The Nasdaq fell 5.5% on Thursday, its worst day since April 2025, as investors sold AI and chip stocks after the US midterm elections. Nvidia dropped more than 9%, which CNBC says cost it almost $450bn in market value.
AI and chip stocks slide after the US midterms
Qualcomm also fell more than 9% after a disappointing forecast for the current quarter, and AMD, Micron and Applied Materials each fell by double digits, CNBC reports. CNBC says the election results raised concerns on Wall Street that the AI buildout is set to slow down.
Why it matters: AI vendors are spending on the bet that demand keeps growing. Pay for tools that save you time or money now.
OpenAI promises to pay for its own data centre power
For data centres it owns and runs, OpenAI will ask for community input, publish construction plans and avoid non-disclosure agreements with public officials, CNBC reports. It also restated its promise to pay the full cost of powering those sites, and asked lawmakers to turn similar commitments into law. It came two days after voters sided against data centres in all but one of 41 local ballot measures in 10 states.
Why it matters: power and land are becoming a local fight, and that cost shows up in AI prices over time. Watch what your AI services cost, not only what they can do.
The US Justice Department sets up a Super Intelligence Task Force
The task force, announced on 5 November, targets misuse of what it calls super intelligence by foreign state actors, transnational criminal groups and people or companies committing fraud, according to the Justice Department. Associate Deputy Attorney General Stephen Weiss said the department wants to "ensure companies exercise vigorous internal controls" so the technology behaves as intended.
Why it matters: if you build on AI or sell it, keep a record of what your AI systems are allowed to do and who checks them.
OpenAI says GPT-6 Cyber found 81 Firefox flaws and built 948 Linux exploits
OpenAI says its researchers used GPT-6 Cyber to identify 81 previously unknown high-severity vulnerabilities in Firefox and develop 948 privilege escalation exploits in Linux. Mozilla has released fixes for many of the Firefox flaws. For Linux, OpenAI says it disclosed 304 critical findings to maintainers with candidate patches and is still getting other vulnerabilities fixed. OpenAI classifies the model as Critical in cybersecurity under its own Preparedness Framework.
Why it matters: these are OpenAI's own numbers, but holes in old software are being found faster than they are all fixed. Keep your updates current.
Sarvam releases Prism, a model that picks instead of writing
Sarvam AI released Prism, an open-weight model built on Sarvam 105B that chooses from a set of options and returns a probability for each, Inc42 reports. The weights are under the Apache 2.0 licence, and it's also on an API. Sarvam says the median response time is 44 milliseconds on an Nvidia B200; one outside engineer measured a 65-millisecond median.
Why it matters: sorting tickets, leads or documents into your own categories doesn't need a chatbot. A small model you can run on your own servers can do it.
ChatGPT can now make animated GIFs
OpenAI is rolling out Images 2.6 to all ChatGPT and Codex users on web, desktop and mobile, The Next Web reports. It can turn a prompt or a photo into a GIF whose speed and single frames you can edit. Developers get two API models, GPT-Image-2.6 Flare for speed and Sunburst for quality.
Why it matters: a looping product GIF for a shop's social posts is now a one-line request.
Mistral adds managed agents to Studio
Mistral says developers can now build agents in Studio that carry out multi-step tasks, with limits on which tools, data and actions each agent can reach, according to Mistral. The models and the platform can run in a company's own cloud or servers. Tata Consultancy Services is among the companies Mistral lists as using Studio in production.
Why it matters: if keeping data in-house has stopped you trying agents, this is one more option.
Worth watching: whether the selloff slows data centre plans, and how the Justice Department's task force defines the "internal controls" it wants from AI companies.
Arthat AI
Working on something similar at your company?
Arthat AI builds these automations for Indian businesses every week. If you want to talk through whether this fits your situation, we'll give you a straight answer — even if the answer is "don't build it."
