Anthropic's Compute Bill Is Eating Its Revenue Growth

Anthropic's Compute Bill Is Eating Its Revenue Growth

Anthropic's amended S-1 tells a story every AI builder should sit with: Q3 revenue hit $17 billion, up from $11.5 billion the prior quarter — genuinely explosive growth — but compute costs climbed to 58% of revenue, up from 48%, and the company swung to a $567 million operating loss from a $974 million profit [1][2]. The culprit was $5 billion in short-term compute commitments, which cost more per unit than the multi-year deals the company would rather be locked into [3]. Revenue per dollar of compute fell to 1.72x, and total compute commitments now exceed $800 billion.

This isn't a crisis so much as a tax on growing faster than your supply chain. Anthropic is buying capacity at spot-market prices because demand is outrunning its locked-in infrastructure — a good problem, but an expensive one.

The lesson for anyone building on top of frontier models: pricing today is not pricing in eighteen months, in either direction. Lock in what you can, and don't build margin assumptions on compute costs you don't control.

The October Jobs Report Didn't Find the AI Apocalypse

The US added 165,000 jobs in October against expectations of 65,000, and unemployment fell to 4.1% — numbers that directly undercut the white-collar-wipeout narrative that's dominated AI commentary all year [1]. Professional and business services added 70,000 jobs, including nearly 33,000 in professional/scientific/technical roles — the exact category supposedly most exposed to AI automation [2]. Healthcare added 42,000-55,000 more.

Economists quoted in the coverage framed AI as an engine of growth rather than a displacement mechanism this cycle, with AI-adjacent construction and manufacturing gains showing up alongside the knowledge-work numbers [3]. That doesn't mean displacement isn't coming for specific tasks and roles — it means the aggregate labor market hasn't cracked yet.

Treat this as one data point, not a verdict. But it's a useful corrective for anyone building AI roadmaps around imminent mass layoffs that haven't materialized.

A Custody Case Won With a Raspberry Pi and Grok

On November 8, a user posted the kind of story that makes AI tangible in a way benchmarks never do: they used Grok to help build "Chasing Your Tail," an open-source Raspberry Pi tool that logs nearby phones and Wi-Fi networks, and used it to catch a private investigator their ex-wife had hired to track them during a custody dispute [1]. The logs showed repeated surveillance at specific locations tied to identifiable networks — evidence that reportedly contributed to a 50/50 custody ruling [2].

Elon Musk amplified the story on November 9, calling out what "Grok Bot can do." Setting aside the promotional angle, the underlying mechanic is the real signal: a non-engineer used a conversational AI to scope, build, and deploy a functioning hardware-adjacent surveillance tool for a specific, high-stakes personal problem.

This is the orchestration era showing up outside the enterprise — individuals directing AI to build custom tools for problems too small or too personal for any software company to ever build for them.

Citi Bets Europe's Lower AI Exposure Is Now an Advantage

Citi strategists downgraded US stocks and global tech to neutral on November 9 while upgrading Europe ex-UK to overweight — and the reasoning is notable: Europe's lower concentration in AI stocks, usually framed as a weakness, is now being pitched as a hedge against US tech volatility [1]. Falling oil prices and Europe's diversified mix of industrials, banks, and defensives add to the case, and the region reportedly outperformed the US during prior AI-led selloffs [2].

For a region that's spent two years being told it's "behind" on AI, this is a strange kind of vindication — not because Europe caught up, but because it never over-concentrated in the first place.

X discussions tied to the move pointed to European players like Mistral and regional infrastructure bets as the actual upside case, not just defensive positioning. Worth watching whether capital reallocation turns into real investment in Nordic/EU AI infrastructure, or just a temporary flight from US volatility.