Anthropic's revenue growth this year has been pretty astonishing https://
simonwillison.net/2026/May/29/an
thropic/
… – and the latest leaked run-rate number is $65 billion for July
ECONOMY
-
Reported Anthropic Revenue Run Rate Reaches $65 Billion in July
By
–
-
Anthropic Expects Continued Profitability Beyond a Timing Benefit
By
–
Anthropic have since said they expect Q3 to be profitable as well, which wouldn't benefit from the SpaceX timing
-
Higher Terra and Luna Demand May Offset Discounted Inference Prices
By
–
If you estimate the revenue, before the cut it was $2m, at discounted prices that would have been more like $0.5m. But because the demand went up 5.6× for Terra and 13.8× for Luna, total revenue was $4m for the discounted period – so double of that before the discount
-
Cheaper AI Models Can Shift Error-Correction Work Onto Readers
By
–
It might soon be disrespectful to use a weaker model for human-facing content: “you saved 6 cents to make me read through error-filled & badly written AI slop? At least send me high quality and low-error slop that doesn’t waste my time.” (This is how I feel about AI X comments)
-
Successor Demo Finds Supplier Spending Discrepancy With Bounded Authority
By
–
AI should not merely answer. It should earn the right to act.
— MONTREAL.AI (@Montreal_AI) 29 août 2026
Watch SUCCESSOR Ω uncover an $80K supplier-spend discrepancy, learn by falsification, pass independent proof, earn Mission Alpha—and operate only within bounded authority.
Superintelligence, made accountable.… pic.twitter.com/Xntud0gAgoAI should not merely answer. It should earn the right to act. Watch SUCCESSOR Ω uncover an $80K supplier-spend discrepancy, learn by falsification, pass independent proof, earn Mission Alpha—and operate only within bounded authority. Superintelligence, made accountable.
-
Changing Reasoning System Prompts Can Invalidate Prompt Caches
By
–
juice is a system prompt so your cache is busted and any savings you think you get are gone.
-
Orbital AI Compute Faces Cooling, Radiation and Data Transfer Constraints
By
–
Putting AI compute in space may indeed be possible and the satellite may be the easy part. Cooling, radiation, data transfer, launch economics and maintaining thousands of compute satellites are the real test. Fascinating idea. Now physics and economics have to agree.
-
Perplexity Leads Reported Search Index With Improved Cost-Quality Tradeoff
By
–
The Perplexity Search API takes the top three spots on the Artificial Analysis Search Index. The medium setting scored five points above the previous leaders, and extended the quality-cost Pareto frontier at about $0.091 per task.
-
Grok Bot Can Make Approved Purchases Through Stripe Link
By
–
Grok Bot can now be linked to a Link account so it can buy things for you online.
— 🚨 AI News | TestingCatalog (@testingcatalog) 28 août 2026
> All transactions still require approval.
> Only available in the US 👀 https://t.co/9XhY6iFOkw pic.twitter.com/YqHyfrHueeGrok Bot can now be linked to a Link account so it can buy things for you online. > All transactions still require approval. > Only available in the US
-
Replit Reports Lower Coding Costs From Automatic Model Routing
By
–
1) Intelligent Model Routing. Spend your time on your idea, not on picking models. Replit now picks the best model for each task, automatically. Just select Auto in the product. Our testing shows you'll get the same quality output at 65% lower cost than the previous version of