Nvidia's $279 Billion Gamble: Why the Record Revenue Forecast Comes With Record Commitments

Nvidia just told investors it expects revenue to grow 70% next fiscal year — about $690 billion, far above the $570 billion Wall Street had modeled. The stock rallied after hours on August 26, as it usually does.

The headline is a growth number. The story is $279 billion in supply commitments, more than double the $119 billion from just one quarter ago. Nvidia is betting that every supplier on the line will keep delivering, and that demand will not pause — because the commitments are on Nvidia's side, not the suppliers'. If the AI buildout slows, the risk doesn't vanish. It gets locked in.

This is the decision under review: whether a 70% growth forecast, fronted by a supply commitment surge this large, still justifies the same allocation as when the forecast was 44%.

The quarterly results are, by now, the familiar pattern. Revenue came in at $96.2 billion for the quarter ended July 26, up 106% year over year and 18% from the previous quarter. Data Center revenue alone was $89 billion. Gross margin held at 75%. Nvidia beat revenue and EPS estimates.

But the number that changes the picture is what followed: $279 billion in multi-year supply-chain commitments, up 135% in a single quarter. Memory is the primary target — the component that constrains how many GPUs Nvidia can ship. Nvidia's response was to buy the bottleneck, years in advance.

That is a demand signal and a leverage signal at once. On one side, you cannot commit this much unless you believe customers will buy the chips you're building. On the other, those commitments sit on Nvidia's books as obligations — and they don't disappear if demand turns.

Huang was blunt about the constraint. He said demand for AI chips is growing at roughly 100% — well above the 70% guidance. "A lot, a lot higher," he put it. The bottleneck isn't customers. It's memory supply. Nvidia is pre-paying suppliers to ensure first allocation of what the industry can physically produce.

But the margin cost of that strategy is visible in the guidance. Gross margins are expected to compress from 75% to 71%–72% by the fourth quarter of fiscal 2027, then settle at 72%–73% for the full fiscal 2028. Memory suppliers are capturing a larger share of the AI value chain. Nvidia's workaround — roughly 15% price increases passed to customers — will cushion the hit, but the floor is lower.

The Wall Street narrative that attached itself to this earnings beat involves SpaceX.

SpaceX held its first public earnings call on August 4, ten days before Nvidia's. CEO Elon Musk used the platform to commit that SpaceX will build its AI computing infrastructure exclusively on Nvidia's Vera Rubin architecture. The numbers behind that commitment are aggressive: SpaceX currently operates 1.4 gigawatts of nameplate compute capacity, targets 2 gigawatts by year-end 2026, and aims for close to 10 gigawatts by the end of 2027. Musk floated a "tentative target" of 20 gigawatts of power and cooling capacity — a figure that, even at a fraction, translates into enormous GPU orders.

SpaceX spent roughly $28.5 billion on capital projects in the first half of 2026, with AI infrastructure as the dominant spend. It has already signed multi-year compute offtake agreements worth $14.1 billion in contracted cloud services, with deals including Anthropic and Google, and monthly contracts worth $150 million per customer.

Here is how analysts have linked the two. A 20-gigawatt deployment — or even a quarter of it — would generate an incremental $100 billion in revenue for Nvidia, according to Melius Research. That would add roughly $2 to Nvidia's estimated earnings per share for the fiscal year. Huang has cited $1 trillion in cumulative Vera Rubin and Blackwell sales as the target through 2027; at that level, Nvidia's earnings power is substantial — but the timing of when that revenue hits is what matters for the near-term return curve.

The arithmetic looks clean. The question is whether the commitments translate to delivered orders fast enough to matter for Nvidia's next few quarters.

SpaceX is not yet a material revenue line for Nvidia. It is a demand signal — one that suggests the AI buildout extends well beyond the hyperscalers who have dominated the headline. And the timing is meaningful: SpaceX can bring massive clusters online in months, not years, using on-site natural gas power and modular systems. Where Microsoft has contracted for 10 gigawatts of capacity that won't be online until late 2027 or early 2028, SpaceX can fill the gap.

