NVIDIA: How Much of Its $200B+ AI Story Is Built on Revenue We Won’t Understand?
Jensen Huang’s handshake deals, billions flowing through Singapore, a history of disclosure problems and an increasingly circular AI financing ecosystem raise an uncomfortable question: are investors seeing the full picture behind NVIDIA’s extraordinary growth?
NVIDIA (NASDAQ: NVDA) has become the financial symbol of the AI revolution. Its chips power the data centres being built by the world’s largest technology companies, and its revenue growth has been extraordinary.
But there is a question investors should be asking beyond how many GPUs NVIDIA can sell:
How much of this reported demand is independent, transparent and ultimately sustainable?
I’m not suggesting NVIDIA has fabricated its revenue. That would be a serious allegation, and the available evidence does not establish it.
The more interesting concern is whether the headline numbers tell the whole economic story — particularly when customers, suppliers, investors and infrastructure providers are increasingly financing one another.
There are several reasons to look more closely.
1. The crypto-mining precedent: when Gaming revenue wasn’t quite what it seemed
This isn’t the first time NVIDIA’s revenue mix has raised disclosure questions.
In May 2022, the SEC announced that NVIDIA had agreed to pay a $5.5 million penalty over inadequate disclosures concerning the impact of cryptocurrency mining on its Gaming business.
During fiscal 2018, demand for crypto-mining GPUs contributed significantly to Gaming revenue growth. However, NVIDIA failed to disclose the extent of that contribution in certain quarterly filings.
The issue wasn’t that the sales didn’t exist. It was that investors weren’t given enough information to understand what was driving them — or how sustainable that demand might be.
That distinction matters today.
Crypto mining created a temporary surge in GPU demand. AI is a much larger and potentially more durable opportunity, but investors should still distinguish between genuine end-user demand and demand that depends on financing, intermediaries or temporary market conditions.
History doesn’t prove today’s AI revenue is misleading. It does demonstrate why the quality and transparency of revenue matter.
2. Singapore: a huge revenue number with an even bigger question
One of the most intriguing developments concerns NVIDIA’s Singapore business.
According to the Financial Times, NVIDIA’s Singapore subsidiary recorded approximately $104 billion in revenue during 2025, rising to approximately $117.8 billion over the 12 months to January 2026.
Yet its reported profit was only around $53 million.
These figures need context.
Singapore is a major international invoicing and trading hub. The revenue does not mean that all those chips were physically delivered to Singapore. NVIDIA had previously indicated that less than 2% of the revenue attributed to Singapore was associated with physical shipments there.
In other words, the billing location and the final destination of the products can be very different.
But that distinction creates a legitimate investor question: how much of the demand attributed to a particular market is actually being generated and consumed there?
The issue becomes more interesting because NVIDIA subsequently stopped separately disclosing Singapore sales after changing its geographic reporting approach.
This occurred amid scrutiny of Singapore-based Megaspeed, a regional customer investigated by US authorities over suspected GPU diversion and possible links to shipments reaching China.
There are important qualifications here. An investigation is not proof of wrongdoing, and there is no established evidence that NVIDIA knowingly booked illegal smuggling activity as revenue.
Nevertheless, the combination of enormous Singapore-linked revenue, limited physical shipments, a very small reported profit and reduced geographic transparency deserves scrutiny.
Investors should not have to guess where demand is ultimately coming from.
3. Jensen Huang’s handshake deals: trust or transparency risk?
Another intriguing detail emerged from Elon Musk.
In November 2026, Musk said that many important NVIDIA supply-chain arrangements were based on personal trust with Jensen Huang — describing them as a handshake rather than a written contract.
Musk also indicated that purchase orders could follow later.
This is not evidence of accounting fraud. Large businesses regularly use preliminary commitments, negotiations and commercial relationships before final contracts are signed.
But it raises an important question about how investors interpret demand.
A handshake commitment is not necessarily the same thing as a firm, enforceable purchase order. Nor does a planned allocation of GPU capacity automatically guarantee that the customer will ultimately take delivery, pay for it and generate a return from using it.
If significant parts of the AI supply chain depend on relationships and commitments that are not yet fully documented, investors need to understand the difference between expected demand and contracted revenue.
The distinction becomes especially important when valuations assume years of uninterrupted growth.
4. The circular AI financing problem
Perhaps the biggest structural concern is the growing relationship between AI investment and AI chip sales.
