Anthropic’s ‘Good But Not Good Enough’ Quarter: Revenue Growth Slams on the Brakes
TREE NEWS reports: AI unicorn Anthropic delivered a third-quarter scorecard that looked impressive on the surface but landed with a thud among investors. At a closed-door pre-IPO investor day in San Francisco, the company disclosed Q3 revenue of $16.8 billion, up nearly 50% from $11.5 billion in the second quarter, alongside an “adjusted profit” and positive cash flow.
Markets were far from cheered. One research shop bluntly summed it up: the numbers are good, but not good enough. Another analyst was more final: not good enough — it’s over.
The Hidden Problem: A Growth Cliff
The concern lies in a dramatic deceleration. Anthropic’s annualized revenue run rate had already reached $65 billion by late July. Backing out the quarterly figure, implied monthly growth collapsed from roughly 33% compounded earlier in the year to just 3.4%–6.6% in August and September. If that trend holds, year-end annualized revenue lands near $89 billion — roughly $10 billion short of the widely expected “over $100 billion” target.
To hit its own goal, Anthropic would need Q4 monthly growth to re-accelerate to 10%–13%. That would have to happen precisely as token prices hit historic lows and open-source competition intensifies.
‘Adjusted Profit’ Comes With Asterisks
The profit claim deserves the most scrutiny — and the most footnotes. Full GAAP figures are expected around October 25 alongside a public S-1 filing. Confidential filings previously indicated a 2025 net loss of roughly $42 billion, including about $34 billion in financing-related accounting charges, more than $8 billion in operating losses, and at least $518 billion in future compute and infrastructure commitments over the next decade — about 80% of which is non-cancellable.
Revenue definitions are also contested. Anthropic uses a near-gross revenue approach that counts all customer spending through cloud partners and books partner revenue share as cost, flattering top-line figures relative to rivals using net revenue. That makes “adjusted” margins highly dependent on accounting treatment.
The Core Contradiction: Volume Up, Price Down
The structural root of the slowdown points to falling token prices. Anthropic’s token price index has hit record lows. Meanwhile, usage data shows AI consumption still growing rapidly — consistent with Jevons paradox logic — yet actual AI spending is flat at best, or even shrinking, as open-source models expand market share and drag down overall token prices.
This volume-price divergence is Anthropic’s central business-model contradiction: token prices are falling faster than the company can convert higher usage into revenue growth. If the August–September slowdown occurred while prices were still relatively firm, Q4 re-acceleration would require vastly higher usage at half the price — a fundamentally different commercial logic. Meta, Google, and Chinese open-weight models are flooding the frontier market at deep discounts, further compressing Anthropic’s pricing power.
What It Means for the AI Trade
- Equities: The AI narrative needs top labs to lift combined annualized run rates from roughly $100 billion in July to $1.8–2 trillion by year-end. Anthropic’s ~$89 billion plus OpenAI’s ~$70 billion target totals ~$159 billion — below the threshold, and that mixes gross and net revenue definitions.
- Sentiment: Markets voted first: the S&P 500 fell 1.11% and Nasdaq 100 dropped 2.02% the day before the disclosure, with AI-linked names sliding and some cloud stocks down over 12%.
- Valuation: At a $2 trillion IPO valuation, Anthropic trades near 27x September-end implied run rate and ~22x December trend — narrower than the ~47x historical revenue multiple calculated two weeks ago.
Key Takeaways for Investors
The $16.8 billion quarter, adjusted profit, and positive cash flow are enough to keep a November IPO on track. But the collapse in monthly growth from 33% to single digits — and the Q4 re-acceleration needed to reach the $100 billion year-end target — are the real questions markets must answer once the public S-1 drops. For the broader AI trade, the message is clear: good is no longer good enough.
