🤖 Nvidia bets $100B on OpenAI data centers 🤖
Nvidia plans to invest up to $100 billion in OpenAI's infrastructure needs for next-generation AI models.

Nvidia and OpenAI announce $100B partnership represents what they're calling the "biggest AI infrastructure project in history," with Nvidia deploying 10 GW worth of systems representing millions of GPUs to power OpenAI's advanced AI models. The first gigawatt goes online in second half 2026, using Nvidia's upcoming Vera Rubin platform. This massive infrastructure investment signals the enormous compute requirements for training and running cutting-edge AI systems.

📧 Perplexity launches AI email assistant 📧
Perplexity introduced an AI-powered Email Assistant that enables users to schedule meetings, organize emails, and draft replies.

Perplexity launches AI email assistant currently limited to $200-per-month Max plan subscribers with Gmail and Outlook integration, enabling users to schedule meetings, organize emails, and draft replies using artificial intelligence. The tool can mimic users' writing styles while maintaining privacy by not training on user email data, representing Perplexity's expansion beyond search into productivity tools. This move positions the company to compete directly with established email management solutions and demonstrates the growing integration of AI capabilities into everyday productivity workflows.

🚀 Alibaba unveils Qwen3 multimodal AI models 🚀
Alibaba's Qwen team released Qwen3-Omni, a 30-billion parameter model supporting text, audio, and video capabilities.

Alibaba unveils Qwen3 multimodal AI models headlined by Qwen3-Omni, which supports text, audio, and video input and output capabilities under Apache 2.0 license for free commercial use. The release also includes Qwen3-TTS-Flash for text-to-speech and Qwen-Image-Edit-2509 for image editing. This represents a significant advancement in open-source multimodal AI capabilities, potentially challenging proprietary models from Western tech giants while demonstrating China's continued investment in competitive AI model development and commitment to open-source approaches.

INTERESTING TO KNOW

🛡️ Google tackles AI shutdown resistance 🛡️

Google DeepMind has released Frontier Safety Framework 3.0, expanding AI risk monitoring to track whether frontier AI models resist shutdown attempts or develop unusually strong influence over human behavior. The updated framework includes sharpened Critical Capability Level definitions and requires safety reviews before both external launches and internal R&D deployments. This addresses recent studies showing AI systems can develop self-preservation behaviors that complicate human oversight.

💕 Meta launches AI matchmaker 💕

Meta integrates AI capabilities into Facebook Dating with features including a "Meet Cute" weekly surprise connections and AI chatbot dating assistant that helps users find relevant matches through text prompts. The platform now suggests matches based on user interests to improve match quality, causing Match Group's stock to drop approximately 5% following the announcement.