Jensen Huang gave his GTC keynote yesterday. Three hours, leather jacket, trillion-dollar demand projections, a Disney robot. Classic Jenson.
Most of the coverage is about chips, data centers, DLSS5, and robots. That is not why I am writing this.
Buried in the middle of a hardware keynote, Huang made a statement that applies to every restaurant operator whether they know it or not. He said every company in the world needs an OpenClaw strategy. He compared it to needing an internet strategy in the late nineties, a Linux strategy in the early 2000s, a mobile strategy ten years ago. He called OpenClaw an operating system, just like what you are reading this on or windows on a PC.
For those who have not been tracking it, OpenClaw is an open-source framework that lets AI agents run on your own hardware. It is like hiring as many bussers, dishwashers, and cooks as you want each acting as an agent. An agent that can navigate files, execute tasks, call external tools, send messages, and work overnight without supervision. It launched in January. It became one of the fastest-growing open-source projects in history within weeks.
Nvidia built NemoClaw on top of it because it realizes how important this is, and it needed security. The pitch: run autonomous AI agents inside your business without your data leaving the building.
Why This Matters If You Run a Restaurant
Huang framed this as the shift from SaaS to agent-delivered services. Instead of logging into a dashboard and pulling a report, an agent pulls the report, reads it, identifies the problem, and tells you what to do about it. Or...wait for it...just does it.
I have built a boutique food cost system. That is a daily email that surfaces the number and the trend without anyone opening a spreadsheet. The difference is I duct-taped it together with Zapier and Twilio in true southern fashion. What Nvidia is describing is the infrastructure to make that kind of thing normal. Standard. Expected.
What did Jack Dorsey @ Square do just a few weeks ago? He cut 4000 people, many I knew personally. Why? He saw this coming, a place where the front line restaurant operator is talking directly to the technology or the person writing the code to transform their business.
The restaurant industry is not going to adopt OpenClaw tomorrow. Most are too busy trying to figure out why the kitchen printer isn't working or orders aren't coming through. Most operators have never heard the word. But the pattern Huang described, software that watches, reasons, and acts on your behalf instead of waiting for you to log in, that pattern is already arriving in every POS and restaurant platform. Square is already on this trajectory with their AI development for sellers. Toast is building toward it. Every platform will have to.
The Part You Don't Hear
The thing Huang did not say, because his audience was data center operators and enterprise CIOs, is that this shift lands hardest on small businesses. Not because small businesses will adopt agentic AI first. Because the gap between operators who have any AI integration and operators who have none is about to become a gap between operators who have a manager that never sleeps and operators who are still checking the bank account at the end of the month hoping payroll goes through.
The operators who started using AI tools two years ago are going to be running circles around the ones who waited. Not because they are smarter. Because their systems see things earlier and surface decisions faster.
Huang compared OpenClaw to Linux. When Linux showed up, most small business owners had no idea what it was. Twenty years later, every server their website runs on uses it. Android is built out of it. The same thing is happening here. The infrastructure is being built right now. The question is whether you are going to build on top of it or get built over.
What To Actually Do
You do not need to install OpenClaw. You do not need to buy an Nvidia GPU. You do not need to understand what NemoClaw is, yet.
What you need to do is start paying attention to the AI features inside the tools you already use. Your POS has them or is building them. Your accounting software is adding them. Your scheduling platform is integrating them. The question is not whether to adopt AI. The question is whether you are going to use what is already in front of you.
The food cost system I built with free tools and an afternoon of work dropped costs over five percent in two months. Not because the AI was sophisticated. Because it made a number visible that had been invisible.
That is the floor. That is the starting point. And the ceiling Jensen Huang described yesterday is agents that run your back office while you run service.
The leather jacket is optional.