1/Anthropic’s S-1 is out.
2/ Revenue, quarter by quarter ($M)
Q1'25: 376
Q2'25: 787
Q3'25: 1,382
Q4'25: 2,040
Q1'26: 4,731
Q2'26: 11,555
Q2'26 revenue was up 1,368% YoY. Compute & infra costs were up 300% over the same period.
Q2'26 was the first quarter with positive GAAP operating income: $974M.
3/ How “Run-Rate Revenue” is defined
Trailing-28-day consumption fees × 13
+ monthly subscription fees × 12
Three caveats from the S-1:
-It includes customers later found to be violating the usage policy
-It is seasonal: usage dips in summer and winter holidays
4/ Revenue is recognized gross
When Claude is sold through the AWS, GCP or Azure marketplaces, Anthropic is the principal. It books the full amount as revenue, and the cloud provider’s fee goes into Sales, marketing & partnerships.
The marketplace share of revenue has climbed fast:
11% (2023) → 32% (2024) → 47% (2025) → 48% (H1'26)
Adjust for this before comparing multiples with companies that report net.
5/ Distillation
The S-1 puts a number on it:
“we estimate that revenue from such customers represented approximately 5% to 7% of our revenue for the nine months ended September 30, 2026”
These are customers violating the usage policy or commercial terms. Applied to ~$33B of 9M revenue, that’s roughly $1.7–2.3B.
Enforcement was stepped up in Aug–Sep, and the company expects the share to decline. The S-1 says the estimate “requires judgment and may change.”
6/ Visibility is low
RPO at 6/30/26: $10.8B
~75% of that is recognized in the next 12 months, so cRPO ≈ $8.1B
Deferred revenue: $1.65B
Q2 annualized revenue is ~$46B, so cRPO covers less than 20% of it.
80% of revenue is consumption-based, and next year’s revenue depends almost entirely on usage.
7/ The compute bill
Disclosed commitments:
Broadcom TPU leases: ~$161B
SpaceXAI (NVIDIA GPUs): up to $121B, ~$3.1B a month from Dec’26
Google: ≥$115B, plus $125–150B more of TPU capacity in 2028–31
Amazon: ≥$110B
Microsoft: ≥$45B
Nscale: ~$45B
Lambda: ~$41B
AMD: >$20B
That adds up to roughly $780–810B.
8/ The mismatch to watch
Revenue is variable and consumption-driven. Compute costs are fixed, multi-year and take-or-pay.
On top of that, the Broadcom leases, the credit agreement and the convertible notes include cross-default provisions.
This is the single most important risk in the filing.
9/ Demand is real
NRR ~900% (ex-Pro/Max)
Customers paying more than $100K / $1M / $10M a year: 11,900+ / 1,900+ / ~250
No single customer was >10% of 9M’26 revenue
Top 5 customers ≈ 25% of 9M revenue
Q3: 79% of revenue came from outside the top 5, up from 64% in Q1
Claude Enterprise users who also use Claude Code spend ~20x more than those who don’t
10/ Dilution
$77B of convertible capacity: Google $20B, Amazon $15B, Broadcom $42B, settled in shares at a trailing VWAP
Google can buy up to $10B more stock at the IPO price
A Public Benefit Fund already holds 5% of fully diluted shares, and can get up to +2% more for every 50% rise in the share price
Daave “3 and 30” Tepper retweeted
Pretty crazy that on a day like today, the S&P 500 nearly triggered a deGraaf Breadth Thrust for the first time since May 2025. 👀
The trigger is 55% of S&P 500 stocks closing at new 20-day highs.
Today, we hit 50.4%.
So close.
@RenMacLLC @granthawkridge @WillieDelwiche
Daave “3 and 30” Tepper retweeted
Potentially some incredible cap structure arb setups made possible if “liquid enough” gpu derivatives emerge
I’d setup
-Long gpu
-short $crwv equity
-long crwv converts (synthetic put)
Ratio tbd
Also why hasn’t $hype @HyperliquidX announced anything in the gpu space?
The mighty A100 fleet are mission-capable from 2020 through 2029. NVIDIA computing is more than chips. CUDA gives developers and NVIDIA engineers a common platform to continually upgrade Ampere, Hopper and Blackwell throughout their useful lives.
CUDA makes NVIDIA computing versatile. Versatility makes it fungible. Fungibility drives utilization and extends durability, making NVIDIA compute a productive asset: rentable, durable and financeable.
Daave “3 and 30” Tepper retweeted
The S&P 500 is now in bear-market territory..
