I went to NVIDIA GTC

GTC is over. Now the real work begins.

I came into the week expecting the usual Jensen cadence — new hardware, big numbers, a leather jacket and a vision. What I didn’t expect was how much of a three-hour keynote would be dedicated to an open-source project built by a single developer.

OpenClaw isn’t just a clever piece of software. It’s the clearest signal yet that the agentic era finally has something to build on — one that didn’t come from a hyperscaler, a foundation model lab, or require a hundred-million-dollar training run. It came from someone who saw what was missing and built something people actually wanted.

Jensen’s comparison to Linux is getting a lot of attention. This is the wrong frame. Linux was built for a narrow technical audience — DevOps, IT admins, the people who lived in the terminal. OpenClaw is designed for everyone. That’s a fundamentally different bet, and it changes everything about what governance, deployment, and enterprise adoption must look like.

And that’s exactly where financial services enters the conversation.

For regulated institutions, the question was never whether agentic AI would be compelling. Of course it is. The question is whether it can be deployed inside the constraints that actually govern these organizations — data security, model transparency, audit trails, and access controls. NemoClaw is NVIDIA’s first answer to that question. It won’t be the last.

What comes next is a race to build the governance scaffolding that makes enterprise agentic deployment real — policy engines that travel with the agent, not bolted on after the fact. For capital markets specifically, that means agents that can operate across trading, risk, and compliance workflows without creating new regulatory exposure.

The firms that move now to define those boundaries — what an agent can see, what it can act on, what it must escalate — will have a meaningful head start. The infrastructure is here. The frameworks are forming. The question is whether your organization is ready to meet them.

The BIG takeaways: Nvidia’s Vera Rubin delivers exponentially higher throughput at a fraction of the power. A credible path to securing agentic AI at enterprise scale. An innovation runway that would have seemed like science fiction eighteen months ago — now within reach. The pieces are aligned in a way they haven’t been before.