This was the week the AI build-out got its receipt and its invoice at the same time. Nvidia reported $96.2 billion in a single quarter, up 106% year over year, and guided to $108 billion for the next one — settling any question about whether the spending is real. Then look at what else landed in the same seven days: more than a hundred companies warning that organisations have months, not years, to prepare for AI-enabled attacks on critical infrastructure; Sony and Warner suing Anthropic and naming its founders personally; Brussels placing ChatGPT under its strictest platform regime; and inside Meta, a plan to replace staff with AI that the company quietly called off the night before it was due. The technology is proven. The terms of using it are being written right now.
1. Nvidia made $96 billion in a quarter and says memory is the limit
Nvidia reported quarterly revenue of $96.2 billion, up 106% from a year ago, with the data centre division alone contributing $89.0 billion — up 117% — and guided to $108 billion next quarter while forecasting roughly 70% revenue growth for the coming fiscal year. The caveat is the part buyers should read twice: Nvidia warned that memory shortages will limit how fast it can expand, which is the same shortage now showing up as higher server prices and thinner margins across the supply chain.
2. Nvidia is reportedly buying Hugging Face for $12.9 billion
Nvidia has agreed to acquire Hugging Face — the repository where most of the world's open-weight AI models are published and downloaded — for $12.9 billion, according to The Information, though no contract has been signed and the deal could still change. Coming a week after Nvidia paid $6 billion to license Poolside's model-building software, it reads as a deliberate move to own the open-model supply chain, and it is why open-weight companies have become the most sought-after acquisition targets in the industry.
3. 100 companies say you have months to prepare for AI-run cyberattacks
OpenAI, Anthropic, Microsoft, Google, AWS, Cisco, Mastercard and more than 100 other organisations signed an open letter warning that AI-enabled attacks will become far more widespread within months, that hospitals and water utilities are most exposed, and that "status quo security won't be enough." It is not hypothetical: the NSA, CISA and the FBI warned in mid-August that attackers are already using AI to write exploit scripts against the industrial controllers running US water, energy and manufacturing sites.
4. Sony and Warner sued Anthropic and named its founders personally
Sony Music Publishing and Warner Chappell sued Anthropic in California federal court over what they call "a brazen campaign of illegally torrenting, scraping, and downloading copyrighted works," naming co-founders Dario Amodei and Benjamin Mann as defendants and seeking up to $150,000 per infringed work plus an order barring further use. The complaint arrives weeks after Anthropic's confidential IPO filing and leans on the $1.5 billion settlement it already paid authors — a reminder that how training data was acquired, not just how it was used, is now the expensive question.
5. Europe just regulated ChatGPT like a search engine
The European Commission designated ChatGPT a Very Large Online Search Engine under the Digital Services Act — the first time an AI chatbot has been placed in that category — after OpenAI declared 159.1 million monthly users in the EU, alongside new designations for Reddit and Roblox. The three services have four months to run annual systemic risk assessments, submit to independent audits and open their data to regulators and vetted researchers, with penalties reaching 6% of global annual turnover.
6. Meta's plan to replace staff with AI collapsed from the inside
A Reuters investigation revealed Project OT, Mark Zuckerberg's plan to cut some Meta teams by as much as 60% and make the workforce "AI native"; the company called off the second wave of layoffs the night before it was due, after its own AI agents and coding tools caused reliability and security problems and failed to lift productivity as expected. The wider evidence is pointing the same way — of 600 HR leaders who made AI-driven cuts in the year to February, only 8.4% said the restructuring delivered what was promised.
7. OpenAI's ad business hit $1 billion a year and moved into Europe
OpenAI's advertising business reached a $1 billion annualised revenue run rate, and the company expanded its ChatGPT Sponsored pilot into 31 European markets — its largest geographic rollout yet, with tens of thousands of marketers now buying placements. The expansion lands in the same week as the EU designation, which means OpenAI is building an ad business directly into the bloc's transparency, profiling and ad-repository rules rather than retrofitting for them later.
8. Google started selling AI by industry instead of by seat
Google Cloud launched Gemini Enterprise for Financial Services and Gemini Enterprise for Legal, the first in a planned series of packaged, industry-specific agentic products built on its governed enterprise platform rather than sold as a general assistant. The market read it immediately — Thomson Reuters shares slid about 3% the same week it announced its own first AI model, as buyers started asking whether their vertical software vendor or their cloud provider owns the workflow.
9. Ottawa put $195 million into a Toronto quantum factory
The federal government signed a C$195 million funding agreement with Xanadu, drawn from Canada's Strategic Response Fund, to build advanced manufacturing facilities for photonic quantum computing in a converted Toronto soup factory — a step toward the roughly US$1 billion quantum data centre the company wants to build. It is a notable contrast with the week's other Canadian technology story, where Alberta's premier spent an hour fielding public objections to AI data centres: capital is flowing to compute Canadians will accept.
10. MIT calls AI a watershed while Chicago's core goes analog
MIT published a responsible-use policy and a set of recommendations urging "AI-aware" curricula, disclosure of AI use and ongoing faculty support, with president Sally Kornbluth telling the community that AI will force practical changes to the education model — while the University of Chicago is making its social sciences core fully analog this autumn and Florida moved to require AI policies at all 28 of its public colleges. A week after Purdue made AI coursework a graduation requirement, the sector is visibly splitting between teaching with the technology and deliberately teaching without it.
