Nvidia's Double-Edged Boom: Record Demand Meets a $200 Billion Credit Question

Nvidia's Q2 revenue surged 106% to $96.2B, but off-balance-sheet credit exposure may hit $200B by 2028, raising risk concerns.

The arithmetic of Nvidia's ascent is getting harder to ignore. On one side sits a demand picture so robust that Amazon Web Services has tripled its order to 2 million graphics processors. On the other sits a financing web so intricate that the company's off-balance-sheet credit exposure could approach $200 billion by 2028 — a figure that raises uncomfortable questions about who ultimately carries the risk of the AI buildout.

The chipmaker's latest quarterly results, covering the second fiscal quarter of 2027, were nothing short of spectacular. Revenue hit $96.2 billion, up 106 percent year over year and comfortably ahead of the $92 billion consensus estimate. Adjusted earnings per share climbed 111 percent to $2.22, beating the $2.09 analysts had penciled in. Management guided third-quarter revenue to $108.0 billion, plus or minus 2 percent, with no assumed data-center sales from China baked into that number.

Yet the stock fell 4.1 percent on Friday, closing at €187.78. The disconnect between blowout fundamentals and a negative market reaction came down to two factors: margin guidance and a hawkish rate commentary from Fed Governor Christopher Waller. Nvidia projected third-quarter gross margins of 74 percent, sliding to 71.5 percent in the fourth quarter, as rising costs for high-bandwidth memory chips eat into profitability. The shares now sit 7.3 percent below their 52-week high of €202.50, reached in May, while maintaining a 3.6 percent premium to the 50-day moving average of €181.21.

The Banker's Balance Sheet

The more structural concern, however, is the growing financial entanglement between Nvidia and its own customers. Chief Financial Officer Colette Kress has said that AI labs co-financed by Nvidia are expected to contribute roughly a quarter of the revenue projected for fiscal 2027. Together with six major investment houses — Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR — the company has established a platform designed to channel more than $500 billion into AI infrastructure.

Morgan Stanley analysts estimate Nvidia's credit exposure could grow to nearly $200 billion by 2028, mostly held off-balance-sheet. S&P maintains an AA rating with a stable outlook, citing a modest leverage ratio of 0.4. But the commitments extend further: roughly 12 gigawatts of Nvidia compute capacity are pledged to OpenAI through 2030.

That scale of interlocking obligations has caught the attention of skeptics. Steve Eisman, the investor who famously bet against US subprime mortgages before the 2008 crisis, has warned that the entire AI trade hinges on the financial stability of OpenAI and Anthropic. If either company's spending momentum falters, he argues, the fallout could trigger a sell-off across the technology sector. The numbers show why: OpenAI posted an operating loss of $12.3 billion on 18 percent revenue growth in a recent quarter, while Anthropic was already operating profitably over the same period.

Supply Constraints and Strategic Moves

For now, demand signals remain overwhelmingly positive. AWS's decision to triple its GPU order to 2 million units — aimed at AI, robotics and agent workloads — underscores the appetite from hyperscalers. AI cloud provider Lambda secured roughly $1 billion in short-term private debt, with Nvidia's backing, to purchase chips for a Microsoft collaboration. SpaceXAI has also committed to Nvidia's Vera CPU for large-scale agentic AI workloads.

The supply side tells a more complicated story. Kress described a "preliminary expectation" of 70 percent revenue growth in fiscal 2028, but cautioned that supply constraints are currently limiting the pace. Long-term purchase commitments have more than doubled in a single quarter to $279 billion, creating a potential inventory risk if component prices decline — even though the current environment remains one of acute shortage.

Nvidia is also deploying capital aggressively across its ecosystem: $18 billion in equity investments in AI model developers and infrastructure financiers is planned for the remainder of the fiscal year, bringing total stakes in private companies to $47.9 billion as of end-July. The $13 billion acquisition of Hugging Face fits the same pattern of capital-intensive expansion.

Washington, Insider Sales and the Road Ahead

Regulatory headwinds are building alongside the commercial momentum. The Trump administration is examining broader semiconductor tariffs that could extend to complete servers, beyond the existing 25 percent levy on certain chips from which US data centers are currently exempt. Nvidia has responded by establishing its own political action committee, NVPAC, to press its case in Washington.

On the governance front, insider sales over the past three months totaled $410.44 million, according to regulatory filings. Ajay K. Puri, formerly Executive Vice President of Worldwide Field Operations, departed on August 24 and is no longer subject to Section 16 reporting requirements.

The stock's valuation offers some perspective on market sentiment: at roughly 23 to 24 times forward earnings, it trades below its own five-year average of about 35. The price remains around 34 percent above last September's low, suggesting the financing debate has not derailed the broader upward trajectory.

A quarterly dividend of $0.25 per share is payable October 1, with a September 10 record date. Third-quarter results are due November 18. Between now and then, investors will be weighing whether margin compression is a temporary supply-driven phenomenon or a structural shift — and whether Nvidia's increasingly bank-like balance sheet is a source of strength or a liability waiting to be tested.