Nvidia

On the same day that Meta confirmed its first rogue-agent Sev 1 security incident, NVIDIA dropped the tool that the enterprise world arguably needed yesterday: OpenShell, an open-source secure runtime environment for autonomous AI agents. Released under Apache 2.0 as part of the NVIDIA Agent Toolkit at GTC 2026, OpenShell is designed to put a hard enforcement layer between an agent and everything it shouldn’t touch. The timing is not a coincidence. Enterprise AI teams are deploying agents that can execute code, read file systems, and make network calls — and the governance infrastructure for those capabilities is, in most organizations, significantly behind the capability itself. OpenShell is an attempt to close that gap at the runtime level. ...

The bottleneck for agentic AI at scale has never really been the models — it’s been the infrastructure to run them cost-effectively at production volume. NVIDIA just addressed that directly with Dynamo 1.0, the production release of its open-source inference operating system, announced at GTC on March 16. The headline number: 7x inference speedup on Blackwell GPUs. The more important story is what Dynamo actually does architecturally. Dynamo as an Inference Operating System Jensen Huang’s framing is precise: Dynamo is the “operating system” for AI factories, not just a performance library. Just as a traditional OS orchestrates CPU, memory, and storage for application workloads, Dynamo coordinates GPU and memory resources across a cluster to handle the unpredictable, heterogeneous demands of production AI inference. ...

If you’ve been building AI agents with LangChain and wondering when the “enterprise-grade” piece would arrive, March 16 was your answer. LangChain announced a comprehensive partnership with NVIDIA to deliver what both companies are calling an enterprise-grade agentic AI development platform — combining LangChain’s LangSmith, LangGraph, and Deep Agents frameworks with NVIDIA’s full Agent Toolkit stack. At the same time, LangChain joined the Nemotron Coalition, NVIDIA’s global initiative to advance open frontier models. ...

The moment the agentic AI industry has been building toward arrived at GTC 2026 today: NVIDIA officially launched NemoClaw, an enterprise-grade stack for the OpenClaw agent platform that adds security, privacy controls, and a single-command NVIDIA model installer to the world’s fastest-growing open source project. What Is NemoClaw? NemoClaw is NVIDIA’s purpose-built software stack that plugs directly into OpenClaw — adding what the platform has been missing for enterprise adoption: a security and privacy infrastructure layer beneath every claw (agent). ...

General-purpose CPUs were designed for a world where software executed deterministic instructions in predictable sequences. Agentic AI — where models plan, call tools, run code, validate results, and loop — doesn’t work that way. NVIDIA recognized that gap and, at GTC 2026 today, launched something genuinely new: the Vera CPU, the world’s first processor purpose-built for the age of agentic AI. What Makes Vera Different The Vera CPU isn’t a traditional server processor wearing an AI marketing hat. It’s architected specifically for the workloads that agentic AI actually runs: ...

If you had any doubts that the age of agentic AI has officially arrived, Jensen Huang just removed them. Speaking today at the SAP Center in San Jose before a packed crowd of more than 30,000 attendees, the NVIDIA founder and CEO declared GTC 2026 the moment the industry crossed an irreversible threshold — from AI as a tool to AI as an autonomous agent. What Is GTC 2026? NVIDIA’s GPU Technology Conference is the company’s flagship annual event, bringing together researchers, engineers, and enterprise leaders to see where AI infrastructure is heading next. This year, the event has taken over downtown San Jose entirely, with workshops, hands-on labs, and what NVIDIA is calling “the largest AI infrastructure gathering in history.” ...

NVIDIA just dropped something that’s going to matter for anyone building real agentic AI systems. Nemotron 3 Super is a 120-billion-parameter open-weight model — but here’s the key detail that separates it from the crowd: it only uses 12 billion active parameters at inference time thanks to a hybrid Mamba-Transformer Mixture-of-Experts (MoE) architecture. The result? Five times higher throughput than comparable-sized models, with a one-million-token context window that changes how agents can actually operate in the wild. ...

The AI agent platform race just got a heavyweight contender. Nvidia — the company that makes the chips powering most AI workloads today — is reportedly preparing to launch its own open-source AI agent platform called NemoClaw, designed to compete directly with OpenClaw in the enterprise market. The news, first reported by Wired and confirmed by Ars Technica, Tom’s Hardware, CNBC, and DigiTimes, arrives just days before Nvidia’s annual developer conference in San Jose. ...

“Linux took, right, some 30 years to reach this level. OpenClaw, in three weeks, has now surpassed it.” That quote came from Nvidia CEO Jensen Huang at the Morgan Stanley Technology, Media, and Telecom Conference on March 4. It didn’t get the headlines it deserved at the time — but today’s Meta-Moltbook acquisition makes it impossible to ignore. A Software Milestone Unlike Any Before It OpenClaw is now the single most downloaded open-source software project in history. That’s not a projection or a trend line — it’s where the download curve actually landed, surpassing Linux’s cumulative install base in under a month. ...

Nvidia has quietly been pitching something significant to enterprise software companies ahead of its GTC 2026 conference next week: an open-source AI agent platform called NemoClaw that lets companies dispatch AI agents across their workforce — regardless of whether those products run on Nvidia hardware. That last part is the headline. Nvidia building hardware-agnostic software is a deliberate strategic move, and it signals that the GPU giant is serious about owning the enterprise agent software layer, not just the silicon underneath it. ...