The robotics industry has been promising us a future that keeps failing to arrive. We're still waiting for machines that can move seamlessly through our homes, our offices, and our streets, and the gap between the demo reel and the daily reality is a chasm. Nvidia's Les Karpas is set to explain why at TechCrunch Disrupt 2026, and his answer will likely be less about hardware failures and more about a fundamental misunderstanding of how we build and train these systems. For anyone who has ever felt stuck with a spreadsheet that can't think, this is the same story: the tools are not the bottleneck, the approach is.
Our take is that Karpas's appearance is a signal, not a sales pitch. Nvidia has become the quiet engine behind the AI boom, and when they send a leader to talk about robotics, they are not asking you to dream. They are asking you to pay attention to the shift from scripted automation to adaptive intelligence. The practical implication for you is direct: if you are a developer, a founder, or a data analyst, the same principles that will make robots useful in the physical world are already changing how we interact with digital tools. You do not need to wait for a humanoid to arrive. You need to start exploring how large language models and vision systems can be applied to your own workflows, whether that is cleaning up messy datasets or automating repetitive reporting. The future is not a product you buy; it is a skill you build.
What would we tell a reader who asks whether this is worth their time? Register before September 25 to save up to $200, but do not go for the discount. Go because the conversation around robotics has been stuck in a loop of "one more year" for a decade, and Karpas's answer to why we are still waiting will give you a framework for evaluating every other AI tool you encounter. This is not about hype. It is about diagnosis. If you understand why robots fail to leave the lab, you will understand why your own automation projects stall. The same reason applies: we train for perfection in controlled environments, then wonder why the real world breaks us. Expect Karpas to push back on that mindset, and we would bet on him arguing for more data, more simulation, and more patience with imperfection.
The specific detail to watch is not the technology itself but the timing. Why does Nvidia choose to make this argument now, at a conference like Disrupt, rather than through a quiet technical paper? Because the market is ready to listen, but only just. The takeaway you can quote is this: "The breakthrough will not come from a better robot, but from a better understanding of how we teach it." If Karpas delivers on that idea, the next year will be spent less on building new hardware and more on rethinking the data pipelines that feed it. For you, that means the competitive advantage is not in owning the robot. It is in owning the question of what you want it to learn.
