NVIDIA has opened its full Earth-2 AI weather stack, giving researchers, energy companies and weather agencies faster global forecasts and hyperlocal storm alerts in minutes instead of hours.
Microsoft has pulled the curtain back on Maia 200, a custom AI accelerator tuned for Azure that promises up to 3x the FP4 inference performance of Amazon Trainium3 and stronger FP8 throughput than Google’s TPU v7, aiming to cut costs while scaling frontier models.
Nvidia CEO Jensen Huang has arrived in Shanghai just as China weighs whether to let the company’s powerful H200 AI chips cross its borders, putting one of the world’s biggest AI hardware bets under regulatory stress.
Meta’s new Superintelligence Labs has quietly shipped its first internal AI models after just six months of work, as CTO Andrew Bosworth tries to reset the company’s AI story beyond Llama 4 and turn big 2025 bets into consumer‑ready products from 2026 onward.
Elon Musk has escalated his legal fight with OpenAI and Microsoft, asking a US court to award up to $134 billion in “wrongful gains” tied to his early backing of the AI lab. The case, headed for a jury trial in Oakland this April, could reset how nonprofit AI missions, mega‑valuations and Big Tech partnerships collide in the next phase of the AI boom.
OpenAI is testing clearly labeled ads in ChatGPT’s free and Go tiers to keep AI access affordable, protect privacy, and leave premium plans ad‑free.