I Called Nvidia a Bubble Ready to Burst. I Was Reading the Wrong Story.

For the better part of a year, my perspective on Nvidia was cemented in skepticism. I confess, I was one of them. One of the voices in the chorus of caution, one of the writers who looked at the dizzying ascent of its stock chart and felt a familiar, chilling echo of history. The narrative, as I saw it, was dangerously simple and deeply compelling. I would read the headlines—and I believed them because they confirmed my bias. The most potent of these was the drumbeat of insider selling. "Nvidia Insiders, Including CEO, Cash Out Over $1 Billion," the reports blared. How could anyone see that and not feel a pang of doubt? The very architects of this supposed revolution were taking money off the table at a historic clip. To me, it was an open-and-shut case: a stunning lack of faith from the people who knew the most.

My second pillar of disbelief was the ghost of Cisco Systems. The analogy was just too perfect, too neat to ignore. A company providing the essential 'picks and shovels' for a world-changing technological shift. A parabolic stock rise fueled by a narrative of limitless growth. A valuation that seemed to defy financial gravity. I saw the pattern, and I was convinced I knew how it ended—not with a gentle plateau, but with a catastrophic collapse, a dot-com-style reckoning that would wipe out fortunes and serve as a cautionary tale for a generation. I saw Nvidia not as a pioneer, but as a symptom of market mania, and I felt it was my duty to point out the emperor’s new, astronomically expensive clothes.

I held these beliefs firmly. I argued them with colleagues. I saw the market’s unbridled optimism as irrational exuberance, and the counter-arguments as hype. My change of mind wasn't sudden, but it began with a single, nagging thought that I couldn't shake. The catalyst wasn’t a press release or a rosy analyst report. It was a term I had initially dismissed as corporate jargon: 'Sovereign AI.'

I first heard it and rolled my eyes. It sounded like a clever marketing phrase designed to invent a new customer base just as fears were mounting that Big Tech’s spending on AI infrastructure might slow down. But during a late-night research session, I stumbled upon a report not about finance, but about geopolitics and national industrial strategy. It detailed how mid-sized nations—from the EU to the Middle East to Southeast Asia—were becoming terrified of being left behind, of having their entire digital future dependent on a handful of American tech giants. They didn't just want to use AI; they needed to own their AI. They needed to train models on their own languages, with their own cultural data, and within their own secure data centers.

Suddenly, 'Sovereign AI' wasn't a buzzword. It was a new, multi-trillion-dollar global arms race, and Nvidia wasn't just selling chips; it was selling national ambition. This realization forced me to re-examine everything I thought I knew, starting with that damning Cisco comparison.

My belief that Nvidia was Cisco 2.0 was predicated on the idea that their market was finite. Cisco sold the plumbing for the internet to corporations. Once the offices and data centers were connected, the explosive growth phase had to end, and competition would commoditize the hardware. I was applying that same template. But Sovereign AI revealed a fatal flaw in my analogy. Nvidia isn't just selling plumbing to corporations. It's selling the foundational engine of 21st-century economic and military power to nations. It's a parallel market, a completely new growth vector that exists alongside the demand from Silicon Valley. The customer isn't just Microsoft or Meta; it's the entire government of France, the sovereign wealth fund of the UAE, the national research labs of Japan. I had been comparing a company that built the roads to a company that is selling the unique, proprietary engine for every car, truck, ship, and airplane that will ever use them. The scale is fundamentally, paradigm-shiftingly different.

What’s more, Nvidia’s product isn’t just a piece of hardware that can be easily replicated. The true defensible moat is its software ecosystem, CUDA. This intricate web of software, developer libraries, and programming models has been built and refined over 15 years. It is the operating system for accelerated computing, and the entire world of AI research and development is built on top of it. A competitor can’t just build a faster chip; they would have to convince millions of developers to abandon the ecosystem they know and rebuild their entire life's work on a new platform. Cisco never had that kind of lock-in. Their hardware was brilliant, but it was ultimately interchangeable. Nvidia’s platform is not.

With this new perspective, I was forced to revisit the insider selling. The headline figure—over a billion dollars—is designed for shock value. But it’s a number without context. I started looking not at what was sold, but at what was kept. In nearly every case, the stock sold represented a small fraction of the executives' total holdings. Many of these individuals are long-time employees, engineers who have been with the company for decades, whose net worth has exploded in ways they could have never imagined. What I had interpreted as a 'lack of faith' was, in reality, prudent financial planning and diversification on an almost unimaginable scale. When your company stock, through its own success, becomes 99% of your personal wealth, selling 5% isn't an act of betrayal; it's an act of responsibility. They aren't betting against the company; they are simply cashing a fraction of a winning lottery ticket so massive it's hard to comprehend, while still leaving the vast majority of their fortunes tied to its future.

I was wrong about Nvidia because I was looking for patterns from the past in a future that doesn't rhyme. I was so focused on the headline-grabbing stock price and the dot-com playbook that I missed the deeper, more complex story unfolding beneath the surface: a story of technological moats, software ecosystems, and the birth of a new global market for national intelligence. My certainty has been replaced by a hard-won humility. It’s a humbling experience to realize you’ve been reading the map upside down, but it’s a necessary one. And it has forced me to ask a different question: what if this isn't a bubble? What if it's the beginning of an infrastructure build-out unlike any we have ever seen before?