassing everyone's expectations. At this point they're resetting them.
Look at the revenue. Look at the gross margin. Once you factor out the convertible note marks, the numbers are enormous.

Gross margin went from -95% in FY24 -> 85.5% in the first 9 months of 2026.
FCF went from a $5.7B burn -> $5.6B of cash coming in. AND they're sitting on a war chest of $200B+ now.
Q2 2026 even printed $974M of GAAP operating income. Q3 slipped back to a $567M loss, and Q4 eats a $14.7B stock comp hit when the IPO RSUs vest.
These are temporary, 1-time events of going public. It's early, but the model has had its first success moment.
Takeaway: The biz model works.
2) They'll have to raise prices
Anthropic has clearly built the Ferrari of AI tools for B2B bizs. The next move they have to make is to raise prices.
Cheap Chinese models are taking out the bottom of the market. Even the S-1 says open-weight models "could exert pricing pressure" on workloads that "prioritize cost efficiency over intelligence."
And I'd bet they're losing money on all the month-to-month plans. Really, they're making all their profit dollars off token usage. 83% of FY25 rev was tokens. Subs were 17%.

They don't reveal the cost breakdown by plan, but 1 insight is painfully obvious: monthly plans have to go up.
Expect them to raise pricing by 20%-50% every year for the next 2 to 3 years to improve their profit margins.
They'll be able to make it work because, let's be honest, who isn't getting $50, $200, or $500/user/mo worth of value out of Anthropic right now?
Takeaway: Monthly plans go up 20% to 50% a year.
3) Too big to fail
The last takeaway, the one no one is talking about but we need to acknowledge: Anthropic is too big to fail within the tech ecosystem.
To a certain extent, the future compute commits everyone is freaking out about don't matter.
Anthropic is creating too much value at too many different parts of the stack. Anthropic's stock markup makes up 40% of Amazon's last year PROFITS.
Not the revenue Amazon is generating from Anthropic as a customer. Or the revenue it's making from selling Anthropic to customers. 40% of Amazon's profits are coming from its stake in Anthropic.
It's a similar model with all the other providers.

I'm not predicting doom, but let's say something did happen to Anthropic. They're such a large customer they can make a call and work out a deal. They're tied too closely together.
And even Google and Amazon will have to accommodate their cash cows.
This is truly the ideal situation for a new biz. To be so successful your largest vendors are so tied to your success and have bought in so much they have no choice but to play ball.
Takeaway: The market will force these bizs to be successful.
Final Thought
It's more important to catch lightning in a bottle than to be disciplined. I can't tell you how many times I've seen this across successful breakout brands, and Anthropic is the epitome of this.
We can talk all about financial discipline, cost spending, and everything else about how a biz should operate.
But when you just catch a moment in the market like this (where everybody wants your product, where you're super hot, and you can just go vertical), the only play is to land grab as aggressively as possible.
This only happens to 1 out of a million companies. It's so rare it completely perverts how business building is thought about. They are the LeBron James of business.
But when it does, everyone is so engulfed in it that they try to become them.

