AI is promising a lot these days, to change how we work, influence how we live, and transform network traffic. It’s interesting, exciting even, but suppose it’s all hype? Wall Street has been increasingly antsy about AI claims, and Nvidia is arguably in the spotlight of their fears, just as it’s in the spotlight of AI. Will our current AI model transform everything, including the financial fortunes of AI giants like Nvidia? It’s a risk, and if you’re facing risk what do you do? Buy insurance…but buy the right kind.

Almost all the current AI investment focuses on cloud-hosted technology, not surprising given that it’s the cloud providers doing the spending. A lot of investors think this AI approach is just hype. Nvidia isn’t publicly repudiating the old model (it makes its money there), but they’re increasingly exploring other AI approaches to ensure that they won’t be caught in a hype-wave collapse. Some are small starts, some are gigantic, and all are important.

The small sign of Nvidia’s “AI insurance policy” is something that many Nvidia GPU users see when they update their driver software. Nvidia is promoting ChatRTX, which lets users run an open-sourced chatbot LLM on their RTX-30 or higher GPU chips. You can link it to your own data and retain complete sovereignty. This has been around for a while, but Nvidia is now making it more visible. It’s a populist alternative to the big cloud-hosted chatbots and also a demonstration of self-hosted AI.

Then there’s real-time computing. Nvidia has been promoting the use of its AI tools to build digital twins, which it calls “world models” for some time. These model physical systems and let applications control real-world, real-time processes. Just this month, they announced a broad initiative to integrate AI and world models with robotics, in every form from mechanical arms and vehicles to humanoid robots.

What’s important about the latest initiatives is the focus on perception, meaning the ability of AI to analyze real-world conditions via sensors and even video, and populate the world model that represents the processes being controlled. Keeping a world model synced with the real world is the most critical single factor in any kind of autonomous behavior; it’s what keeps things from running into each other, facilities, or even people, and it ensures the systems do what’s intended.

This is a big-money opportunity, something that could add up to as big a business case as all the past business cases that have justified IT spending for 70 years, combined. But it’s slow to develop. Less than 20% of enterprises have any current projects to exploit these autonomous-system world models. It’s probably what Nvidia and other AI companies are betting on for 2028.

If you think 2028 is far-out thinking, you’ll love the next thing: quantum computing. Yes, there are systems around today. No, you won’t find enterprises using quantum computing even in specialized missions. Yes, a recent set of reports suggests that quantum computing is not infinitely scalable, as was once believed. Still, the potential for quantum computing is enormous. A quantum system could, in theory, rival a supercomputer in something no bigger than a decent server, but nobody really knows how long it will take to get to that point. So, Nvidia isn’t waiting.

Right now, the only way to test out quantum applications is to simulate them, and GPUs are the accepted optimum platform for that. The Nvidia idea is to first support that simulation (their CUDA-Q platform and cuQuantum library), and then to link quantum systems to GPU servers with low-latency paths (NVQlink and DGX Quantum). They even have an initiative to create a coalition of labs and quantum players, similar to what gave rise to the Internet, in their Nvidia Accelerated Quantum Computing Research Center. All this is available in early-access form via Nvidia’s quantum cloud.

The real-time and quantum initiatives are getting the most attention, but it’s that little personal chatbot that’s the most critical. AI world models and quantum computers are to AI insurance what a health insurance policy that’s effective in five years is to your own financial planning. That little chatbot can bridge the gap and can be real AI insurance and not just another wave of AI hype.

It’s fair to ask why you should care about all of this, if you’re not an Nvidia investor. The answer is that, behind the hype, Nvidia is actually pushing the right AI buttons to help build future business cases. There is real enterprise value being demonstrated—it’s just that the value is going to take a long time to develop, and AI hype sustains interest while the development happens. At least that’s the theory. The challenge is to build interest in the boring aspects of AI business cases when the market’s focus is elsewhere. Can Nvidia do it? The answer to that will determine how far, and how fast, real AI transformation happens. Insurance is great, but staying healthy and alive is better.