Circular Financing of AI: What It Is, Why It Concerns Wall Street, and What Risks It Poses to Markets

The boom in artificial intelligence is generating increasing doubts on Wall Street about the sustainability of investments and technological valuations, in a context of high interest rates and increased financing costs.

One of the main concerns is the so-called circular financing, a mechanism by which chip manufacturers and large tech companies invest or lend money to AI companies that, in turn, use those funds to buy hardware or hire their services, thus financing part of their own demand.

Microsoft, one of the leading cloud infrastructure providers, committed $13 billion to OpenAI, the developer of ChatGPT, which in turn agreed to contract $250 billion in Azure services to train and operate its artificial intelligence models.

A similar scheme involves Amazon and Google, which invested billions in Anthropic, a competitor of OpenAI and developer of Claude. The latter committed to spending at least $110 billion on AWS and $114.5 billion on Google Cloud to access the processing capacity it needs.

In this way, the companies financing the development of AI also sell the necessary infrastructure to make it work, raising doubts about how much of that demand can be sustained without new capital contributions.

Investor concern is that this circuit is generating a demand dependent on its own financing, increasing the risks of overinvestment and indebtedness. These doubts deepened after the earnings reports from Microsoft, Alphabet, and Amazon this week, which showed a strong growth in AI infrastructure spending and increasing pressure on cash generation.

The reaction was particularly negative for Alphabet, whose shares plummeted 12.6% on Thursday, while the Nasdaq fell 1.37%. This Friday, after the results from Amazon and Intel, the tech index fell more than 3%, with heavy losses in semiconductor manufacturers like Nvidia and AMD.

Circular Financing and the Risk Behind the Growth of Artificial Intelligence

The growing dependence on financing among the companies themselves raises questions about the sector's ability to sustain its expansion without new capital contributions. Although these agreements allow for the acceleration of data center construction and the expansion of processing capacity, they can also lead to overinvestment in infrastructure if the final demand for AI services does not grow at the expected rate.

According to Tomás Sisto Bourel, research analyst at Fortress Capital, there is currently a level of circular financing higher than usual within the tech ecosystem.

"The closest comparison is probably the telecom boom of the late 90s, when the providers themselves financed their customers to accelerate infrastructure construction," he noted. However, he highlighted a difference from that period: today many of the companies involved have stronger balance sheets and cash generation.

"The risk would be that this dynamic ends up generating installed capacity far above final demand," he explained. However, he considered that this will not be the main problem, given that the potential demand for artificial intelligence is still enormous and difficult to gauge.

Román Eduardo Gutiérrez, portfolio manager at Báltico Investments, also acknowledged the risks of the mechanism, although he downplayed their magnitude. "The so-called circular financing is certainly a risk for the tech sector, but in our opinion, it is magnified," he stated.

He explained that these operations are partly a response to the way tech companies do business and to the growing competition for access to capital in the United States, where the Treasury's financing needs coexist with those of the private sector, in a scenario of inflation and high rates.

As an example, he mentioned the recent agreement between Nvidia and Meta, one of its main clients. The chip manufacturer announced an investment of $3 billion in a stock issuance from Meta, aimed at financing its expansion in artificial intelligence, while both companies deepened their alliance to expand the use of Nvidia's infrastructure.

For Gutiérrez, this type of operation, where a supplier helps finance its own client's investments, could add downward pressure on Meta's shares.

Tech Balances and Doubts About Investment Profitability

The results presented this week by Microsoft, Alphabet, and Amazon confirmed the strong growth of the artificial intelligence business, but also deepened doubts about the profitability of investments in infrastructure. The increase in spending and pressure on cash generation caused significant drops in tech stocks, especially after the earnings reports from Alphabet and Amazon.

For Sisto Bourel, the earnings reports left a heterogeneous outlook, although the operational performance of AI-related businesses continued to be solid.

