TODAY IN AI
Nvidia CEO Jensen Huang / Source: I-Hwa Cheng/Bloomberg
1. Nvidia doubles down on OpenAI: The chip giant is planning to invest up to $100B in OpenAI to build large-scale data centers for training and running AI models. The deal includes deploying 10 gigawatts of Nvidia systems, with the first phase expected to go live in late 2026. The move could reduce OpenAI’s reliance on Microsoft, which relaxed its exclusivity deal earlier this year.
2. Alibaba debuts powerful all-in-one model: The Chinese tech giant has unveiled Qwen3-Omni, a fully multimodal AI that handles text, image, audio, and video in one model. Unlike OpenAI’s GPT-4o or Google’s Gemini 2.5 Pro, Qwen3-Omni is open source under Apache 2.0 license, allowing free commercial use and modification. Available on Hugging Face, the model charges $0.25 per million API tokens. Alibaba claims the model is state-of-the-art on several performance benchmarks.
3. Perplexity adds AI-powered email assistant: The startup’s new Email Assistant lets users schedule meetings, organize emails, and draft replies using AI. The tool is limited to the $200-per-month Max plan and currently supports only Gmail and Outlook. Max subscribers can access the AI assistant by emailing Perplexity’s dedicated address. The AI doesn’t train on user emails, but it can mimic their writing style when drafting responses. See it in action here.
via Superhuman