Frontier AI shows a path to profit
Anthropic's revenue went from about $4.8B in Q1 to $16.8B in Q3 with an 11% adjusted margin — the benchmark for the coming lab IPOs.
Frontier leader, deregulating and building fast
The four labs above 50 on the Intelligence Index are American (Anthropic 58, Meta 56, OpenAI and Google 53), Anthropic told investors its Q3 revenue reached $16.8B, and Amazon and Alphabet alone plan over $400B of 2026 capex. Washington stays light-touch on rules, but enforcement is arriving through other doors: an FTC probe into whether lab security incidents harmed Americans, a Senate probe into data centres, and voluntary independent audits. Gartner and the IMF warn that the spending is running ahead of revenue.
Anthropic's revenue went from about $4.8B in Q1 to $16.8B in Q3 with an 11% adjusted margin — the benchmark for the coming lab IPOs.
The FTC is examining whether lab security incidents harmed people, and Anthropic gave METR inside access with the right to publish.
Muse Wonder Full (56) and the open Muse Spark Open make four US labs at the frontier and give the US a large open model again.
Opus 5.5 and GPT-6 Astra arrived within three weeks of each other; the leaders iterate quarterly.
OpenAI's Dots and Temporal's $550M round show the US stack moving from chat to delegated work.
If you sell into the US, map which state AI laws apply to you (they still do), keep a record of AI incidents in case a regulator asks, and price your AI products assuming US compute stays scarce and expensive into 2027.
Ranked by relevance. EU-wide signals are included for EU members.
IDC data for the first half of 2026: US export controls limited Nvidia's supply while domestic accelerators gained ground. China's parallel AI stack is now measurable in market share.
The 816B-parameter model (42B active) handles text, image, video and audio and runs on two Mac Studios. It is free to use; API providers earning over $20M a year need a separate deal. The strongest US answer yet to Qwen, GLM and Kimi.
Per the FT: about $4.8B in Q1, $11.5B in Q2, $16.8B in Q3 (+46%); $1.70 of revenue per $1 of compute and an 11% adjusted operating margin. Frontier AI now shows a path to profit.
Artificial Analysis places Meta's new model two points behind Claude Opus 5.5 (58) and ahead of OpenAI and Google (53). For now it is only available inside Muse Code, with no public API or price.
Argon scores 53 on the Intelligence Index, level with GPT-6 Astra. The introductory API price of $2/$10 per million tokens doubles afterwards, and Argon uses many output tokens per task — budget on the full price.
The inquiry opened in late September after incidents including OpenAI agents breaching Hugging Face in July; reports say fines could mirror those for late data-breach notices. It is still an investigation, not a charge.
No more one-off partnerships: any agent can now complete purchases inside Shopify stores. Agentic commerce goes from pilot to default channel.
AI providers spend nearly $1T in 2026 and another $6T to 2030; a 12% return would need about $8.5T of revenue, but Gartner sees at most $3.6T. Neoclouds are the most exposed; hyperscalers can absorb it.
Tim Cook confirmed the October beta after Beijing approved Alibaba's models for iPhones. Global products now ship with different model providers per market.
Its Global Financial Stability Report counts 14 deals linking hyperscalers and AI firms: about $300B of equity and $650B of debt against $1.4T of revenue commitments. Hyperscalers now issue about 10% of global investment-grade bonds.
The terms appear to cover websites 'associated' with businesses using Meta's platforms; legal experts call it an AI power grab. Check the terms you accept on any platform.
'Small weight-based replicators' could behave like a new generation of computer worms. Connected devices with weak security become part of the AI risk surface.
Founded by ex-Google researchers, it builds AI that designs chips and learns from each iteration — a bet that compute demand will keep outrunning human design capacity.
1,000 outside contractors joined 3,000 staff to patch flaws that AI surfaced. When AI finds bugs faster than people fix them, remediation capacity becomes the bottleneck.
