Nvidia briefly poised to become the most valuable company in history

Wall Street’s newest trillion-dollar darling isn’t a social platform, an e-commerce empire, or a software suite. It’s Nvidia—an infrastructure company. On Thursday, Nvidia’s stock reached a market capitalization of $3.92 trillion, briefly overtaking Apple’s all-time high before closing just shy of that figure. At a glance, this looks like another milestone in the generative AI gold rush. But beneath the record valuation is a more important question: what kind of tech economy are we building when one company, originally designed for gaming GPUs, becomes the heartbeat of modern AI?

Nvidia’s transformation is no accident. Its graphics processing units (GPUs), once prized by gamers and crypto miners, are now central to training and deploying large language models. From OpenAI’s ChatGPT to Microsoft’s enterprise-scale Copilot, nearly every major AI product is built on Nvidia’s compute hardware. The result? A product-market fit that isn’t user-led—it’s ecosystem-mandated. Nvidia is the toll booth through which all serious AI investment must pass. That’s less a growth story and more a supply chain chokehold.

This explains why Nvidia’s stock is now trading at nearly 32 times its projected earnings—actually below its five-year average of 41, according to LSEG data. Not because the hype has faded, but because earnings expectations are sprinting ahead even faster than the stock price.

Let’s be clear: Nvidia is not a platform in the traditional sense. It doesn’t control app ecosystems, user relationships, or monetization funnels. Instead, it sells the shovels in the AI gold rush—hardware, not software. But its current valuation implies something more. It suggests Nvidia has durable platform leverage—that it can extract margin from every layer of the AI value chain, indefinitely.

That logic starts to break when we consider the risks. Model compression is accelerating. Emerging players like China’s DeepSeek are building smaller, cheaper models that perform at near-parity. If more enterprises opt for optimized efficiency over brute-force compute, Nvidia’s stranglehold weakens. Moreover, dependency on one vendor—especially in a geopolitically sensitive category like semiconductors—creates systemic fragility. The April tariff shock, which temporarily knocked Nvidia’s stock, showed just how tightly the company is tethered to global trade alignment.

There’s another quiet implication behind Nvidia’s surge: passive investors are more exposed to the AI narrative than they realize. Nvidia now accounts for 7% of the S&P 500. Add in Microsoft, Apple, Amazon, and Alphabet, and five companies make up 28% of the index. That’s not diversification. That’s a thematic bet disguised as a balanced portfolio. And the theme isn’t tech. It’s AI infrastructure, concentrated in one link of the stack.

This raises a hard question for institutional allocators: are we over-allocating to a phase of the technology cycle, rather than to resilient systems? Because if the next evolution of AI—say, on-device models or alternative silicon—arrives faster than expected, that exposure flips from tailwind to trap.

We’re already seeing a divergence in AI infrastructure strategy. In the US, the model is clear: more compute, more power, more capex. Microsoft, Amazon, and Meta are racing to build billion-dollar AI data centers, all powered by Nvidia chips.

In China, the strategy is evolving. Rather than chase top-end GPU performance, players like Huawei and DeepSeek are focusing on smaller models that can run on domestically available hardware. This isn’t about beating Nvidia at its own game—it’s about changing the game entirely. If these efficiency-first strategies scale—especially in cost-sensitive emerging markets—it could undercut Nvidia’s biggest advantage: being “too critical to skip.”

Nvidia’s $3.92 trillion moment is remarkable. But it also highlights a fundamental truth of today’s tech ecosystem: we’ve confused critical infrastructure with platform durability. Yes, Nvidia is indispensable today. But moats built on current dependency are vulnerable to model shifts, policy backlash, and innovation that changes the stack order. Just ask Intel.

Founders building in the AI space shouldn’t assume Nvidia’s dominance is permanent. Neither should investors. Because in infrastructure-driven cycles, today’s bottleneck is tomorrow’s commodity. The real risk? When every player builds around the same dependency, the fragility compounds. Pricing power masks volatility. Demand masks exposure. And when a single supplier dictates velocity across a high-growth vertical, innovation bottlenecks upstream.

Nvidia isn’t “too big to fail”—it’s too concentrated to ignore. If training gets cheaper, or inference moves to edge devices, its leverage erodes fast. This isn’t platform economics. It’s supply chain imbalance with a premium multiple. And that’s not a moat. That’s a mirage.