NVIDIA and Foxconn Collaborate to Develop AI-Focused New Data Center


NVIDIA announces its first custom CPU for Agentic AI, Vera, now shipping to top AI firms and in mass production. Featuring 88 self-developed Olympus cores, 1.2TB/s memory bandwidth, and 50% performance boost over Grace, it supports large-scale agentic AI workloads, laying the foundation for future development.....
Netflix announces new internal studio 'INKubator' focused on generative AI to produce high-quality short animations, blending art and technology with feature-film standards. Currently recruiting producers, software engineers, and CG artists to build a diverse team.....
Artificial intelligence startup Andon Labs conducted a six-month experiment where four AI models - Claude, GPT, Gemini, and Grok - autonomously managed a radio network under the same initial conditions, including identical prompts, a $20 budget, and full control. The results showed drastically different extreme behaviors from each model after being left unattended, ranging from chaos to efficiency, highlighting the unpredictability of AI operating independently.
NVIDIA CEO Huang Renxun emphasized the importance of AI hardware in a virtual lecture at Stanford University, reiterating his opposition to export controls on AI chips, which he believes are ineffective and may backfire. He refuted the analogy of high-end AI chips to nuclear weapons, calling it absurd and baseless, and pointed out that billions of people are already using AI technology.
After the merger of SpaceX and xAI into SpaceXAI, over 50 top researchers and engineers have left since February this year, affecting core teams such as programming assistants, world models, and Grok voice interaction. The pre-training team has been severely poached by competitors, raising concerns about the sustainability of the technology.