But SpaceX's contracts contain termination provisions. The AI revenue is real — $3.2 billion in 2025, expected to top $60 billion by 2027 on Wall Street estimates — but it is not guaranteed recurring revenue. The orbital data center vision, where SpaceX launches Nvidia racks into space to sidestep land and cooling constraints, is a decade-out thesis that depends on Starship achieving rapid reusability. Near-term value is tethered to Earth.

The SpaceX story is color, not the core. The core is Nvidia's own commitment to $279 billion in supply — a number that tells you something more concrete about the company's conviction, its leverage, and its risk.

Let's look at what those commitments mean on a different lens.

Cisco Systems entered the dot-com bubble with massive supplier commitments for networking equipment. When demand collapsed, those commitments became write-downs — and the stock lost 99% of its value. The mechanism was identical: a company locked in supply capacity on the assumption that demand would keep accelerating. Demand didn't. The commitments stayed.

Nobody is comparing Nvidia to Cisco. The AI buildout is real, the customers are flush with capital, and Nvidia's platform has moats that Cisco never enjoyed. But the structural risk is the same: commitments are fixed obligations. Revenue is variable. The gap between them is what makes or breaks the thesis when the cycle turns.

Nvidia's CFO Colette Kress defended the exposure on the call, calling it "low-risk, high-reward." She noted that frontier AI labs have proven traction and that Nvidia's platform is "fungible and durable." The argument is that even if one customer slows, demand from the next fills the gap. The platform is so broadly adopted that a single downturn in one segment doesn't crater the whole machine.

That defense holds — while the machine keeps accelerating.

There is another risk that runs parallel to the supply commitments. Nvidia has invested nearly $50 billion directly in frontier AI labs and partnered with private equity firms including Apollo, BlackRock, Blackstone, and Goldman Sachs to mobilize over $500 billion in third-party capital for AI infrastructure buildout. The maximum gross guarantee exposure stands at $108.5 billion, primarily from credit support for data center projects leased to customers like OpenAI.

This is what critics call "circular financing": Nvidia sells chips to companies that borrow to buy them, and Nvidia sometimes guarantees the borrowing. The market has reacted to this structure with skepticism. When reports surfaced of potential $250 billion in OpenAI capacity guarantees, Nvidia's credit-default swap pricing nearly doubled from 40 to 82 basis points and the stock shed $250 billion in market cap. Kress's defense — that the platform is "fungible and durable," so even a defaulted customer means Nvidia can redeploy the hardware — is plausible. But it's also untested at this scale.

So where does this leave the investment case?

Nvidia's operating metrics are still extraordinary. Revenue grew 106% year over year. Free cash flow for the trailing twelve months hit $127 billion. Return on invested capital is 87%. The company returned $26 billion to shareholders in the quarter alone and has $99 billion remaining under buyback authorization. On a trailing P/E of 27, the stock looks cheap — until you look at the forward P/E of 58, priced against next year's earnings.

The 70% growth guidance is the key variable. If Nvidia delivers on it — which the supply commitments, the customer base, and the AI demand cycle all suggest is plausible — the current valuation compresses rapidly. Revenue of roughly $690 billion at the guided rate, with margins settling at 72%–73%, would generate non-trivial earnings growth that justifies holding through volatility.

But the opportunity cost question remains. At a $5.2 trillion market cap, Nvidia is no longer the dark horse. The thesis is priced in at a level where the margin for error has shrunk. The memory bottleneck that constrains supply to 70% when demand runs at 100% is also the thing that compresses margins. The commitments that secure supply also create exposure. The circular financing that accelerates adoption also ties Nvidia's fortunes to the solvency of its customers.

What separates the bull case from the bear case is not whether AI demand slows — it doesn't need to. It's whether the acceleration is enough to absorb the leverage. If supply constraints ease in 2028 and the memory bottleneck clears, margins recover and the commitments convert to revenue. If they don't, Nvidia sits on $279 billion of obligations and a margin floor that hasn't found its bottom yet.

The SpaceX angle is a useful reminder that the demand base is broader than the three hyperscalers — but it's one customer in a buildout that involves hundreds. The real question is whether you believe Jensen Huang's assessment that demand is "a lot, a lot higher" than the 70% guidance. If you do, the commitments are a strategic moat, not a balance-sheet risk. If you don't, they're the other side of the coin.

Either way, the decision has been made. Nvidia has committed. Now the supply chain has to deliver — and the customers have to pay.