I Was Convinced Nvidia's Reign Was Ending. Here's What Changed My Mind.

For the longest time, my perspective on Nvidia was crystal clear, and it was deeply skeptical. You’ve probably read the articles because I’ve certainly written my share of them, or at least echoed their sentiments in my columns and private conversations. My thesis was built on two pillars that seemed, to me, unshakable. First, I saw the headlines about premier customers like OpenAI turning to Google's TPUs and I saw the beginning of the end. It was the classic innovator's dilemma in reverse: the king becomes so expensive that the court finds a cheaper way to live. I was convinced Nvidia’s pricing power was a mirage, its market lock-in an illusion about to be shattered.

Second, I looked at the stock chart and saw not a triumph of innovation, but a textbook bubble. Every financial instinct screamed that the valuation was unsustainable. I pointed to the endless parade of articles from outlets like The Motley Fool and the starkly bearish warnings from Seeking Alpha, particularly the one ominously titled 'The Music Is About To Stop.' For me, the smoking gun was the insider selling. When a billionaire hedge fund manager like Philippe Laffont, someone rightly dubbed 'smart money,' unloads over 1.4 million shares, you don’t need to be a market wizard to see the writing on the wall. I saw it as a five-alarm fire, a clear signal that those in the know were cashing in their chips before the inevitable collapse. I was certain, and I was vocal.

My conviction was so strong that I began outlining what I believed would be a definitive piece—a deep-dive into the fragility of Nvidia's empire. This was the catalyst for my change of heart, though not in the way I expected. The research process, intended to gather ammunition for my bearish thesis, instead forced me into a state of profound cognitive dissonance. The turning point wasn't a flashy press conference or a glossy annual report. It was a late-night phone call with an old colleague, now the CTO at a mid-sized AI startup—exactly the kind of company that should be feeling the sting of Nvidia's prices most acutely.

I was probing him for confirmation, expecting to hear complaints about GPU costs and plans to migrate to a competitor's hardware. Instead, he laughed. “You’re thinking about it all wrong,” he told me. “We don’t just buy a chip. We buy an ecosystem. We buy speed.” He went on to explain something the headlines about OpenAI had completely missed.

One of the central pillars of my argument had been that OpenAI’s use of Google TPUs was the crack in the dam. If the leader in the field was diversifying to save money, it was only a matter of time before everyone else followed suit. But my CTO friend laid out a different truth. “Sure,” he said, “if you have a massive, perfectly defined, and mature model that you just need to run at scale, a specialized chip like a TPU can be more cost-effective. OpenAI is optimizing a specific, high-volume workload. But that’s not what the rest of us are doing. We’re in a knife fight of innovation.”

He explained that for developing new models, for experimenting, for the frantic race to the next breakthrough, there is no substitute for Nvidia’s CUDA platform. It’s not about the silicon alone; it’s about the decades of software, the libraries, the developer tools, and the massive community. Switching would mean retraining their entire engineering team, rewriting their codebase from scratch, and sacrificing the agility that is their only competitive advantage. “The cost of the GPU is a line item,” he concluded. “The cost of falling behind is extinction.”

It was a difficult realization. What I had perceived as a mortal threat—a key customer turning to a competitor—was, from another perspective, a sign of a maturing market segment, not a wholesale rebellion. The headlines were focusing on the single tree while I, and many others, were missing the forest. Nvidia wasn't selling a commodity; it was selling an entire platform for innovation, and its lock-in wasn't based on a lack of alternatives, but on the sheer productivity it enabled.

With my first pillar crumbling, I turned a more critical eye to my second: the overvaluation and the insider selling. The story of Philippe Laffont selling 1.4 million shares had been my Exhibit A. It was the perfect, simple narrative: the smart money is getting out while the getting is good. But armed with my new perspective, I forced myself to ask a different question: Is it a sign of panic, or is it prudent portfolio management?

If you owned a stock that had grown thousands of percent, representing an ever-larger portion of your fund, wouldn't you rebalance? Selling a fraction of a holding after such a historic run isn't a vote of no-confidence; it's basic financial discipline. I had been so focused on the act of selling that I ignored the context. I was looking for confirmation of a bubble and found it, all while ignoring the mountain of evidence suggesting something else was happening entirely.

That’s when I started looking at the bigger picture—the forward-looking narrative I had previously dismissed as corporate marketing. I looked at the Blackwell chip ramp and the new ‘Parakeet’ AI model, not as individual products, but as components of a relentless forward march. Most importantly, I reconsidered the partnership with Foxconn. I had initially seen it as just another press release. Now, I saw it as a strategic masterstroke. Nvidia isn't just powering data centers for tech companies anymore. It’s working to embed its AI and robotics platforms into the very heart of global manufacturing. They are moving from powering the digital world to powering the physical one.

My mental model of Nvidia’s Total Addressable Market shattered. I had been valuing it as a semiconductor company, subject to the industry's historical cycles and valuation metrics. This was a profound error. You don’t value a foundational utility like electricity or the internet using the metrics of its component manufacturers. Nvidia is becoming the foundational layer for the next industrial revolution. The argument that it's 'overvalued' is based on a rearview mirror understanding of what the company is and what it is becoming.

It’s a humbling thing to admit you were wrong, especially when you were so certain. The narratives of customer defection and overvaluation are seductive because they are simple and they fit familiar patterns. But they miss the tectonic shift happening underneath. They miss the ecosystem, the software moat, and the expansion from a niche market to the core infrastructure of the global economy. I don't pretend to know whether Nvidia's stock will hit $4 trillion or $5 trillion, and the path forward will undoubtedly be volatile. But I do know this: I was wrong to focus on the ripples of insider selling and customer optimization, because I completely missed the powerful, once-in-a-generation tide of technological transformation that Nvidia is not just riding, but creating. My journey has forced me to question my own biases, and I can only invite you, the reader, to do the same.