Google broadens its AI hardware portfolio with two new specialized chips

techcrunch.com Nov 10, 2026

Google broadens its AI hardware portfolio with two new specialized chips

Curated by Shubham · TechCapsules · Nov 10, 2026

AI-summarised brief · reviewed before publication

Google unveiled two new tensor processing units—TPU 8tx and TPU 8i Max—expanding its eighth‑generation AI chip lineup. The TPU 8tx, co‑designed with AMD, adds hardware‑level security for confidential training and inference, allowing models to be refined on proprietary data without data movement, and includes a custom inferencing network for on‑the‑fly model distillation. The TPU 8i Max merges two TPU 8i inference chips into a single package, doubling memory, compute and bandwidth to support larger query workloads. Both chips aim to lessen Google Cloud’s reliance on external silicon, particularly Nvidia, as hyperscalers and AI labs develop their own processors. The announcements arrive amid heightened scrutiny of Nvidia’s leadership following CEO Jensen Huang’s personal controversy.

💡 Why It Matters

· By embedding security and scaling memory in its own silicon, Google can offer cloud customers private‑data AI services that competitors cannot match without third‑party chips.

AI Capsules Google Semiconductors

Read full article on techcrunch.com

Meta’s AI agent Muse lands on Apple Watch and Android Wear OS

Meta announced that its Muse AI agent is now available on Apple Watch and Android Wear OS, enabling users to interact via voice, text, or dictation directly from their smartwatches. The rollout follows a month‑old iPad app launch and expands Muse’s presence across Apple devices, underscoring Meta’s push to embed its AI assistant within its ecosystem. Muse can manage emails, meetings, bills, and more, and offers customizable avatars, video chat, and new “Spaces” for focused topic conversations. The smartwatch launch positions Muse among dozens of competing consumer AI agents.

💡 By extending Muse to wearables, Meta taps a growing market for on‑the‑go AI assistance, potentially reshaping how users access productivity tools outside their phones.

9to5google.com · Nov 10

Google and Adobe partner to add app integration in Google Drive, Gemini

Google announced a partnership with Adobe to streamline PDF handling in Google Drive and Gmail. Users will be prompted to open PDFs in Adobe Acrobat, creating a direct transfer path that bypasses Drive’s limited editing tools. The deal also grants Gemini, Google’s AI model, access to Adobe’s creative suite, potentially enabling near‑real‑time editing of PDFs and images within Google’s AI interfaces. The Drive and Gmail integration is slated for immediate rollout, with Gemini features to follow.

💡 By embedding Adobe’s robust PDF and image editing capabilities directly into Google’s core services, the partnership addresses a long‑standing user pain point, positioning Google as a more competitive document‑management platform.

androidcentral.com · Nov 10

Google Gemini’s Memory summary allows users to edit or delete stored personal…

Google has introduced Memory summary for Gemini, a feature that displays the personal data the chatbot has collected from users’ chats. The summary is organized into distinct sections—Essentials, career, interests, etc.—and users can edit or delete entries that are incorrect. Currently, the tool is available only to U.S. users over 18 with an active Google AI plan and in English, though the web version appears to work outside the U.S. and for some free users. The feature is accessed via Settings > Personal Intelligence > Memory summary.

💡 By enabling users to curate the data that powers Gemini, Google addresses growing concerns over AI privacy and data accuracy, offering a practical way to correct misrepresentations before they influence future interactions.

9to5mac.com · Nov 10

Anthropic upgrades Claude with new Fable 5.2 model, details here

Anthropic released Claude Fable 5.2, the latest upgrade to its top-tier Mythos-class model. The update improves coding, reasoning, and vision, and delivers clearer communication than previous versions. Performance gains include up to 30% faster output generation, with pricing at $8 per million input tokens and $40 per million output tokens. Fable 5.2 is positioned for ambitious technical tasks, while users are advised to start with Claude Opus 5.5 for general coding and knowledge work. Anthropic also offers additional Fable usage credits for eligible paid plans.

💡 The faster, cheaper model enables developers to prototype and deploy complex AI solutions more efficiently, potentially reshaping how businesses integrate advanced language models into production workflows.

thequantuminsider.com · Nov 10

Google Researchers Use AI And Mathematical Proofs to Cut Quantum Computing Requirements

Google DeepMind and Google Quantum AI researchers demonstrated that AI agents, combined with computer‑checked mathematical proofs, can redesign quantum algorithms to use far fewer resources. Their system generated programs that cut the combined logical‑qubit‑and‑T‑gate cost by about elevenfold for a fusion‑simulation task, eightfold for a photochemistry calculation, and 2.5‑fold for a drug‑metabolism model. The approach employs the Lean proof assistant to specify correctness, accuracy and resource budgets; AI agents iteratively propose code and receive proof‑based feedback, revising solutions until they meet the constraints. While the optimized programs still exceed the capabilities of near‑term fault‑tolerant quantum hardware, the method shows a pathway to bridge the gap between algorithmic demands and hardware progress, potentially accelerating scientific applications of quantum computing.

💡 By proving quantum programs meet strict resource limits before hardware exists, the technique accelerates the readiness of quantum computers for real‑world scientific problems.

techcrunch.com · Nov 10

AI hardware startup Nominal triples valuation to $3B in 8 months

Nominal, a four‑year‑old Austin startup specializing in AI platforms for engineering and testing complex hardware, secured a $240 million Series C that raised its valuation to $3 billion. The round was led by Lightspeed Venture Partners and co‑led by Insight Partners, following a $155 million Series B in March that valued the company at $1 billion. Investors include Sequoia, General Catalyst, and Lux Capital. Nominal serves 80 commercial and government customers, including defense firm Anduril, and recently won a $71.2 million Air Force STRATFI contract.

💡 The rapid valuation jump underscores growing demand for AI‑driven hardware validation, positioning Nominal as a key player in defense and industrial automation.