Nvidia's Financial Dominance and Growth

When CFO Colette Kress told analysts to expect roughly 70% year-over-year revenue growth in fiscal 2028, she was guiding well above what Wall Street had priced in - the average analyst estimate stood at just 44% ^[1]. The gap matters because of what came next: management frames the 70% figure as a supply-constrained outlook, arguing that absent bottlenecks in the current supply chain, revenue could actually more than double year-over-year ^[1]. That reframes the growth story entirely - Nvidia isn't describing a market it needs to go find, it's describing a market it can't yet fully serve.

Two proof points give the guidance teeth. First, Amazon's recent commitment to buy 2 million Nvidia GPUs for delivery across 2027 and 2028 ^[1]is the kind of multi-year forward order that only makes sense if a hyperscaler believes compute demand keeps compounding well past this earnings cycle. Second, coverage framing the 70% figure as putting Nvidia on track to become one of tech's largest companies by some measures ^[2]underscores how unusual it is for a company already this large to guide toward accelerating, not decelerating, growth. The same quarter that produced this guidance saw Data Center revenue climb 117% year-over-year to $89.0 billion ^[3]- the base the 70% number has to build on.

Strip away the quarter-to-quarter noise and the long-run projection is the more startling number. Wall Street's compounded EPS growth forecasts imply Nvidia's annual net income could reach approximately $851 billion by 2029 ^[4]- more than five times Saudi Aramco's $161.1 billion 2022 profit, itself the largest annual profit any company has ever reported ^[4]. Run the same trajectory through revenue instead of profit and the comparison gets stranger still: a projected $1.35 trillion in 2029 revenue would rank as the 18th-largest economy on Earth, just behind Saudi Arabia's GDP and just ahead of Switzerland's ^[4].

The market has already started pricing pieces of this in. Nvidia became the first company to cross a $5.5 trillion market capitalization ^[5], sitting roughly $1 trillion ahead of the next-largest company, Alphabet ^[5]. Social commentary has picked up on just how large a share of the market that represents - posts circulating on X this quarter pointed out that Nvidia's roughly $5.2-5.3 trillion valuation alone equals something like 8% of the entire S&P 500 and is comparable in scale to 16% of US GDP, a framing that treats one company's balance sheet as a macroeconomic variable in its own right. Whether or not the 2029 figures hold up, the fact that a single chipmaker is now being measured against sovereign economies by professional forecasters and retail traders alike is itself the story.

Growth guidance this aggressive raises an obvious question: who is actually paying for all the compute Nvidia says it's about to sell? Part of the answer is that Nvidia itself is. The company's customer guarantees carry a maximum exposure of $108.5 billion, and roughly $105 billion of that is tied to a single 4.25-gigawatt OpenAI data center project in Ohio ^[6]. In practice, that means Nvidia is backstopping a meaningful share of the very demand it reports as revenue, a structure critics say inflates both Nvidia's top line and its customers' balance-sheet valuations at the same time ^[6].

There's a second, quieter risk sitting underneath that arrangement: the hardware itself loses value as newer chips ship, so guarantees written against today's GPUs are, to some degree, a bet that the underlying collateral won't lose value faster than the debt gets repaid. Community discussion on YouTube and Reddit this quarter leaned hard into exactly this mechanic, describing reports of Nvidia arranging large financing packages with outside asset managers to back GPU-collateralized loans, and in some threads drawing explicit comparisons to the vendor-financing loops that preceded the 2008 financial crisis. None of the specific financing figures circulating in that discussion are confirmed in on-the-record reporting used here, but the sentiment itself is notable - a meaningful slice of Nvidia's own audience is now openly asking whether its growth engine and its financing arm are becoming the same thing.

Not everyone is buying the story at these prices, and some of the clearest signals are coming from people who used to own the stock. Peter Thiel sold down a roughly $100 million Nvidia stake, and SoftBank exited a position worth roughly $6 billion ^[6]- moves that skeptics read less as routine profit-taking and more as an early exit ahead of a broader unwind. That skepticism isn't limited to insiders selling shares. Morningstar's Brian Colello argues the opposite case - that AI demand doesn't appear to be slowing at all ^[7]- which is precisely the split defining the current debate: is this a company growing into its valuation, or one whose backers are quietly heading for the exits?

The more structural version of that worry comes from commentary on the AI data-center financing boom itself, which warns that once the excitement around AI fades and investors start demanding results, the broader AI bubble will likely deflate ^[8]. Unlike the circular-financing mechanics in Nvidia's own contracts, this critique targets sentiment and positioning across the whole AI trade - Nvidia included - rather than any single balance-sheet line. Read together, the stock-sale signal and the bubble-deflation warning describe the same risk from two different angles: one is about who's already stepped away, the other is about what happens if everyone else decides to follow.

Nvidia's growth strategy is no longer just about selling more chips - it's about owning more of the stack around them. The company agreed to acquire open-source AI platform Hugging Face for roughly $12.9-13 billion, its second-largest acquisition ever, trailing only a roughly $20 billion licensing and technology agreement it struck with inference-chip startup Groq ^[9]. Both deals point the same direction: buying capability Nvidia doesn't want to build in-house, whether that's an open model ecosystem or competing inference IP.

That same expansionist instinct shows up on the investing side of the business. X and Reddit posts this quarter converged on a claim that Nvidia's equity-investment portfolio has grown to roughly $99 billion - plus another $25 billion in future commitments - up sharply year-over-year, with Reddit's finance community framing it as evidence Nvidia is running something closer to a venture fund than a component supplier and asking outright whether that makes it a moat or a monopoly. Whichever framing wins out, the pattern is consistent: Nvidia is spending Q2's record profits to buy influence over every layer of the AI stack it doesn't already control.