Both sides aim to formalise a dialogue ahead of a presidential summit next month — the first sign of AI risk being managed between the two powers.
A multimodal open model for always-on agents, plus two compact 'Bolt' models for local machines — part of a wave of US open-weight releases.
Production deployment is planned for 2027; Maia 200 already powers Copilot and Teams in US, European and Australian data centres. Hyperscalers are moving inference onto their own silicon.
On the TheoPhysEval benchmark the model matched leading specialists; in some 'failures' the experts' own reference answers were wrong. Research-grade work is becoming a per-task cost.
AI demand keeps lifting the foundry; it also confirmed talks about a role in Elon Musk's Texas 'Terafab', insisting on operational control.
Together with Shopify's move, it means product data must be readable by agents, not only by people.
The non-profit will publish its own assessments of risks and incidents, and Anthropic gave up the right to block them — a template for independent oversight.
Access to power now sets land value more than location. Energy, not plots, is what developers are really buying.
Dragoneer and Thrive led the round for the AI-driven spend-management company; the figures are reported, not yet confirmed. AI-native finance software is pricing like infrastructure.
Nuclear SMRs move from slides to permits — a long-term answer to AI's power hunger, with GE Vernova Hitachi's boiling-water design.
The theme was agents that keep working after the prompt ends, plus ChatGPT as a shared workspace for people and agents. Expect clients to ask for always-on agents, not chatbots.
Frontier capability keeps getting cheaper: input drops to $4 and output to $20 per million tokens. Projects that were too expensive in spring are worth re-quoting.
Data-centre revenue rose 117% year on year and management called next year's outlook supply-constrained. AI infrastructure demand shows no sign of slowing, which keeps GPU prices — and AI costs — firm.
Released first to vetted business users, then to paid ChatGPT plans and the API. OpenAI positions it as able to operate a whole computer autonomously, with limits on its most advanced cyber capabilities.
A template for regulated industries: humans stay in the loop on sensitive claims and every AI interaction is logged for traceability.
Reported on the Q4 earnings call, with a goal above 30% of tickets worldwide within a year. Customer service is where agent ROI is easiest to prove.
The Series H nearly triples February's $380B mark. Capital is concentrating in two or three frontier labs — expect their platforms to set the price and pace for everyone else.
After years of anomaly detection with C3 AI, agents now investigate causes and propose fixes — the natural next step for any company with sensor data.
The range was $180–190B before. Together with Amazon and Meta, Big Tech's AI build-out is running well above $500B this year.
Higher memory costs pushed the figure up from about $200B; AWS revenue grew 37% to $42.2B. Cloud AI capacity stays tight into 2027.
The AI-agent platform behind deployments at DHL, Kuehne+Nagel, Naturgy and Repsol says it has grown 5× since its Series B. Agents for operations are a real business.
Reliability infrastructure for agents is now a multi-billion category: making agents finish the job matters as much as the model.
Generative Engine Optimisation is officially a market. If customers ask ChatGPT before Google, your visibility there needs measuring.
Coding agents remain the hottest enterprise AI category after chat assistants.
Hyperscalers are co-locating with large power sources, turning former power-plant sites into AI campuses.
A second 500 MW phase arrives a year early: former crypto miners keep pivoting into AI hosting.
Labs are building their own infrastructure teams — and Europe is on Anthropic's capacity map.
The small, fast model closes the 5.5 family. Pricing this low makes high-volume classification and routing almost free — the economics behind news triage, ticket routing and document tagging just changed.
Power, not chips, is becoming the bottleneck. Infrastructure investors are positioning upstream of AI capacity.
Neoclouds keep raising at scale to buy GPUs before going public — watch how the market prices them after recent IPO jitters.
Education, job training, water and energy conservation: social licence is now part of the cost of AI capacity.
The report argues companies overstate benefits and understate costs to communities. Expect tougher permitting and tax-break debates — on both sides of the Atlantic.
Tell us what slows your business down. In 30 minutes we'll show you where AI pays back first — and what it costs.