Nvidia Has a $350 Target and a 54% Problem
Nvidia reported 81.6 billion in Q1 fiscal 2027 revenue on Wednesday — an 85 percent year-over-year jump, a record quarter, and another quarter where the company described demand as far exceeding supply. Bank of America raised its price target to $350, citing the Data Center segment's 92 percent year-over-year growth and a new CPU line called Vera Rubin that Nvidia says opens a 200 billion total addressable market the company has never touched. The bull case writes itself.
But there is a structural problem buried in the same filing that Wall Street seems inclined to look past.
Three unnamed customers accounted for 54 percent of Nvidia's Q1 revenue — 21 percent, 17 percent, and 16 percent respectively, all in Compute & Networking. A year ago, two unnamed customers accounted for 39 percent of revenue. The concentration has deepened, not narrowed. Nvidia is not selling GPUs to a broad industrial base; it is fulfilling the compute ambitions of a small number of hyperscalers, and those ambitions are structurally constrained by what Nvidia can physically produce.
On the earnings call, CEO Jensen Huang said Nvidia will be supply constrained throughout the entire life of Vera Rubin — not a temporary cycle, a permanent condition until manufacturing capacity catches up. That is not a softening demand signal. It is a hard ceiling on revenue capture regardless of how large the TAM grows. Nvidia tells investors it can reach a 200 billion CPU market; it simultaneously tells investors it cannot fulfill current demand. The execution paradox does not resolve cleanly into a price target.
The Cisco comparison is not original — it surfaces in every conversation about infrastructure-layer value capture in tech — but it is worth sitting with. Cisco in 2000 was the essential infrastructure of the internet boom. Its routers connected every enterprise, every carrier, every emerging dot-com. The market cap peaked near 500 billion dollars. The companies that built on Cisco — Amazon, Google — captured the durable economics of the era. Cisco did not disappear; it became a solid infrastructure company. The transformative value of the internet layer it enabled accrued elsewhere.
The question Nvidia poses is whether the AI application layer — the reasoning models, the agents, the enterprise workflows — settles its economics above or below the chip layer Nvidia occupies. Right now, the chip layer has all the momentum. But the supply constraint is also the tell. When you cannot make enough of something to meet demand, you are pricing and distributing based on manufacturing capacity, not market desire. That is a different business than one where you capture the full value of what your customers build on top of you.
A Seeking Alpha analysis published this week made the contrarian case directly: Nvidia stock has traded essentially flat after the beat-and-raise quarter. The market is not repricing the stock higher despite extraordinary numbers. The analyst argued this reflects a growing institutional view that H200 will be the peak margin product before custom ASICs from Google, Amazon, and Meta begin to meaningfully displace GPU-only training workloads. The supply constraint may actually accelerate that displacement by giving hyperscalers a direct incentive to build their own inference silicon.
The 119 billion in long-term manufacturing commitments on Nvidia's balance sheet suggests the company is betting the supply constraint resolves on its terms — that when capacity catches up, demand remains. That bet is coherent if the AI buildout continues at its current trajectory and custom silicon does not capture a significant share of inference workloads. It is less coherent if the application layer rationalizes its compute spend and the hyperscalers' own ASIC programs mature faster than expected.
The BofA price target of 350 implies Nvidia is worth 26 times estimated calendar year 2027 earnings. That multiple bakes in a world where the concentration risk is manageable, the supply constraint is temporary, and Vera Rubin's 200 billion TAM converts to revenue at high margins. The structural evidence — deepening customer concentration, a CEO who says supply will be constrained forever, 119 billion in fixed manufacturing commitments — suggests a company that is extraordinary in the near term and structurally uncertain over a longer horizon.
Nvidia is almost certainly the most important technology company in the world right now. Whether it is the best long-term value capture is a different question — and one the quarterly record does not answer.
Sources: NVIDIA Q1 FY2027 earnings press release | SEC 10-Q filing | TechCrunch earnings call coverage | Bank of America note via Yahoo Finance | Seeking Alpha: "Count Me Bored"