💰 Nvidia seals $20B Groq chip deal 💰
Big Tech makes its biggest AI infrastructure bet yet

Nvidia has signed a licensing agreement worth approximately $20 billion with AI chip startup Groq, acquiring rights to the company's LPU (Language Processing Unit) technology. As part of the deal, Groq's CEO, president, and key engineers are joining Nvidia's team. Groq's LPU chips are recognized for running AI models with exceptional speed and cost efficiency, making them highly attractive for inference workloads. The deal gives Nvidia critical pricing leverage as GPU competition intensifies across the inference market. The move represents one of the largest AI infrastructure acquisitions to date.

🎯 Google's 2025 AI review 🎯
Search giant highlights conversational AI, coding tools, and autonomous agents

Google has released its comprehensive 2025 AI accomplishments report, covering major updates to models, products, scientific breakthroughs, and research initiatives. Key themes include conversational AI integrated into Search, advanced coding assistants, and the evolution toward autonomous agents that take action rather than just answer questions. The company also published 40 practical AI tips covering productivity workflows, creative applications, and business use cases. The recap demonstrates Google's accelerating push toward agentic capabilities that could reshape how users interact with its entire product ecosystem.

🛡️ OpenAI hires for self-improving AI safety 🛡️
New leadership role signals proactive approach to frontier AI preparedness

OpenAI CEO Sam Altman has posted a job listing for a "Head of Preparedness" role focused on planning for and securing increasingly advanced AI models. The position specifically mentions preparing for "systems that can self-improve," signaling OpenAI's proactive stance on AI safety as capabilities advance. This hire will be critical in developing protocols and safeguards as OpenAI continues pushing toward more powerful models. The strategic move comes as global regulatory scrutiny around AI safety intensifies, forcing frontier labs to formalize capability-risk planning.

INTERESTING TO KNOW

🤖 China's TARS robot masters embroidery 🤖

Robotics startup TARS has demonstrated a world-first achievement: a humanoid robot threading a needle and stitching embroidery with sub-millimeter precision using coordinated two-handed movements. The breakthrough tackles one of automation's most challenging problems—delicate bimanual manipulation—and opens doors to precision manufacturing previously reserved for skilled human workers. The technology could transform industries requiring fine motor skills, from textile production to electronics assembly and medical device manufacturing.

⏱️ Claude Opus 4.5 handles 5-hour tasks ⏱️

AI evaluation firm METR has published new analysis finding Anthropic's Claude Opus 4.5 capable of completing tasks requiring nearly 5 hours of sustained work—the longest duration of any AI model evaluated to date. The finding demonstrates substantial improvements in maintaining context across complex, multi-step workflows, signaling a leap toward replacing professional knowledge work that requires extended concentration and task persistence.