Microsoft exceeded expectations, with revenues of $92.3 billion and a year-on-year growth of 48% in Azure. Alphabet, on the other hand, presented revenues practically in line with projections, while Google Cloud grew by 103%. However, revenues from the search business disappointed the market, and the increased spending on AI exacerbated concerns about its profitability. As a result, its shares plummeted by 12.6% on Thursday.

In this last case, Sisto Bourel explained that the lower earnings per share were mainly due to non-operating losses, rather than a deterioration of the core business. However, the strong growth in infrastructure spending is beginning to represent a financial challenge: Alphabet allocated $51.4 billion to capital investments during the quarter, while Microsoft disbursed $39.4 billion, amounts that are starting to pressure free cash flow.

Amazon also came under pressure after releasing its results. Although its cloud services division, AWS, increased its revenues by 39% year-on-year, to $46 billion, the market reacted negatively to the magnitude of its investments in AI infrastructure.

The company recorded a negative free cash flow of $42.4 billion in the last twelve months and projected capital expenditures close to $220 billion for 2026, which deepened doubts about how long it will take to recover those investments and translate business growth into greater cash generation.

For Gutiérrez, the results require adopting a more selective strategy, as not all tech companies have the same capacity to finance their investments.

The specialist distinguished a first group of companies with solid balances and good growth prospects, consisting of Apple, Amazon, and Microsoft. In a second segment, he placed Meta, Alphabet, and Oracle, which maintain financial strength but are beginning to show greater pressures on their liquidity and cash generation.

"These companies have financial maneuvering room, but their financial costs are rising and they will need to slow down their capital expenditure in the future," he warned.

Finally, he identified a third group composed of smaller and more aggressive companies, such as CoreWeave and Nebius, where he observes greater risks of over-indebtedness and dependence on external financing.

The difference is fundamental for evaluating investment prospects: while some tech companies can sustain AI growth with the resources generated by their own businesses, others depend more on access to credit and new capital contributions to finance their expansion.

Risks for Wall Street and Investment Perspectives

Beyond individual results, one of the main questions for Wall Street is to what extent the growing financial interconnection between AI companies could amplify a potential market correction, especially in a context of high rates and increased indebtedness.

For Sisto Bourel, this is the main risk of the current growth model. "Today, suppliers, hyperscalers, and AI companies are linked not only commercially but also through investments, financing, and guarantees," he warned.

As he explained, if any of the main participants face difficulties in financing their expansion or significantly reduce their investments, the consequences could quickly ripple through the rest of the chain, affecting both suppliers' revenues and the value of their financial stakes.

Gutiérrez also highlighted the differences in the cost of securing the debt of tech companies through CDS contracts, an indicator of the market's perception of default risk. In particular, he pointed out Oracle as one of the major tech companies with the highest credit coverage costs, reflecting greater investor concern about its financial exposure.

However, Sisto Bourel considered that these risks must be weighed against the strength of operational results. As an example, he noted that Nvidia increased its revenues by 106% year-on-year in its last quarter, while its data center division grew by 117%, demonstrating that the demand for AI infrastructure continues to accompany the high pace of investment.

Additionally, he stated that part of the concerns is already incorporated into the valuations. According to his estimates, Alphabet trades at about 20 times its projected earnings, Nvidia around 15 times, and Micron approximately five times, multiples he considers attractive given the growth prospects of these companies.

Therefore, Fortress Capital maintains a favorable outlook for investments in semiconductors and large tech companies, particularly Nvidia and Alphabet, although it warns that high volatility can quickly alter investment opportunities depending on the entry price.

From Báltico Investments, Gutiérrez also maintains a positive outlook on Wall Street and anticipates a possible advance of the S&P 500 towards 8,000-8,200 points by January 2027, following the midterm elections in the United States. However, he recommends diversifying portfolios among selected stocks, ETFs, and bonds to reduce exposure to specific sectors.

In this way, the main uncertainty for investors will be which companies will be able to sustain the growth of artificial intelligence with their own revenues and cash flows, and which will continue to depend on external financing, in a scenario where cross investments could amplify risks amid a sector slowdown.