Transcription
Nvidia Nema changed OpenClaw forever. Here's the thing that nobody's talking about. One developer, one weekend, one side project, and in less than 60 days, it became the fastest growing piece of software in the history of the internet. Right? That is open claw.
And then 3 weeks later on March 16th, 2026, Jensen Huang, the CEO of Nvidia, walked out on stage in front of 30,000 people and said something that should have stopped every business owner in the world dead in their tracks. He said, "For the CEOs, the question is, what's your open claw strategy?" And then Nvidia dropped Nema claw, and that's what we're talking about today. Because what happened at GTC 2026 did not just change open claw, it changed the game for every single person watching this, whether you're a freelancer, business owner, manager, or someone who just wants to understand what is going on in AI right now.
Right? So, the difference between open and your typical AI is like a consultant and an employee. A consultant tells you what to do. An employee actually does it. And open core is the employee. It runs on your own machine, on your own hardware. But the problem is that open has a massive problem, right? A massive, dangerous, company-destroying problem. And until last week, nobody had really solved it. Let me explain.
Imagine you hire a new employee, smart, fast, hardworking, does everything you ask. You hand them the keys to the office, the filing cabinets, the email, the customer data, the bank accounts. You say, "Go ahead, handle everything." And then you find out that the new employee had absolutely no background check, no ID, no references, no way to know if they're going to help you or rob you, right? And that is what OpenClaw was for businesses, right? It was built for individual users, but not for companies, right? Um, the difference here is that this had a huge gap in security. So you have this incredible, world-changing tool that can automate almost anything, but businesses can't really touch it because it has massive security gaps. And that is the gap that Nvidia just stepped in to fill.
So, what they built is Nema Claw. It was announced at GCC 2026 on March 16th by Jensen Tuang, the CEO of Nvidia himself. And Neoclaw is a software stack that integrates directly with OpenClore and installs in a single command. Along with it came Nvidia OpenShell, an open-source security runtime designed to give autonomous AI agents the guardrails they need to operate inside real enterprise environments.
Now, let me translate that into plain English. Open floor is a P car. It's powerful. It goes fast. But it had no seat belts, no brakes, no airbags, and the steering wheel was a little loose. Necclaw is what you get when you take that car to the factory, install all the safety systems, the seat belts, the brakes, the airbags, the GPS that make sure the car does not drive somewhere it's not supposed to go with one command. And then you go from a powerful but dangerous tool to a powerful and safe one. Necore installs Nvidia Netron models and the newly announced Nvidia Open Shell runtime in a single command. So you can just install it in one click. Adds privacy. Adds security controls to make self-evolving autonomous AI agents more trustworthy, scalable, and accessible to the world.
Jensen Hang said something on stage that I keep coming back to. He said, "Look, for the CEOs, the question is, what's your open core strategy? We need it. We all have a Linux strategy. We all needed to have an HTTP HTML strategy, which started the internet." Right? Read that again. He's not comparing Open Core to another chatbot. He's comparing it to the internet. He's comparing it to Linux, the operating system that runs most of the world's servers. He's comparing it to HTML, the language that made the web possible. Those are not small comparisons. Those are the kind of comparisons when someone makes them, you either roll your eyes or you wake up. I'm not rolling my eyes. And I do not think the 30,000 people in that room were rolling their eyes either.
So, let me dig into what Neoclaw actually does, because the security layer is only part of the story. The first thing to understand is Open Shell. Open shell contains several security layers, including kernel-level sandboxing and a privacy router that monitors open claw's behavior and communication with our systems. Think about a bank vault, right? A regular open court agent is like a person who has a key to everything in the building, the main doors, the office, the filing room, the safe, everything. Open shell is what puts different locks on different doors and hands the agent only the specific keys it needs for a specific job. The agent can still work. It can still get stuff done, but it can't go somewhere it's not supposed to.
Nemlaw also installs Nvidia Nemetron open models locally on whatever dedicated hardware is available. So a privy router then allows agents to reach cloud-based frontier models when needed, whilst keeping the guardrails in place. The combination is designed to let AI agents develop and learn new skills without ever stepping outside defined boundaries. And this is huge for business, because the old fear was, if my AI agent is connected to cloud models, is my private data going to cloud servers? Is it going to end up in someone else's training data? And Neoclaw solves that. Your agent can use powerful cloud AI when it needs to, but it runs through a privacy router that keeps your sensitive stuff locked down before anything leaves your building. So you got local power plus cloud intelligence, and so the data stays yours.
