NVIDIA

Jensen Huang announced at GTC Taipei 2026 that the Agentic AI era has arrived, shifting AI from content generation to autonomous task execution. NVIDIA launched infrastructure products like Vera Rubin and Vera CPU, driving a computing paradigm shift where AI becomes a direct generator of profit and GDP.

NVIDIA launches Cosmos 3, an open omni-model for physical AI based on a novel mixture-of-transformers architecture, capable of generating physics-accurate synthetic video, serving as a world model and simulator, and enabling training for robotic and mobile intelligent systems.

NVIDIA today launches Nemotron 3 Ultra, a 550B-parameter open model built on the same architecture as Nemotron 3 Super, optimized for long-running AI agents. It employs LatentMoE to quadruple the number of experts at the same inference cost, introduces multi-token prediction to boost single-user inference speed, and is released under the Linux Foundation’s Open MDW license to enable enterprise deployment.

NVIDIA introduces the 550B-parameter Neotron 3 Ultra, a mixture-of-experts agent model trained for task orchestration, outperforming many trillion-parameter open agents on benchmarks, with full data and recipe transparency to enable enterprise on-prem deployment and fine-tuning.

The platform cycle has turned, and partners are becoming increasingly important in helping customers navigate change. Elastic's Elasticsearch Platform sits at the intersection of today's largest enterprise priorities, offering AI-powered search, unified observability, and modern security operations on a single data layer. This platform creates a significant opportunity for partners who move early, as it naturally expands within accounts, providing a compounding services opportunity that point so

The Microsoft Surface RTX Spark Dev Box, designed in Mac mini form factor, features NVIDIA's latest RTX Spark GPU with 128GB memory and 1 petaflop computing power, enabling local execution of 120 billion parameter models without cloud GPUs. Its anodized aluminum 3D-printed chassis serves as a passive cooling system through 1,000 top vents, while pre-installed Windows 11 Pro includes developer tools like WSL, VS Code, and GitHub Copilot. Priced between $3,000-$3,500, it targets AI workloads and w

NVIDIA deploys Nous Research's Hermes Agent via NemoClaw, enabling one-time workflow teaching with persistent memory, integration with Slack, Outlook, GitHub, and turning chat corrections into reusable skills.

NVIDIA introduces self-evolving Hermes agents that autonomously improve through continuous learning and feedback, significantly enhancing task efficiency and accuracy in enterprise AI.

NVIDIA implemented a self-evolving Hermes agent system using NemoClaw blueprint and Open Shell policies, balancing internal data access with external security boundaries.

NVIDIA Nemotron 3 Ultra is now available on Amazon SageMaker JumpStart with one-click deployment. This 550B-parameter MoE model is designed for long-running agents, delivering 5x faster inference, 30% lower cost, and 1M token context support.

Nvidia's RTX Spark chip is ill-suited for 2026 Agentic AI due to its GPU-heavy, CPU-light design. Ben Thompson argues local agents now require strong CPU plus cloud inference, not just local GPU power; the chip targets 2023 chatbot use cases and faces Windows on ARM compatibility risks.

NVIDIA announces Cosmos 3, positioning it as the leading breakthrough in physical AI worldwide.

Nvidia announced Neotron 3 Ultra open-source model (550B parameters) and Vera CPU at GTC Taipei, with 5x faster inference and 30% lower cost for the former, and AI agent-optimized CPU for the latter.

NVIDIA released Nemotron 3 Ultra, a 550B MoE open model for long-running agents, offering 5x faster inference and 30% lower costs.

NVIDIA's DGX Spark and NemoClaw provide a one-click AI agent deployment solution, simplifying the traditionally complex process of sourcing models, configuring inference backends, installing runtimes, and integrating systems into a single command, significantly improving development efficiency.

NVIDIA launches Cosmos 3, the first omni-model for physical AI, supporting multimodal inputs and complex physical scene understanding, applicable to robotics, autonomous driving, and industrial simulation.

NVIDIA introduces Forward Dynamics, a physics-based simulation technology that enables AI training in virtual environments for autonomous driving, robotics, and industrial design.

NVIDIA DGX Station is being delivered to developers and researchers, enabling enterprise teams to achieve local data center-level performance with GB300-powered systems.

NVIDIA AI introduces Inverse Dynamics, a technique that infers actions from state changes, applicable in robotics, physics simulation, and AI control.

Ethan He shared in the Latent.Space podcast that video generation models derive most intelligence from language, not video data; idea-to-code is fast now, but backend development remains a bottleneck, with future frontiers focusing on world models, continual learning, and agents.

Satya Nadella thanks Jensen Huang for joining Build and expresses gratitude for the deep partnership with NVIDIA across cloud and edge.

NVIDIA released the post-trained Ultra open frontier model, optimized for agent frameworks like OpenClaw, Hermes Agent, and LangChain, enabling developers to build specialized cross-domain agents.

NVIDIA has fully open-sourced Nemotron 3 Ultra, including weights, synthetic data, and post-training recipes, now available on Hugging Face.

This post summarizes recent developments from Anthropic, Microsoft, NVIDIA, OpenAI, and Google, including Anthropic filing an S-1 with the SEC, Microsoft and NVIDIA launching new laptops, OpenAI announcing a $13 billion AI Resilience program and building a 1GW data center in Michigan, and Google planning to raise $80 billion for new data centers.

Lex Fridman documented his day-long activities with NVIDIA CEO Jensen Huang in Taiwan, including interactions with thousands of engineers and enjoying street food at a night market. Huang was warmly received by the local tech community, drawing comparisons to Beatlemania, though the post lacks detailed technical insights or practical implications for engineering work.

Today's top AI news includes NVIDIA threading agents across the stack, Bernie Sanders proposing a public AI stake bill, turning Claude sessions into reusable skills via daily audit, hackers accessing Instagram accounts through Meta AI, and four new AI tools released.

The Cosmos Coalition is a global initiative launched by NVIDIA and leading AI labs to build and open-source frontier world models for physical AI, with Runway joining as a founding member.

NVIDIA Nemotron 3 Ultra achieves leading accuracy in agentic tasks covering productivity, coding, and long-horizon planning. This tweet is a product teaser lacking benchmarks and technical details, offering limited engineering value.

NVIDIA released Nemotron 3 Ultra, an open-source model for long-running agents claiming 5x faster inference and 30% lower costs, yet community feedback suggests it still lacks competitive edge.

NVIDIA's Nemotron 3 Ultra model is now available for free on the OpenRouter platform, enabling developers to integrate and test this high-performance model via API at zero cost.