..the valuation, that is. P/E has contracted from 23.2x a year ago to 18.7x today.. .that's just about a 20% de-rating @dualityresearch
Multiple is the cheapest since Liberation Day April 2025
Daave “3 and 30” Tepper retweeted
For now, I am giving this cyclical (and secular) bull the benefit of the doubt, assuming (hoping?) that the stellar earnings momentum is not about to mean-revert. I suspect that the “melt-up” phase of this bull (since April 2025) may have ended in early June, but perhaps that is just a pause in the momentum during the seasonally weakest timeframe (July-October). I’ll stick to my 60/20/20 playbook, which has held up quite well, even though it diversifies away from the mega cap theme.
Daave “3 and 30” Tepper retweeted
Volatility on misunderstood newsflow seems endemic to AI:
Deepseek.
Liberation Day.
Kimi.
SALP.
Gross to net confusion for Anthropic.
OpenAI taking share from Anthropic.
Gross to net confusion for OpenAI.
A mistranslation of Hock Tan’s comments yesterday.
Daave “3 and 30” Tepper retweeted
Two wolves inside the market present are fighting each other
AI - this once a century technological revolution - about to go into recursive self improvement and radically change the structure of knowledge economy forever leading to a productivity boom….
Juxtaposed with
bond yields - step function change higher relative to a large global debt stock we haven’t seen this combo in the modern global economy.
Equities discount the residual value of everything.
“Everything” has a wide cone of outcomes due to these two wolves.
be prepared for anything, which is to say, be diversified
Daave “3 and 30” Tepper retweeted
Replying to @arakharazian
Hmm. Let’s think for a second.
Why are the frontier models cutting price? Might it possibly, just possibly, be a result of the fact that they are losing share to open-weight models?
Given positive elasticity, their price cuts are clearly good for AI infra demand.
Open weight models are positive for AI infra demand, exhibit 1002.
And obviously the semiconductor companies and cloud companies would fund open-weight training if necessary. Jensen already is!
OpenAI and Anthropic have locked in years of compute at below market prices. Other companies would happily assume these commitments at *higher* prices.
Also I’ve always wondered if “missing the forest for the trees” is an idiom, a proverb or an aphorism. Do you have a nuanced view on this?
I am going to ask everyone this question from now on when appropriate.
Daave “3 and 30” Tepper retweeted
Hewlett Packard Enterprise ($HPE) targets low-to-high 50s% annual revenue growth in data center networking through F2029, outpacing an addressable market projected to expand at a 44% CAGR. This marks a sharp acceleration from earlier 15% market growth estimates, driven by hyperscaler demand for liquid-cooled 800G and 1.6T switching architectures across AI clusters. Underpinning this expansion, the group secured a $1.2 bn AMD Helios rack order from Vultr for scale-up switching and expanded a gigawatt-scale infrastructure contract with Oracle. Management expects Helios to generate over $1 bn in revenue across the next two years, though realization speed hinges on component delivery schedules and data center deployment timetables. #HPE #Networking Report date: 2026-10-05 · GS
Daave “3 and 30” Tepper retweeted
It is becoming obvious that AI is impacting the labor market.
Going back to mid-2025:
-Finance + Tech industries -246k jobs.
-All other industries +812k jobs.
This dynamic will most likely intensify next year and expand into other industries.
Daave “3 and 30” Tepper retweeted
$HPE SECURES $1.2B VULTR ORDER FOR $AMD HELIOS AI RACKS; RAISES NETWORKING OUTLOOK
HPE secured a $1.2B order from Vultr for AMD Helios AI Rack systems, its first order for the new platform.
HPE raised FY27 Networking revenue growth guidance to high-teens to low-20s %, with a high-teens CAGR now expected through FY29.
By segment through FY29:
Data Center Networking: low-to-high 50s % CAGR
Routing: low-to-high 20s %
Campus & Branch: high single digits
Security: high single digits
HPE also raised expected Juniper cost synergies to $800M annually by FY28, from at least $600M previously.
Daave “3 and 30” Tepper retweeted
cerebras is schrodinger's cat
$CBRS
Daave “3 and 30” Tepper retweeted
Replying to @jake_ayes @insane_analyst
yeah kind of weird and intentionally misleading.
cerebras has some of the fraudiest marketing. reminds me of Nikola.
Daave “3 and 30” Tepper retweeted
cerebras may be the most hf-held trashco with de mininus terminal value regardless of success probability
Daave “3 and 30” Tepper retweeted
Replying to @The_AI_Investor
cerebras installing chips for seed equity stakes not cash
Sad.
Daave “3 and 30” Tepper retweeted
Bernstein on CXMT: “We believe a new fab of max capacity at 100k wpm is being built in Shanghai now, & is due to begin production early next year. There is no mention of this fab in any company filings”
Models capex rising 70% YoY through ‘28 to $35B
CXMT capex is $ACMR revenue




























