Chain of News 27/07/2026

**Top Story** The development of NVIDIA Nemotron 3 Ultra has led to significant advancements in open models, achieving higher accuracy and efficiency in agentic RTL coding. This breakthrough is crucial as modern chip design is increasingly limited by engineering time, and RTL development and verification require specialized hardware. The Nemotron 3 Ultra's improved performance has the potential to revolutionize the field of chip design, enabling faster and more efficient development of complex AI systems. As a result, developers can expect to see improved performance and reduced development time for AI-powered systems. The implications of this development are far-reaching, and it will be interesting to see how the industry responds to this new technology. With the Nemotron 3 Ultra, NVIDIA is poised to lead the charge in agentic AI development, and its impact will be felt across the industry. SOURCES: [1] **AI Models & Research** The ability to run a 28.9M parameter LLM on an $8 microcontroller is a significant achievement, demonstrating the potential for AI models to be deployed on low-power devices. This development has important implications for developers, as it enables the creation of more efficient and cost-effective AI systems. The use of large language models on microcontrollers can also enable new applications, such as edge AI and IoT devices. Additionally, the Inside NVIDIA Rubin GPU Architecture provides valuable insights into the development of agentic AI, highlighting the importance of always-on AI factories and the role of GPUs in powering these systems. The NVIDIA Vera CPU is also noteworthy, as it is designed to provide maximum single-thread performance in agentic AI, enabling agents to operate in sandboxes and execute code more efficiently. SOURCES: [2], [3], [4] **Developer Tools & Frameworks** The development of ModelExpress is a significant advancement in distributing model artifacts, enabling faster and more efficient deployment of AI models. This technology has the potential to greatly reduce the cost and complexity of moving large model checkpoints, making it an essential tool for developers working with large-scale AI systems. The NVIDIA NVLink is also an important development, providing a scale-up network for AI factories and enabling the deployment of larger and more complex AI models. Furthermore, the NVIDIA GB300 NVL72 has set a world record for MoE pre-training, demonstrating the potential for large-scale AI training and the importance of mixture of experts (MoE) in this field. SOURCES: [8], [9], [10] **Industry & Business** Anthropic has presented Claude Opus 5, a model AI that can correct itself, marking a significant development in the field of AI. This technology has the potential to greatly improve the accuracy and efficiency of AI systems, and its implications will be closely watched by developers and industry leaders. The development of brain wave readings as a potential unlock for physical AI is also an interesting area of research, with potential applications in frontier physical AI models. As the industry continues to evolve, it will be important to monitor these developments and their potential impact on the field of AI. SOURCES: [5], [7] **Worth Watching** The advancement of semiconductor innovation across materials engineering and manufacturing is an important area of research, with significant implications for the development of AI systems. As AI workloads increase, the demand for high-performance semiconductors will continue to grow, and developments in this field will be crucial in meeting these demands. The use of brain waves as a potential unlock for physical AI is also an interesting area of research, with potential applications in frontier physical AI models. These developments are worth watching, as they have the potential to greatly impact the field of AI and its applications. SOURCES: [6]