Nvidia’s Groq Play: License, Talent, and a $20B Rumor
Nvidia has struck a non-exclusive licensing deal with AI chip rival Groq, and will hire Groq founder Jonathan Ross, president Sunny Madra, and other employees. CNBC also reported Nvidia is acquiring Groq assets for $20B, but Nvidia told TechCrunch this is not an acquisition of the company and didn’t clarify the full scope. If the reported figure is accurate, it would rank as Nvidia’s largest-ever purchase, while still leaving Groq as a separate company.
Groq is pushing its LPU (language processing unit) chips as an alternative to GPUs, claiming they can run large language models 10× faster while using one-tenth the energy. Ross has deep pedigree in AI hardware—he previously helped invent Google’s TPU—which helps explain why Nvidia would want both the tech and the team. Groq also raised $750M in September at a $6.9B valuation, and says it now supports AI apps for 2M+ developers, up sharply from about 356,000 last year.
Waymo’s Robotaxi Upgrade: Gemini in the Passenger Seat
Waymo appears to be testing Google’s Gemini as an in-car “ride assistant” for its robotaxis, based on what researcher Jane Manchun Wong found inside Waymo’s mobile app code. The discovery includes a 1,200+ line internal “meta-prompt” that lays out how the assistant should behave, even though the feature isn’t live in public builds yet. Waymo says it’s always experimenting with rider features, and hinted that what’s in testing may or may not ship.
The prompts suggest Gemini would do more than chat: it could answer questions, control select cabin features (like temperature, lighting, and music), and reassure riders when needed—while staying brief and unobtrusive. It’s also designed with firm guardrails: it should not comment on real-time driving decisions, avoid being a spokesperson for the driving system, and can’t handle emergencies or real-world actions like ordering food or making reservations. Waymo has already used Gemini’s “world knowledge” behind the scenes to help train vehicles for complex, rare scenarios, so this looks like the passenger-facing layer of a broader AI strategy.
taly’s WhatsApp AI Crackdown: Meta Told to Hit Pause
Italy’s competition watchdog (AGCM) has ordered Meta to suspend a policy that would block rival general-purpose AI chatbots from being offered via WhatsApp’s business tools/API. The authority says it has enough grounds to suspect Meta is abusing a dominant position, potentially restricting market access and technical development in AI chatbot services, and risking serious, irreparable harm to competition while the investigation continues.
The policy stems from an October change to WhatsApp’s business API rules and was set to take effect in January, affecting bots from providers like OpenAI and Perplexity—while still allowing business customer-service bots that use AI. The European Commission has also opened an investigation into the same policy, and Meta says the decision is “fundamentally flawed,” arguing WhatsApp isn’t an app-store-like route to market and that chatbot usage strained systems not designed for it; Meta says it will appeal.
Alexa+ Gets a Bigger “Do Stuff” Button
Amazon is expanding Alexa+ with four new integrations — Angi, Expedia, Square, and Yelp — rolling out starting in 2026. The additions let people do practical tasks through the assistant, like comparing and booking hotels, getting home-service quotes, and scheduling appointments (including salon bookings), without bouncing between apps.
These join Alexa+’s existing partners (including services for dining, rides, and more), pushing Alexa toward an AI “app platform” model where you talk naturally and refine requests as you go. Amazon says early usage shows strong engagement with some home and personal-service integrations, but the big test will be whether this feels faster and easier than using websites and mobile apps — without the assistant’s suggestions coming off like ads
Your Year with ChatGPT: The Wrapped-Style Recap for AI Chats
ChatGPT is rolling out a Spotify Wrapped–style year-end recap called “Your Year with ChatGPT” for eligible users in select markets, including the U.S., Canada, the U.K., Australia, and New Zealand. It’s available for Free, Plus, and Pro users who have reference saved memories and reference chat history turned on (and meet a minimum activity threshold).
The recap is designed to be lightweight and user-controlled, with catchy graphics, personalized “awards” (like “Creative Debugger”), plus a generated poem and image reflecting your year. It’ll be promoted on the home screen but won’t auto-open, and it’s available on both web and mobile—while Team, Enterprise, and Education accounts won’t get access.