The very big caveat to the report that Anthropic is profitable for a second straight quarter
THE SO WHAT
Profitability headlines for frontier labs with major caveats highlight how dependent these businesses still are on accounting treatment, prepayments, and cloud barter — not just unit economics. When you evaluate model vendors, push past topline profitability claims and ask about gross margin by product, infra commitments, and how long current pricing is underwritten.
READ THE SOURCE
MORE FROM THE WIRE
Applied AIBRKZ secures $31m to scale AI-enabled building materials procurement
AI-native procurement in heavy materials is moving from pilot to balance sheet line item — $31 million into BRKZ’s pricing, fulfillment, and embedded finance stack is a bet that margin lives in coordination, not inventory. If you run construction or industrial supply, assume your counterparties will soon have better price discovery and credit intelligence than your ERP exposes today.
Applied AIClosing the gap between AI investment and impact: the rise of Open Data Infrastructure
The bottleneck on AI ROI is shifting from models to whether your data is actually accessible, labeled, and governed for reuse — “open data infrastructure” is just the new name for getting your pipes and contracts in order. If your AI budget is outpacing your data engineering and stewardship spend, you’re prepaying for disappointment.
Applied AIAI Reality: The Race to (P)doom End-times
Talk of a “frontier slowdown” until a major lab IPO is a reminder that capital markets, not just research breakthroughs, will shape the pace of frontier model deployment. For operators, the practical takeaway is to hedge roadmaps on frontier access — invest in smaller models and internal capability, not just a single external bet.
Applied AIAI Use for Schoolwork Leaves Students 1.5 Years Behind, OECD Finds
If OECD data holds, unsupervised chatbot use for schoolwork is trading short-term convenience for a 1.5‑year learning deficit — especially in regions where usage is most pervasive. Any organization using AI for training or assessment should assume the same dynamic applies: without structured scaffolding and verification, you’re eroding capability, not building it.