Nvidia’s founder, Jensen Huang, has lost billions of dollars on paper. It is that financial markets are becoming increasingly sensitive to any evidence that the enormous investments in artificial intelligence may not generate revenues quickly enough to justify the valuations placed on AI companies.
This directly reinforces the concern about overvaluation that we discussed earlier: the AI revolution can be real and economically transformative while the prices investors pay for companies associated with it are nevertheless excessive.
1. What happened?
Nvidia shares
−3%
Closing price: $230.48
Jensen Huang's estimated wealth
$199.3bn
Down approximately $6.1bn in one day
Oracle shares
−5.6%
Larry Ellison lost about $8.9bn
Source: Forbes, 8 October 2026. These are reported market movements and estimated paper wealth, not realised investment losses.
forbes.com
+1
The immediate trigger was a report that OpenAI's annualised revenue was approximately $50 billion at the end of September, rather than the previously reported $68 billion figure. The difference was reportedly related in part to how partner revenues were counted and compared with competitors. The figures should therefore not automatically be interpreted as evidence that OpenAI's underlying business had suddenly deteriorated.
forbes.com
+1
Nevertheless, the market reaction was substantial. Nvidia and Oracle are both closely connected to OpenAI's expansion: Nvidia supplies its AI processors, while Oracle provides computing infrastructure. Investors consequently reassessed not just one company's revenue, but the prospects of a much wider network of AI-related investments.
There is a revealing contrast here: Nvidia reported quarterly revenue of $96.2 billion, up 106% year on year, with $89 billion coming from its data-centre business. Yet even exceptional growth did not insulate its share price from doubts about the economics of the broader AI ecosystem.
Forbes Australia
2. Why this matters for the overvaluation argument
I would connect this episode to our earlier analysis of the AI investment boom through four related risks.
A. Investment is running ahead of demonstrated returns
Technology companies are committing enormous sums to chips, data centres, electricity supply and cloud infrastructure. The critical question is whether the revenues and profits generated by AI applications will ultimately justify that investment. Strong demand for computing capacity today does not, by itself, guarantee adequate returns on all the infrastructure being built.
B. The same expectations are embedded in many valuations
Nvidia may be a highly successful business, but its valuation also reflects expectations about future AI spending. The same spending supports data-centre operators, cloud providers and other technology companies. If expected AI expenditure disappoints, several valuations can come under pressure simultaneously.
C. A small change in expectations can produce a large change in value
When investors price a company for exceptional future growth, even a modest revision to revenue expectations can trigger a disproportionate share-price decline. The issue is not necessarily that the company is bad, but that its previous price may have left too little room for disappointment.
D. The risk extends beyond listed technology stocks
Data centres require capital-intensive construction, power infrastructure and financing. If anticipated demand or prices weaken, investors and lenders may discover that the underlying assets produce lower returns than projected. The consequences can spread to private equity, infrastructure funds, lenders and institutional investors.
3. The most important issue: the circular investment dynamic
The deeper question is whether the AI ecosystem is creating enough independent economic value to support its investment cycle.
1. Massive investment in AI infrastructure
Chips, data centres, cloud computing and power
2. AI companies and customers need more capital
Revenue growth and future demand support new financing
3. Infrastructure suppliers report exceptional growth
Strong orders reinforce confidence in the entire ecosystem
4. The critical test: final customers' economics
Are end users generating enough recurring revenue, productivity gains and profits to sustain the expenditure?
This is not proof of a bubble. Investment cycles routinely involve infrastructure being built ahead of demand, and Nvidia's actual revenue growth is substantial. The concern is that the financial success of the suppliers may be easier to demonstrate today than the eventual returns on the whole system's investment.
If AI customers eventually generate strong, sustainable profits, the current investments may be justified. If the commercial returns disappoint, the market could face excess capacity, falling prices, reduced capital expenditure and significant valuation adjustments.
4. An additional warning: the risk is visible in private markets and IPOs too
A related development reported by Reuters on 9 October strengthens the argument. Firmus Technologies, a data-centre developer backed by Nvidia and Blackstone, withdrew its planned IPO after seeking a valuation of nearly A$44 billion—about US$31 billion—which would have been roughly three times its August funding-round valuation. The company cited market volatility and prevailing conditions.
Reuters
The withdrawal does not prove that the proposed valuation was objectively wrong. But it illustrates an important distinction:
Private-market valuations can rise rapidly during funding rounds, especially when investors compete to gain exposure to a fashionable sector.
Public-market valuations must withstand a broader set of investors examining the business, the price, the financial structure and the potential exit.
An IPO that cannot attract sufficient investor confidence can reveal a gap between the price sellers hope to obtain and the price buyers are prepared to pay.
This is particularly relevant to our earlier discussion of AI infrastructure investment, including the role of large institutional capital and data-centre financing. The key risk is not restricted to speculative technology start-ups. It can also emerge in capital-intensive businesses with tangible assets if their valuations assume unusually strong growth and high utilisation rates.
5. The conclusion
I would avoid saying that the fall in Nvidia's share price proves that the AI market is a bubble. One trading session cannot establish that, and the reported revenue discrepancy itself requires careful interpretation.
A more robust conclusion is that the market is beginning to test the assumptions underpinning the AI investment boom. The distinction between technological potential and financial value is becoming increasingly important.
AI: the danger is not the technology, but the valuation
The recent fall in Nvidia shares, following questions about OpenAI's reported revenue, illustrates a fundamental risk in the current AI investment cycle. The technology may transform entire industries, and companies such as Nvidia may continue to achieve extraordinary growth. But neither technological progress nor exceptional revenue growth automatically justifies every valuation placed on the companies supplying the infrastructure.
The deeper concern is that enormous investments in chips, data centres, computing capacity and energy infrastructure are being justified by expectations of future AI demand and profitability. These expectations are interconnected: a reassessment of one major AI company's prospects can affect the perceived value of suppliers, infrastructure operators and their financiers.
The decisive test will be whether the economic returns generated by AI applications ultimately justify the capital committed to building the ecosystem. If they do, today's investments may prove well founded. If they do not, the adjustment may extend far beyond a handful of technology shares.
The danger, therefore, is not necessarily that AI is another technological illusion. It is that a genuine technological revolution can still produce excessive valuations, overinvestment and substantial financial losses when expectations of future returns become detached from the economics of the businesses expected to deliver them.