Anthropic claims 1,100% revenue surge as Goldman curbs AI tools

Anthropic says its revenue surged 1,100% ahead of an IPO, while Goldman Sachs is restricting traders and bankers from using some AI tools, in the freshest evidence that artificial intelligence is becoming both a revenue engine and a control problem at once. The revenue figure was reported by The Economic Times. The Goldman move was reported by Bloomberg.

Goldman restricts traders and bankers from some AI tools

Bloomberg reports that Goldman Sachs has blocked traders and investment bankers from using certain artificial intelligence tools. The restriction sits on the risk side of the same technology that Anthropic is treating as a growth engine. The report does not say which tools stay available and which specific desks are covered.

The tension is already priced into market attention. The Kobeissi Letter, a financial markets commentary and research platform, says 42 AI stocks have generated 67% of the S&P 500’s price return since the start of 2024. That means a narrow group of names is doing most of the index’s lifting, even as one major bank tells parts of that complex to stop using some AI tools.

42 AI stocks drive 67% of S&P 500 gains

The skew is specific. Since the beginning of 2024, the 42 stocks identified as AI names account for 67% of the S&P 500’s price return. Moneycontrol attributes the claim to The Kobeissi Letter. It doesn’t prove those gains are sustainable, but it does show how concentrated the AI trade has become inside the broad index.

Trump moves to ban the term artificial intelligence

CNBCTV18 reports that Donald Trump announced plans to ban the term artificial intelligence in documents and promote super intelligence instead. Business Today says the executive order followed a high-level luncheon with Elon Musk, Sundar Pichai, Mark Zuckerberg and Jensen Huang. News18 links the rebranding to a White House tech summit where Trump met leading industry figures.

The language fight is spreading into voter-facing territory. KTVU FOX 2 reports that more Americans are turning to artificial intelligence for answers, including questions about politics and elections, and asks how those apps ensure accuracy. Trump’s preferred rename is not a neutral label change; it comes with a warning in News18’s headline that anyone using artificial intelligence instead of super intelligence is an enemy.

Merz links Germany’s AI era to digital sovereignty

Friedrich Merz says Germany must preserve digital sovereignty in the AI era. The Economic Times reported the line without attaching a detailed roadmap. The point ties neatly to the Goldman and Trump stories: states and banks are both deciding where AI can be used, named and controlled.

Connectivity deals, a Meta robot and how to learn AI

CNBC reports that Lumentum’s CEO sees an enormous opportunity in a massive AI connectivity deal with Marvell. NDTV reports that Zuckerberg wants to give AI a body through Meta’s new humanoid robot. Those moves treat artificial intelligence as something to be built into networks and machines, not just software screens.

The same cluster of reports carries a sharper warning. Mint on MSN quotes Nate Soares, a machine-learning researcher, president of the Machine Intelligence Research Institute and a former Google AI expert, saying he doesn’t save for retirement now. Soares estimates the probability of AI-driven human extinction over the next 10 years; the probability figure itself is not included in the summary. The timing is still specific: the next 10 years.

For people trying to work in the field rather than only worry about it, Mint reports advice from IISc Bangalore professor Prathosh AP. He recommends studying math, following a three-semester learning path and putting weight on mentorship, clear reasoning and research that addresses real-world problems. Analytics Insight’s practical AI glossary for 2026 frames the same endpoint: modern AI can use tools, access information, complete multi-step tasks and work with other AI systems, while agentic AI adds goal-based action. The path is finite and concrete: three semesters, with mentorship and research on real-world problems.