Nvidia doubles its sales, but investors have their eye on the future
By Nikhil Raghavan ·
The numbers are, by any metric—and I mean this in the most literal sense—staggering.
The $96 Billion Quarter and the Illusion of Infinite Demand
The numbers are, by any metric—and I mean this in the most literal sense—staggering. Nvidia posted $96.2 billion in quarterly revenue, exceeding forecasts, with data center sales alone accounting for $89 billion, according to reports cited by CNA. Jensen Huang declared that AI has reached an "inflection point," a phrase designed to sound like physics rather than economics. Colette Kress backed this up by projecting 70% growth into fiscal 2028, a figure significantly above the 45% analysts had projected, as Euronews reported. The narrative is one of unstoppable acceleration: "compute is revenue, and demand is accelerating," Huang stated.
When Growth Projections Outpace Supply Chains
What you are reading here—the $5 trillion market cap, the commitment of up to $105 billion for an OpenAI data center in Ohio—is not a statement about engineering capacity; it is a declaration of capital availability. The company’s own CFO noted that demand acceleration was visible "even at our scale." This is where the mechanism breaks down. When every quote focuses on how much money is flowing into compute, and how little attention is paid to who actually implements the infrastructure, we are looking at an operational bottleneck masquerading as a market opportunity. The current growth is driven by Blackwell chips; the successor, Vera Rubin, is merely the next iteration of the same race for foundational dominance.
Cycles Repeat: From TMT to AI Infrastructure
The shared mechanism here—the cycle where foundational technological adoption creates hyper-accelerated demand and speculative capital investment—is not new. It is a perfect echo of the dot-com bubble. The late 1990s saw investments in the World Wide Web create an exponential, almost mythical valuation curve, leading to a massive overextension of venture capital into TMT startups. What was priced today as inevitable, based on technical capability alone, has historically proven susceptible to systemic correction when the speculative layer finally detaches from tangible utility.
The difference between then and now is merely one of nomenclature: instead of dot-coms, we have AI clouds; instead of bandwidth, we have teraflops. But the underlying pressure—the belief that this time's technological breakthrough fundamentally changes the rules of economics forever—is identical. The sheer magnitude of these projections suggests a market pricing not on steady utility growth, but on perpetual narrative momentum.
The current valuation is built upon the premise that 70% year-over-year growth can be sustained through multiple generations of chips and geopolitical maneuvering. This assumes perfect execution across global supply chains, zero regulatory friction (especially given the China restrictions mentioned), and an endless pool of capital willing to fund the next speculative layer. The history of technology shows us that while foundational shifts are real, the financialization of those shifts always overshoots reality, leaving a massive vacuum when the funding dries up.