Higgsfield's AI can now make videos over 10 minutes long, up from one
Why it mattersMarketers and creators using Higgsfield can now produce 10-minute videos in one tool, where the limit was one minute.
Why it mattersMarketers and creators using Higgsfield can now produce 10-minute videos in one tool, where the limit was one minute.
What happened. Higgsfield now builds AI video from a minute up to more than 10 minutes, using OpenAI's GPT-6 Astra for video, GPT-Image-2.5 for visuals and GPT-Live for voice.
Why it mattersGrok 4.7 is now a option, so you can test it for agents without a new vendor contract.
What happened. Grok 4.7, built by SpaceXAI, can now be called through Amazon Bedrock alongside other providers' models. SpaceXAI says it was trained to spend longer on hard tasks and to verify its own work, and it has a reasoning effort setting.
Why it mattersFrench startups building AI assistants now have NVIDIA as an investor, which matters if you sell to or compete with them.
What happened. NVIDIA says it has been investing in French startups, and that French companies now run agentic AI in aerospace, banking, government and research. It ties this to compute commitments made at the February 2025 AI Action Summit and the 2025 and 2026 Choose France events.
Why it mattersThe DOE is treating mid-Atlantic grid stress as a blackout risk, so plan for outages now.
What happened. Energy Secretary Wright issued an emergency order for the mid-Atlantic power grid. The Department of Energy says it aims to stabilize the grid and reduce the risk of blackouts as stressed conditions are anticipated.
Why it mattersFederal money is heading to 2.5 of new grid capacity, and it is aimed at power costs.
What happened. The Energy Department will back the Tennessee Valley Authority's plan to build 84 power-line and local distribution upgrades spanning eight states, adding about 2.5 GW of capacity. The award totals roughly $581M: $250M from the federal government and about $331M from TVA and its partners.
Why it mattersArm says trust checks must work across vendors, so ask suppliers how their interoperates.
What happened. Arm argues that confidential computing moves attestation, the check that a server is running trusted code, from an internal data-centre task to something customers need to verify. AI systems that mix CPUs, GPUs, accelerators, firmware and software from different suppliers make a shared standard for those checks necessary.
Why it mattersLaptops with 1 of local AI compute arrive Oct. 16, so on-device agents become shippable.
What happened. NVIDIA and Microsoft say RTX Spark laptops go on sale Friday, Oct. 16. The chip pairs 128GB of shared memory with 1 petaflop of AI compute, and NVIDIA says its partners are readying apps ahead of launch.
Why it mattersOne central subscription now replaces per-account acceptance of private pricing offers for third-party models.
What happened. Amazon Bedrock now supports managed entitlements, so a company subscribes once to a third-party AI model from a central AWS account and shares access across its other accounts. AWS describes two ways to activate Marketplace Private Pricing Offer entitlements in AWS License Manager, including a lightweight script.
Why it mattersAWS now has one supported route for agent-driven SageMaker work, replacing the older AWS Labs plugins.
What happened. AWS added an AI/ML skill to its Agent Toolkit, letting a coding agent handle SageMaker AI model work from choosing a model through to going live, next to the toolkit's EC2 and Bedrock skills. The toolkit now replaces the earlier AWS Labs skills and plugins as the one supported integration.
9 stories, about 3 min 23 sec. Nothing was left out to fill the time.
“Make something people want.”