Now, let me talk about who's already in on this, because when I tell you the partner list of Nemoclaw, I want you to really hear these names. We're talking about Adobe, Salesforce, SAP, Crowdstrike, and Dell, among the launch partners integrating Neoclaw into their enterprise workflows. Companies including Cisco, Crowdstrike, Red Hat, Seaman, Synopsis, Cadence, and IQVIA are building expanding agent systems with the toolkit across business software, security, chip design, and life sciences.
So, let me put a bow on something here, because the detail is wild. Cisco and Crowdstrike, the same companies that issued security advisories calling opencore a significant security risk, are now partnering with Nvidia on the enterprise version. That's not a contradiction, that's a validation. And Cisco and Crowdstrike spent months looking at open core from the outside and saying, "This thing is dangerous." And then Nvidia built the safety layer, and those same companies turned around and said, "Now we are in." This is what validation looks like, when the people who are the loudest critics become the earliest adopters. And then there is Dell, right? Dell was the first hardware partner to ship the GB300 desktop with both Neml and Open Shell pre-installed for enterprise buyers. That means you'll be able to buy a computer that already has all of this built out of the box, day one, no setup required. Just turn it on, and your enterprise-created AI agent is ready to go. That's a very different world from where we were 6 months ago.
Now, I want to talk about something that gets lost in all the technical coverage of this story. Because if you're sitting there thinking, "Well, you know, enterprise companies, they're big. Like, that doesn't really involve me or my life." Well, here's why you should care. Because Neml didn't just change things for big corporations. It changes things for everyone. Before Nemlaw, OpenClaw was powerful, but genuinely risky. If you were a small business or a solo operator, you could try to use it, but the security issues meant you were taking a real chance every time you connected it to your real business data. After Nemlaugh, those risks go away, or at least they go way, way down. Now you can run an AI agent that actually does your work inside a secure sandbox, using whatever AI models you want, local or cloud, without giving up your privacy or your data or your safety.
Think about what that means practically. Nemo claw is tailored for enterprise use cases like client onboarding, invoice extraction, contract management, and RFP sourcing. But strip out the corporate language, and what are we really talking about? Well, we're talking about an AI agent that can welcome new clients for you. Pull data from invoices so you don't have to read through contracts and flag the parts that actually matter. Go through long proposal documents and pull out the key info. Every single one of those tasks is something a person is doing manually right now. Probably you or someone on your team. And every single one of those tasks is something a Nemo claw agent can do instead. Faster, without breaks, without mistakes caused by tirelessness, and also around the clock, 24/7.
So, I don't say this to scare you. I say it because the data is very, very clear about what happens to the people who learn this stuff early. Organizations that scope and deploy Agentic AI against well-prepared operational foundations report average ROI of 171%, approximately three times higher than traditional automation. So, 171% return on investment for AI automation. Three times higher than traditional automation tools. It's not a projection. That's what's already happening at companies that are ahead of the curve.
Now, let me talk about what makes this different from other AI announcements we've seen in the last 2 years, because you're probably tired of hearing of all of these different updates. Well, every major technology shift has a moment where it goes from being a thing that has to play with to being a thing that real businesses build on. And today is that line. It's enthusiast adoption versus enterprise adoption. Linux was loved by programmers for years before businesses started running their servers on it. The internet existed in universities for decades before companies started selling things on it. The moment Nemlaw launched, with Dell shipping hardware preloaded with it, with Salesforce integrating it into agent force, with Cisco using it for network operations, with Crowd Strike building on top of it. OpenClaw crossed that line. It went from enthusiast to enterprise infrastructure in a single announcement. Gartner projects that 40% of enterprise applications will embed task-specific AI agents by the end of 2026, up from less than 5% in 2025. So it's not a gradual shift. This is a cliff, and we are standing right at the edge of it, right? And Neoc is a reason that jump can happen, because without the security layer, enterprises couldn't trust this stuff. It, the, the gates are open. And the global agentic AI market is tracking from $7.6 billion in 2025 toward a projected $199 billion by 2034. A 43% compound annual growth rate from $7.6 billion to $199 billion in less than a decade. So, if you want a reference point for what kind of growth looks like in practice when it comes to these things, think about smartphones. 2007, smartphones were a novelty. By 2012, they were everywhere. By 2017, the whole world had changed because of them. And that kind of shift happened in 10 years. We are at the 2007 moment for AI agents, and Nemla just gave us the iPhone moment.
