Elon Musk Warns of ‘All-Out War’ in AI, Starting With Nvidia’s Blackwell

Nvidia has acquired SchedMD, the lead developer of the Slurm workload manager, and launched the Nemotron 3 family of open-source AI models. This acquisition and product launch signify Nvidia's commitment to enhancing its open-source offerings in the AI sector.

Korean AI startup Motif Technologies has introduced its latest model, Motif-2-12.7B-Reasoning, which has achieved remarkable benchmark scores, surpassing even OpenAI's GPT-5.1, according to independent assessments by Artificial Analysis. This development highlights Motif's growing influence in the competitive landscape of generative AI.

The Trump administration has granted Nvidia approval to export its H200 AI chips to China, following extensive negotiations. This decision includes a stipulation of a 25% fee on sales, which will be charged upon shipment. The approval reflects a significant shift in U.S. export policy amidst ongoing tensions between the two nations.

xAI, the artificial intelligence startup founded by Elon Musk, has established an enterprise AI sales group to promote its Grok AI models. However, the company's inexperience in selling to large businesses is reportedly hindering its ability to attract potential customers. This challenge highlights the difficulties xAI faces in transitioning from a startup to a competitive player in the enterprise AI market.

Nvidia has released the Nemotron 3 open models, which are built on a hybrid mixture-of-experts architecture aimed at facilitating the implementation of multi-agent systems in enterprises. This release underscores Nvidia's ongoing commitment to advancing artificial intelligence technologies.

Nvidia has launched Nemotron 3, a significant advancement in its AI model offerings, featuring a hybrid mixture-of-experts architecture aimed at improving accuracy and reliability for agentic AI applications. This move positions Nvidia as a major player in the open-source AI landscape, responding to competitive pressures from rivals.

A new system named CUDA-L2 has been introduced, which leverages large language models and reinforcement learning to optimize Half-precision General Matrix Multiply (HGEMM) CUDA kernels. This system has demonstrated superior performance compared to existing matrix multiplication libraries, including Nvidia's cuBLAS and cuBLASLt, achieving significant speed improvements in various configurations.

Nvidia has launched new monitoring software that provides a global overview of where its AI GPUs are deployed, utilizing a customer-managed agent to collect system data and relay it to a centralized dashboard on NGC. This allows operators to analyze performance across various layers, enhancing operational efficiency.