Nvidia investors have spent years watching GPU demand, hyperscaler spending and competition from custom AI chips.
But one number in the company’s latest filing deserves equal attention: $108.5 billion.
That is Nvidia’s maximum gross exposure to guarantees supporting AI infrastructure, including a $105 billion commitment tied to an OpenAI data-centre project in Ohio.
It is not an immediate cash liability and becomes effective only in stages under specific conditions.
Still, the disclosure marks an important change, as Nvidia is no longer simply selling hardware into the AI boom. It is increasingly using its balance sheet to help ensure the infrastructure behind that demand actually gets built.
Nvidia is putting its balance sheet behind GPU demand
The Ohio commitment supports roughly 4.25 gigawatts of IT load at SB Energy’s PORTS-Pike Technology Campus under 20-year leases to OpenAI.
Subject to limited exceptions, the campus will exclusively host Nvidia compute, giving the chipmaker a direct commercial benefit from helping the project secure financing.
The potential payoff is enormous. Nvidia estimates that each generation of infrastructure deployed at PORTS-Pike could represent around 1.5 million GPUs and $150 billion to $200 billion of revenue.
Over a 20-year lease, multiple hardware refresh cycles could follow. The economics could stretch across decades.
Analysts say the financing strategy could strengthen Nvidia’s competitive position by making it easier for customers to deploy its GPUs at scale. In that sense, access to capital becomes another way for Nvidia to deepen customer dependence and widen its moat against rival chipmakers.
That is increasingly part of Nvidia’s competitive strategy. Financial strength can make adopting Nvidia infrastructure easier while raising the barriers for rival chipmakers trying to win the same workloads.
The $108.5 billion is not debt, but it carries risk
The headline number needs context. Nvidia’s $108.5 billion maximum gross exposure consists of $105 billion linked to SB Energy and OpenAI, plus $3.5 billion of guarantees covering certain AI-cloud partners.
The OpenAI-linked guarantee does not become fully effective at once. Nvidia says obligations increase as nine data centres enter service, beginning around fiscal 2029, then decline as OpenAI makes lease payments.
The guarantee is also triggered only under specified circumstances, including tenant default or insolvency.
That limits the immediate financial impact, but it does not eliminate risk.
Bill Birmingham of REX Financial, quoted by Kiplinger, argued that “the quality of Nvidia’s future revenue has declined” because Nvidia is increasingly supporting the financial structures that create demand for its products.
If the projects succeed, Nvidia could capture extraordinary GPU sales, but if demand weakens or counterparties struggle, the company could face both weaker sales and financial exposure tied to infrastructure it helped support.
Wall Street is debating how much risk to price in
Bank of America analyst Vivek Arya takes a less alarmed view. In comments reported by TipRanks, Arya said Nvidia’s “off-balance-sheet risk matters but resists easy valuation,” while maintaining a Buy rating and $350 price target.
BofA argues that Nvidia’s valuation may already discount more vendor-financing risk than the fundamentals justify.
The company still has exceptional financial firepower: second-quarter revenue reached $96.2 billion, up 106% year on year, with Data Center revenue at $89 billion.
Cash, cash equivalents and marketable debt securities stood at $56.6 billion at quarter-end.
Those figures explain why the debate is not about whether Nvidia can currently support the guarantees.
It is about what kind of multiple investors should assign to earnings that are becoming more closely connected to infrastructure Nvidia itself helps finance.
The $108.5 billion figure is neither an automatic red flag nor something to dismiss. The guarantees could secure enormous future GPU sales and deepen Nvidia’s moat.
But they also make the company more financially intertwined with the AI buildout it supplies.
The post Thinking of buying Nvidia stock: one $108 billion number should be on your radar appeared first on Invezz