Afternoon edition — Monday 9 November 2026

WordPress.com owner Automattic launched Spacefast on 9 November as a new Automattic company. People keep building in their own AI tool, such as ChatGPT, Claude, Claude Code, Codex or Cursor, and Spacefast supplies hosting, custom domains and a database, then puts the work online behind a private link that a team can comment on.

App authentication, transactional email and shared team knowledge are coming in the next few weeks, Automattic said. The service is free to start, with a Go plan at $5 a month and a Pro plan at $15 a month per member, and it connects through official plugins for Claude, ChatGPT and others, or through an API, a CLI and an MCP server.

TechCrunch reported that Automattic soft-launched the product in September and that it competes with Lovable and Emergent. Its founder, Batuhan İçöz, told TechCrunch that enterprises face IT sprawl as marketing teams turn out landing pages on such services.

At the Conference on Robot Learning in Austin, Nvidia on 9 November introduced GR00T 2, which it describes as an open frontier world action model built on its Cosmos platform. Its research page says the model predicts future video and robot actions together, and was pretrained on 402,730 hours of egocentric human video, 10,184 hours of handheld-gripper demonstrations and 23,500 hours of robot teleoperation. Nvidia says 100,000 hours of that human video are openly available on Hugging Face.

Nvidia claims that on 50 tasks never demonstrated on the target robot, GR00T 2 completed 40 at least once without task-specific demonstrations, and that five demonstrations per task raised success to 95% after a 30-minute fine-tune on a single RTX 5090. A smaller GR00T 2 Turbo runs on the Unitree G1 humanoid itself, at 124 milliseconds per prediction, according to the research page.

Nvidia also announced Isaac Lab 3.0 and new open benchmarks at the conference, and said two of its papers, GRAIL and RoboTTT, are among six finalists for the Best Paper award.

Google said on 9 November that it has expanded its eighth-generation TPU line with two new systems, TPU 8tx and TPU 8i Max, joining the TPU 8t and TPU 8i it introduced in April. It is the first time Google has introduced four TPU systems in one year. In a post by Amin Vahdat, Google said TPUs now have nearly 100 customers, including frontier labs, trading firms, retailers and universities.

TPU 8tx uses the same silicon as TPU 8t, paired with AMD CPU hosts, and is aimed at regulated enterprises and governments. Google says it keeps sensitive data and model weights protected and isolated even while in active use, and claims up to 2x lower latency than TPU 8t with a near-zero performance penalty. Google pitches it for private fine-tuning, distillation and reinforcement learning on one system.

TPU 8i Max puts two TPU 8i chips on a single package. Google says that doubles compute, SRAM and HBM bandwidth over TPU 8i within the same rack footprint, for workloads such as many agents working across a whole codebase or hours of video. Google also said its Cloud Pathways runtime and GKE dynamic slicing are now generally available.

Disney and Adobe announced on 9 November a multi-year deal to co-develop custom generative AI models for Disney, Reuters reported. Using Adobe Firefly Foundry, which lets companies build and deploy their own models, the two will train models on Disney's intellectual property; the models will be private and proprietary to Disney.

The work starts with Disney's Experiences division, which runs its theme parks, resorts and cruise lines, and with concept exploration at StudioLAB, the innovation hub of Walt Disney Studios. Adobe will also work with ESPN on fan engagement and its content archive. Walt Disney Imagineering has already deployed a custom image model and a sketch-to-image model that turns hand-drawn sketches into rendered concept art.

Kyle Laughlin, Disney's senior vice president of R&D technology and engineering, told Reuters the tools cut work that took months to hours and days, and were "in no way meant to replace" roles at Imagineering. Reuters noted a Bloomberg report earlier this month that Disney was considering cutting thousands of jobs.

Meta said on 9 November that it is expanding its AI age-estimation technology globally to place millions more teens on Instagram into Teen Account protections, even when they gave an adult birthday. Teen Accounts limit who can contact teens, the content they see and their time on the apps.

Meta said it is still rolling the technology out on Facebook and will begin launching similar technology globally on Threads later this year. In Australia it has started using Apple's age-range signal as an extra method, and parents in the US will see an in-feed video on Instagram this month explaining the approach.

Meta tied the expansion to its commitments under an agreement with 52 bipartisan attorneys general. The Verge reported that the system looks at signals such as the profiles and content a user interacts with and the birthday messages they receive, and that it launched in the US, the EU and Brazil last year.