Consider the basic cycle:
* NVIDIA invests in or supports an AI company.
* That company uses capital to expand its computing capacity.
* It purchases or rents GPUs and infrastructure.
* NVIDIA and its partners benefit from the resulting demand.
* The AI company must eventually generate enough revenue from its own customers to sustain that spending.
This is not automatically improper. Strategic investments can accelerate innovation, create new markets and help customers build businesses that would otherwise struggle to obtain funding.
But it can make demand harder to evaluate.
The Congressional Research Service highlighted this issue in its November 16, 2026 analysis of circular financing in the AI industry.
It described arrangements in which chipmakers and infrastructure providers invest in AI businesses that then use some of that capital to purchase computing services and equipment from those same investors.
The Federal Reserve has also discussed the financial-stability implications of these relationships, including the risk that AI companies may eventually need to replace supplier-backed funding with revenue from paying customers.
That is the critical point.
Revenue generated by a customer spending money supplied or supported by its own supplier may be economically different from revenue generated by an independent customer with a profitable business.
The sale may be genuine. The accounting may be correct. But the underlying demand could be less robust than the headline figures suggest.
The real test is whether those AI businesses can eventually stand on their own.
5. What investors should be watching
There are several indicators that could help separate genuine long-term demand from a potentially fragile spending cycle.
Independent customer demand: Are AI customers generating enough external revenue to fund their infrastructure spending without relying heavily on fresh investment?
Receivables and cash conversion: Is reported revenue turning into cash at a healthy rate, or are payment terms and receivables becoming increasingly important?
Customer concentration: How much demand depends on a relatively small number of hyperscalers, AI laboratories and infrastructure providers?
Geographic transparency: Can investors clearly understand where NVIDIA’s products are being billed, shipped and ultimately deployed?
Firm orders versus expectations: How much future demand is supported by enforceable contracts rather than preliminary commitments, strategic relationships or verbal assurances?
Capital commitments: Are investments, guarantees and financing arrangements creating demand that might weaken if external funding becomes more expensive?
None of these indicators alone proves that NVIDIA’s revenue is overstated. Together, however, they can tell investors considerably more about the durability of its growth.
6. The bull case: why NVIDIA could still be an extraordinary investment
There is a very strong counterargument.
AI adoption is real. Data-centre construction is accelerating, computing requirements are expanding, and major technology companies are investing heavily in infrastructure.
NVIDIA has a powerful competitive position, an extensive software ecosystem and a product portfolio that remains central to much of the AI build-out.
Circular financing can also be commercially rational. Early-stage AI companies need capital, infrastructure providers want customers, and strategic investment can help establish an ecosystem that eventually generates substantial independent demand.
It would be a mistake to assume that every investment-backed purchase is artificial or that every handshake arrangement is suspicious.
The long-term opportunity could still be enormous.
But an excellent business and an excellent investment are not necessarily the same thing at every valuation.
7. My verdict: the question isn’t whether NVIDIA sells chips
NVIDIA clearly sells enormous quantities of advanced computing hardware. The more difficult question is how much of the current AI spending cycle represents durable economic demand.
The crypto-mining disclosure case reminds us that the composition of revenue matters.
The Singapore figures raise questions about geographic transparency and the difference between billing location and physical deployment.
Musk’s comments highlight the role of personal relationships and informal commitments in securing supply.
And circular financing raises the possibility that some AI demand is being supported by capital flowing around the same ecosystem that benefits from the resulting sales.
None of this proves that NVIDIA is manufacturing revenue. But investors should be careful not to confuse reported sales with independently funded, sustainable end-user demand.
My biggest concern isn’t that NVIDIA’s revenue is fake. It’s that investors may not fully understand what is funding the demand behind it — or how much of that demand will survive when the investment money eventually has to generate real returns.
That is the revenue we won’t understand until the money starts flowing back from the end customers.
And by then, the market may already have repriced the stock.
What do you think? Is NVIDIA’s AI demand fundamentally stronger than anything we’ve seen before, or are circular financing, opaque customer relationships and geographic reporting creating risks that investors are underestimating?
@Tiger_comments @TigerStars @TigerClub
Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.
- chocoee·09:32IV near 69% says the market is pricing uncertainty hard. If demand is truly clean, that premium has to collapse eventually1Report