Let me stop here and address something, because I can hear the skeptical voice in the back of your head saying, "Come on, every AI announcement is called the iPhone moment, right? Every new tool is going to change everything. How is this different?" And it's a fair point. It deserves a real answer. So here's what's different. The barrier for the last two years of AI tools has been trust, not capability. Trust. So, Chat GBT was capable in late 2020. Um, Claude was also capable. Today, these models can do extraordinary things. But every time you use a cloud AI tool, you're handing your data to someone else, right? And for most legal companies, healthcare companies, law, regulated industries, etc., that's a line they can't cross, right? Especially with Neoclaw. Um, but now that Neoclaw is released, and it's sort of created a more secure version of OpenClaw, this changes the trust equation. You can have an AI agent that does things inside your building, on your own hardware, without your data going anywhere, because it runs on local models. So, it's capability plus trust that equals adoption. And that's the formula. And we just got both sides of the equation.
Now, I want to talk about what this means for you. Depending on who you are, you know, if you're a business owner, within the next 12 months, the businesses in your space that are running AI agents are going to be able to do things at constant speeds that you simply cannot match manually. The businesses that figure this out early probably not going to fire everyone, but they are going to redeploy those people onto higher-value work, and that's where you get the most leverage. If you're a freelancer, well, this stuff is going to automate your creative work and chasing up clients and that sort of thing. And I see this already, the gap between AI-native freelancers and traditional ones is widening every month. A year from now, that gap is going to be a canyon. So, if you're a manager or a leader in a company, Neclo is going to start showing up in your IT conversations, your vendor conversations, your strategy conversations. Jensen Gran just told every CEO in the world they need an open port strategy. Your board is going to hear that. Your CFO is going to hear that. You need to know what this is and have an opinion on it before someone else in the room does. And if you're just trying to keep up with where AI is going so you do not get blindsided, here's the bottom line. The era of AI as a chatbot is ending. The era of AI as an agent, as something that actually does things in the world, is beginning. And Nemlaw is the moment that era became safe for the mainstream.
I know a lot of people in our community, the AI profit boarding, who were excited about open claw but nervous about the security issues. That conversation changes now. The security excuse is gone. The adoption excuse is gone. The "I will wait until it's safe" moment is here, right? And if you've been on the fence about learning this stuff, about really learning how AI agents work and how to use them to make money, the Neoclaw announcement is your signal. Not next year, not when it gets better, now. And the people who learn this in the next 6 months are going to have a massive head start over everyone who waits, because Nemlaw is bringing a flood of enterprise adoption. That means new tools, new clients, new integrations, new use cases, new opportunities. And the people who already understand this stuff, work, how it works, etc., are the ones who are going to be able to move fast when those opportunities arise. So, come join us at the AI profit boardroom. That's what we do there. We take this stuff, all the confusing technical stuff, all the big corporate announcements, we translate it into practical steps for real people, business owners, freelancers, creators, consultants, people who want to use AI to actually make more money and build something real, not just read about it. So, link in the comments description where you can go to the AI profit boardroom to get access.
Now, let me talk about the technical architecture of model will, because I think there's a detail here that most coverage has glossed over, and it's really important. At its core, privacy router lets AI agents um use frontier models running in the cloud alongside local models. So, it strips personally identifiable information before sending data to external services using differential privacy technology. That's a mouthful. Let me make it simple. You're running an AI agent that's doing research for you, right? It needs to ask a powerful cloud model a question, but the question contains private client information. The old way of doing this was, you either do not use the cloud model at all and miss out on its power, or you send the whole question to the cloud and your private data goes with it. The Neoclaw privacy router sits in the middle. It takes your question, strips out all the private info, sends the cleaned-up version to the cloud model, gets the answer back, and puts the context back in. The cloud model never sees your private data. You get the full power of the best AI models in the world without giving up any sensitive information. That's genuinely clever. It's the kind of thing that makes enterprise adoption not just possible, but confident.
