Nvidia's AI Empire Now Runs on a Circular Logic That Investors Can't Ignore

Nvidia posts record revenue and strong guidance, yet shares dip 4.1% as investors question the quality of growth tied to its own financing programs.

The numbers were, by any standard, extraordinary. Nvidia reported $96.2 billion in revenue for its second fiscal quarter of 2027, up 106 percent year over year, with the data center segment alone surging 117 percent to $89 billion. Earnings per share of $2.22 cleared the $2.09 consensus, and the company's guidance of $108 billion for the current quarter blew past market expectations of roughly $104.9 billion.

Yet the stock fell 4.1 percent on Friday in German trading, closing at €187.78 — a curious response to what should have been a triumphant report. The shares had initially rallied in the days following the release before giving back those gains. The stock now sits 7.3 percent below its 52-week high of €202.50, though it remains comfortably above its 200-day moving average of €168.76.

That disconnect between stellar fundamentals and a skittish tape is the real story here. Investors are no longer asking whether Nvidia can keep growing — they're asking whether that growth is built on solid ground or on a self-reinforcing financial loop that could unwind just as quickly as it was constructed.

The Bank of Nvidia

The central question concerns the quality of future revenue. Nvidia has poured roughly $50 billion into AI labs that, in turn, purchase Nvidia chips. Alongside partners including Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR, the company has assembled financing commitments exceeding $500 billion. CFO Colette Kress has disclosed that about 25 percent of the business expected in fiscal 2028 will come from customers Nvidia itself helps finance.

The company has also committed up to $105 billion in leasing guarantees for OpenAI's Ohio campus, alongside a $1.5 billion equity investment in SB Energy. Add it all up, and the question writes itself: Is Nvidia still a chipmaker selling products, or has it quietly become the banker for its own AI infrastructure boom?

Critics call it circular financing — Nvidia lending money to customers so they can buy Nvidia products, with the resulting revenue booked as organic growth. Defenders counter that without these capital bridges, many AI cloud providers simply couldn't place the orders their long-term roadmaps require. Both arguments carry weight, and the market's nervousness reflects that ambiguity.

The key metric investors should watch, therefore, isn't the growth rate itself but the shifting balance between organically financed demand and circularly financed demand. If that ratio tilts further toward subsidized customers, a meaningful slice of the reported growth rests on shakier foundations than the headline numbers suggest.

A Changing Customer Base

What's arguably more remarkable than the growth rate itself is who's doing the buying. The established hyperscalers grew their contribution 102 percent to $48.7 billion, but a newer class of customers — AI-native clouds, industrial enterprises, sovereign projects and specialized providers like CoreWeave and Nebius — expanded even faster, up 138 percent to $40.3 billion.

This diversification reduces the concentration risk inherent in relying on three or four tech giants for the company's future. Amazon Web Services has ordered two million additional GPUs for delivery through the second quarter of fiscal 2029, on top of an earlier commitment exceeding one million units. OpenAI is reportedly planning 12 gigawatts of Nvidia capacity by 2030. With the Vera Rubin platform now in mass production, the value per gigawatt of compute capacity is expected to jump from $18 billion to $40 billion.

That growth, however, comes with a literal price tag. Memory chip shortages have prompted Nvidia to raise prices on its Vera Rubin and Grace Blackwell AI servers by 15 to 17 percent for deliveries starting in 2027. Memory now accounts for roughly a quarter of a rack's cost, while the GPU share has fallen from over 80 percent to about 50 percent. New memory fabrication capacity won't come online until 2029 or 2030, according to industry estimates, and TSMC — Nvidia's key manufacturing partner — is planning price increases of 10 to 15 percent across all process nodes.

The Bear Case Has Teeth

The optimistic narrative rests on the breadth of demand and a valuation that looks reasonable by historical standards. The forward price-to-earnings ratio of 34.9 sits well below the ten-year average of 61.5, suggesting the market has already priced in some caution. Analysts at Raymond James (target: $515) and Evercore ($465) point to persistent memory shortages and fully booked foundry capacity at TSMC as evidence that demand will remain robust.

But the bear case is built on more than speculation. Insider selling totaled over $664 million in 2026, with a single director accounting for 67 percent of all sales — including one transaction of 885,000 shares at $210.17. Such sales don't prove a loss of confidence, but they feed concentration concerns, particularly with prominent skeptics like Michael Burry reportedly voicing criticism.

More telling is the cash flow picture. The free cash flow margin collapsed from 59.5 percent in the prior quarter to 22.2 percent, a dramatic decline driven by massive investments in capacity and financing programs. Should hyperscalers like Google, Amazon, Microsoft or Meta accelerate their custom chip efforts, Nvidia's dependence on its largest customers could become a vulnerability — the very clients driving today's growth are potentially tomorrow's fiercest competitors.

What to Watch

The bull case holds as long as order books — reportedly exceeding $2 trillion in the cloud segment — keep growing and gross margins stay in the 71 to 72 percent range. Under those conditions, the stock should drift toward the upper end of analyst targets.

The bear case gains traction if the share of circularly financed revenue keeps climbing or if hyperscaler custom silicon strategies become visibly more aggressive. At that point, the current 34.9 times earnings multiple would quickly come under scrutiny.

The immediate test is the promised $108 billion revenue mark for the third fiscal quarter. Hit that number without further deterioration in the cash flow margin, and confidence in the business model should return. Miss it, or show further signs of financial engineering strain, and the questions now swirling around Nvidia will only grow louder.