By 
deVere CEO Nigel Green warns that turning AI chips into financeable assets could create new risks as Nvidia seeks to mobilise more than $500 billion in third-party capital for AI infrastructure.
Nvidia, the world’s most valuable company, is pushing the boundaries of AI infrastructure financing by working with selected financial institutions to mobilise more than $500 billion in third-party capital, a move that deVere Group CEO Nigel Green says deserves close scrutiny.
RELATED: Nvidia revenue surges 85% to $81.6 billion as AI demand soars, but market wants more
The initiative represents a significant evolution in the way artificial intelligence infrastructure is financed, with Nvidia effectively helping position its chips and related computing capacity as assets against which large-scale financing can be arranged.
Nvidia CEO Jensen Huang has described the development as a landmark moment, arguing that technology chips are becoming an investable asset class as computing increasingly takes on the characteristics of essential infrastructure.
But Green, chief executive of one of the world’s largest independent financial advisory organisations, has warned that the model carries similarities to financial structures that have previously amplified market risks.
“What Nvidia has done here is elegant, and that is exactly what worries me,” Green said.
Nvidia Seeks $500 Billion in AI Infrastructure Financing
Nvidia’s strategy comes as demand for AI computing infrastructure continues to drive enormous investment in GPUs, data centres and related technologies.
Under the new financing model, Nvidia is partnering with financial institutions to help mobilise more than $500 billion in third-party capital for AI infrastructure.
The scale of the initiative is significant because it potentially changes the relationship between technology manufacturers, customers and lenders.
Rather than simply selling chips to technology companies and data-centre operators, Nvidia is becoming involved in financing structures that enable customers to acquire large quantities of AI computing infrastructure.
Green argues that this distinction is important because the financing effectively places long-term financial expectations on technology assets that have traditionally depreciated rapidly.
From Computer Chips to Investable Assets
According to Green, the central question is whether GPUs can reliably retain enough economic value over a long enough period to support the debt being raised against them.
“Chips have never been treated as a bankable, long-duration asset before, because chips depreciate fast and lose value the moment a newer generation arrives,” he said.
The rapid pace of semiconductor innovation means that newer generations of GPUs can potentially reduce the economic value and competitiveness of previous generations.
Green therefore questions whether financing structures traditionally associated with long-lived infrastructure assets can be safely applied to rapidly evolving technology.
“Turning that into something institutions can lend against, the way they lend against a building or a highway, only works if the underlying asset actually holds its value over time,” he said.
Nvidia’s Role in Customer Financing Raises Questions
Another issue highlighted by Green is Nvidia’s expanding role in the financing ecosystem surrounding its products.
“This is not simply Nvidia selling chips anymore,” he said. “It’s Nvidia helping its own customers borrow enormous sums to buy those chips, then helping structure the financing that makes the borrowing possible in the first place.”
He warned that the scale of the arrangements makes the relationship between supplier, borrower and financier particularly important to monitor.
“When the seller starts underwriting the buyer’s debt, at this scale, that’s worth watching closely,” Green said.
AI Debt Is Already Growing Rapidly
The proposed financing comes against a backdrop of substantial borrowing by major technology companies investing in AI infrastructure.
Green estimates that hyperscalers have already borrowed roughly $250 billion in 2026, several times their normal annual borrowing.
Adding another $500 billion of financing would represent a substantial expansion of leverage across the AI infrastructure ecosystem.
The concern, according to Green, is not simply the size of the debt but the assumption underpinning it: that AI infrastructure will generate sufficient and sustained returns to service the borrowing.
“Layering another $500 billion in financing on top of that is a genuinely large amount of leverage building on leverage,” he said.
Echoes of the 2008 Financial Crisis
Green draws a comparison with the period leading up to the 2008 global financial crisis, while acknowledging that the assets and markets involved today are fundamentally different.
He argues that the historical lesson is the danger of building significant leverage around assumptions about the future value of an underlying asset.
“Loans were packaged, leveraged, and sold on the assumption that the underlying asset would hold or increase in value,” Green said.
When those assumptions failed during the financial crisis, leverage amplified losses.
Green said the AI market is different, but believes the underlying principle warrants similar scrutiny where substantial borrowing is based on assets whose future economic value remains uncertain.
“This is a different asset and a different market, but the underlying structure, borrowing heavily against something whose future value is genuinely uncertain, deserves the same scrutiny,” he said.
AI Infrastructure Could Still Deliver Strong Returns
Green stressed that his concerns should not be interpreted as a prediction that Nvidia’s financing strategy will fail.
If demand for AI continues to expand and data centres generate the revenues anticipated by investors and operators, the financing structures could function as intended.
He also acknowledged the broader argument that computing is increasingly becoming a fundamental component of economic infrastructure.
“Jensen Huang’s argument that compute is now infrastructure, similar to electricity or the internet, is not unreasonable on its face,” Green said.
The critical question, however, remains whether the economic lifespan of the underlying technology matches the duration and scale of the financing attached to it.
GPU Depreciation Is a Key Risk
According to Green, the rapid development cycle of GPUs creates a potential mismatch between asset life and debt maturity.
GPU technology has historically evolved rapidly, with new generations offering significant improvements in performance and efficiency.
That means companies financing today’s hardware over longer periods must consider whether those assets will continue generating sufficient economic returns as newer technologies emerge.
“Borrowing long-term against an asset with a genuinely uncertain shelf life is a real gamble, even when the borrowing is dressed up in the language of infrastructure investment,” Green warned.
The issue could become more significant if AI infrastructure investment continues to grow faster than the underlying revenue generated from AI applications and services.
Investors Urged to Examine AI Exposure Carefully
Green said the development should prompt investors to look beyond headline valuations and investment flows and understand where risk actually sits within the AI ecosystem.
He distinguished between owning companies and infrastructure that generate durable cash flows and gaining exposure to debt backed by assets whose long-term value may be uncertain.
“There’s a real difference between owning the infrastructure and technology generating durable returns, and owning exposure to the debt piled on top of assets that may or may not hold their value,” he said.
He urged investors to examine the specific assets, companies and financing structures underlying their AI-related investments rather than treating the entire sector as a single risk category.
AI Investment Boom Faces a New Test
Nvidia’s financing initiative highlights how rapidly the AI boom is transforming traditional investment and lending models.
As computing capacity becomes increasingly essential to businesses, governments and technology platforms, the industry is attracting levels of capital previously associated with major infrastructure projects.
The opportunity is enormous, but so are the financial commitments required to build the infrastructure.
The emerging challenge for investors and lenders will be determining whether the extraordinary demand for AI computing can translate into sufficiently durable revenues to support the growing mountain of financing.
For Green, that question makes the evolution of AI chips from technology products into financeable assets one of the developments investors should watch most closely.