Let's also talk about the hardware side of this, because Nvidia didn't just drop a software stack at GTC. The DGX station is built on the same GB300 superchip architecture as Nvidia's data system. Right? It packs 748 GB of coherent memory, up to 20 petaflops of FP4 performance, and supports up to one trillion parameters in the desk-side form factor. Translation in simple English: Nvidia is selling a computer that sits next to your desk and is as powerful as the kind of hardware that used to be in a server room the size of a warehouse. And that computer ships with Neoclaw and Open Shell pre-installed. So, we're at the point where a single person sitting at a desk with a relatively normal-looking computer next to them can run enterprise-grade AI agent infrastructure locally, privately, and independently. And a year ago, that sentence would have sounded like science fiction. Now, it's a product you can actually order. And that's the acceleration we're living through.
So, this is this is crazy stuff. I mean, Open it really just launched in February and it really got, you know, viral around the last few weeks and around early March, and now on March 16th, 2026, Nvidia drops a Nemo claw, right? It's the world's most powerful technology company, um, announcing that every company in the world needs an open claw strategy, that builds the security layer that makes enterprise adoption possible, and then you've got people like Microsoft and Google coming along and starting to use it. So, that whole sequence from side project to global enterprise infrastructure in less than 4 months. Let that sink in, 'cause that's the speed of the world we're living in now. The timeline between interesting idea and foundational infrastructure used to be decades. Now it's months. If you're not paying a close attention, if you're not proactively learning and adapting, you can miss an entire technological shift in the time it takes to plan a vacation. That is not to make you anxious. That is to make you focus. The opportunity inside this shift is enormous, but it will not wait.
Let me talk about the economic implications of this as well, because I think this is where it gets real for some people. Right? Companies using AI agents are capable of autonomous action and can have automated up to 95% of their accounts receivable work compared to 38% at firms without this capability. Right? This is a process of like sending emails, tracking invoices, tracking payments, following up on late invoices, reconciling records, is one of the most time-consuming admin tasks in any business. And with something like Nemore, you could automate that like 95%. A company's already using, uh, capable AI agents. So, if you have a team member or a part of your day that you spend on that kind of work, that's what's on the table right now, not some theoretical future. In fact, 43% of CFOs expect a significant impact from AI agents that can automatically shift budget dollars based on real-time spending data, with another 47% expecting some impact, too. That means 90% of CFOs, the people in charge of money at major companies, expect AI agents to significantly change how their finances are managed. 90%. It's not a fringe idea. This is like, just mainstream view. The question I keep coming back to is, if 90% of CFOs think agents are going to transform finance, and Gartner thinks 40% of enterprise apps will have agents by end of 2026, and Nvidia just made it safe and easy to employ them, what does your industry or my industry look like in 18 months? Not dramatically different in broad strokes, but different in the details, different in the parts that currently require human hours, right? The reports that take someone three hours to compile can be automated. The client follow-ups that fall through the cracks should be automated, too. The invoice reconciliation at the end of the month should be automated. None of this is going to like destroy jobs overnight, but it changes what these jobs look like. And I've said it before and I'll say it again. This is not doom. This is a skills gap, right? And skills gaps have always been opportunities for the people willing to fill them early.
Now, let me be honest about where Namocllor is today, because I want to not oversell this. I want to really tell you where we're at, right? Nvidia is describing Nelor as an early-stage alpha release. They want developers to expect rough edges and noted the release focuses on environment setup rather than production-ready deployment. So, it's not fully cooked yet. There will be bugs. There will be things that don't work the way they're supposed to. Enterprises with very specific compliance requirements are going to need more work to do, um, before they can fully deploy this. That's the honest part. But Nemoclaw is a massive step forward. It's not a finished product, and the government's question, you know, how do you make sure an AI agent can learn new skills and take new actions is always and always doing what you want to. That question doesn't have a complete answer yet. But I always want to give you the real picture, not just the exciting headline. Neoclaw is genuinely important, is also genuinely early. And both of those things can be true at the same time. So, what matters right now is not whether you're going to deploy Nemco on your business tomorrow. What matters is, do you understand this? You understand where it's going? What kind of position you want to be in when it's production-ready in 6 months from now. And the people who understand it now, the ones who will be ready to move faster later. The people who ignore it now are the ones who will be playing playing a catch-up, right? They're, you know.
So, the other thing that was noted was Nvidia's VP of Generative AI Software, Carrie Briskey, described claws as a new application layer for AI and said they are driving orders of magnitude more demand for compute. And that's a very specific face, because Nvidia makes chips, right? The whole business is selling hardware that runs AI workloads. And what they're saying right here is that AI agents, claws, create dramatically more demand for compute than chatbots do. And a chatbot answers a question and goes quiet until you ask the next one. And the agent is just running 24/7. It's constantly on. So, that is why Nvidia cares so deeply about OpenClaw. It's not true altruism, it's business. The more widely agents are deployed, the more hardware companies need to buy to run them. And Nvidia sells the hardware. Nvidia now sees at least $1 trillion in purchase orders for its Blackwell and V Rubin platforms through 2027. Double the $500 billion figure Wang quoted at GTC last year. So they've gone from $500 billion to $1 trillion in projected orders in one single year. And a big part of that reason that number doubled is AI agents, because the demand for compute that agents create is enormous. And Necloro is Nvidia's bet that they're the company that builds the infrastructure those agents run on. And this is what smart companies do at moments of transition, right? They build the infrastructure layer. They make themselves indispensable to whatever comes. Nvidia builds the chips that trained all the AI models. Now they're building the software stack that runs all the agents. And if they succeed, and the early signs suggest they probably are, then they will be central to the Egyptic era as they have been to the model of training, right? That matters for everyone watching this, because when Nvidia bets this hard on a platform, it tends to become the default. And building skills around the default platform is always a good investment.
So, let's wrap up. In the next 3 to 6 months, I expect Nemaclaw to move from alpha to beta to something approaching production readiness. The Dell hardware is already shipping pre-loaded with it. The enterprise partnerships for Salesforce and Cisco and Crowdstrike are expected to go live in Q2 and Q3 of 2026. The integration work is happening now. In the next 12 months, I think we'll see the first wave of businesses reporting real, meaningful productivity numbers from Neoclaw deployments, not like internal projections, actual results, case studies, specific dollar figures. And this is when the second wave of adoption happens. When early movers publish their numbers, and everyone else says, "Wait, that company cut their admin time by 60% using AI agents. Why aren't we doing that?" Second wave is when the job conversation gets louder. Not because agents are replacing everyone at once, but because the businesses running AI agents are able to operate leaner and faster than the ones that don't. And in competitive markets, leaner and faster usually wins.
Now, in the next 24 months, I think we'll look back at March 2026 and GCC 266 as the moment the agent era officially started. Not in research labs, but in actual enterprise companies at scale. And the people who are paying attention right now, who are learning the tools, who are building the skills, who are positioning themselves in front of this wave instead of behind it, those people are going to be in a very different place than the people who dismissed it. And that's the pattern every single time. You know, look at computers, the internet, smartphones, social media. Every single time the people jumped in early, who figured out what the thing was before everyone else did, they got the best opportunities. The doors were widest open before everyone else showed up. And Neml just opened a very big door. So, if you haven't already, I would check out Nema Claw. Take a look at it. Have a look at it. Um, see what things in your business you could actually automate. And if you want to go deeper, if you want to understand not just what this is, and how to actually use it and build with it and make money with it, that's exactly what we do in the AI profit boardroom. So, you can learn real strategies, get real roadmaps, learn step-by-step, get weekly automation coaching calls, and learn exactly how to implement this inside your business. It's available, link in the comments description, or just go to the AI profit boardroom.
Peter Steinberger built Openfloor over a weekend, a side project, a to hack, something he was doing just to see if he could, and it became the fastest growing piece of software in the history of the internet. Jensen Huang, who runs the most valuable tech company on the planet, soon in front of 30,000 people and said, "It is the operating system for personal AI." Then Nvidia built an entire enterprise security stack on top of it. Dell was shipping computers pre-loaded with it. Salesforce and Cisco and Crowdstrike and Adobe and SAP are building on top of it. All of that in just a few months. We are not watching a trend. We're watching a transition. The kind of transition that rewrites industries. The kind of transition that creates new winners and new losers. The kind of transition that in 10 years people will talk about the way they talk about the early internet. And the question is whether you're going to be there when it arrives. I want you to be ready. That's why I made this podcast. That's why I built the AI profit boardroom, because the people who understand what's happening right now, before it is everywhere, before it's obvious, those are the people who end up on the right side of what comes next. Stay curious, stay learning, and I'll see you in the next.