Who’s paying whom? The circular money machine behind the AI boom

By the Editorial Team · Thursday, October 15, 2026

Leading AI companies like Nvidia, Microsoft, and Google are increasingly investing in their customers who buy their products, creating a circular financing model that drives rapid sector growth but raises concerns about market sustainability. This interconnected funding approach heightens the risk of a broader market downturn if anticipated AI demand and revenue fail to meet expectations, similar to the telecom bubble of the 1990s.

Several leading artificial intelligence companies are engaging in complex financial relationships that blur the lines between supplier, investor, and customer, raising questions about the sustainability of the AI sector’s recent growth. This circular financing model, increasingly prevalent among firms such as Microsoft, Amazon, Google, and Nvidia, involves companies investing in their customers, who in turn purchase products from the investors, creating a self-reinforcing cycle of investment and sales.

A recent central bank analysis found that more than half of all investments in AI firms—55.2%—originated from other AI companies themselves. This arrangement has helped accelerate the expansion of AI infrastructure but has also prompted concerns about the true demand fueling the industry and the potential formation of a market bubble.

Nvidia, the world’s largest chipmaker, exemplifies this trend. The company committed to investing US$30 billion in OpenAI during a private funding round and guaranteed up to US$105 billion for an OpenAI data center in Ohio. Nvidia also provides technology and financing to the company constructing the facility. These moves have solidified OpenAI as a top customer of Nvidia, but they also raise questions about whether Nvidia is inadvertently propping up demand in a tightly interlinked ecosystem.

Beyond OpenAI, Nvidia has invested up to US$10 billion in Anthropic, another AI developer, and purchased US$2 billion in shares of the cloud provider CoreWeave to support its AI infrastructure. In total, Nvidia engaged in over 50 investment deals with AI firms in 2024, many structured over multiple years or still open-ended, facilitating quicker build-outs than might otherwise be possible.

A regional example of this circular financing model is found in Firmus Technologies, an Australian data center builder. Last month, Firmus abruptly withdrew plans for an initial public offering on the Australian Securities Exchange. Nvidia is simultaneously Firmus’ investor, supplier, and customer. Following Nvidia’s initial investment in a A$330 million capital raising in September 2025, the chipmaker participated in subsequent funding rounds, including Firmus’ US$2 billion round in August 2026. Firmus uses Nvidia’s DRS AI factory platform for its facilities and has signed Nvidia as a customer for a data center in Batam, Indonesia, with contracts worth up to US$30 billion over six years. OpenAI is also an anchor customer at Firmus’ Malaysian data centers.

This tight financial integration raises concerns about risk concentration. A financial setback for one company, such as OpenAI, could ripple through connected firms, especially given questions surrounding OpenAI’s reported revenues. Its shares declined recently after reports indicated its annual revenues might be US$20 billion lower than previously suggested.

The Reserve Bank of Australia highlighted such risks in a recent Financial Stability Review, pointing to “pockets” of higher-risk firms utilizing opaque or circular funding mechanisms where chip manufacturers extend financing to their customers. The RBA cautioned that a shift in investor sentiment toward the AI sector—fueled by expectations of rapid earnings growth and increasingly complex debt cycles—could trigger a broader market correction.

Wall Street’s performance in 2026 has been heavily influenced by AI-related stocks, with semiconductor companies accounting for approximately 70% of gains in the S&P 500 index. Nvidia alone comprises about 8.06% of the index, making the market particularly sensitive to developments affecting the AI industry.

Industry stakeholders argue that the AI boom reflects genuine innovation and demand, but the timing of expected revenue growth remains uncertain. Nvidia, OpenAI, and Firmus are wagering that future business and consumer use of AI will generate sufficient income to justify current large-scale investments in data centers and chips. Until then, the sector relies heavily on internal financing to bridge the gap.

The situation has drawn comparisons to the telecommunications boom of the 1990s, when equipment manufacturers like Cisco Systems financed customers to build extensive infrastructure. Although the overbuilt fiber networks eventually proved valuable, many investors suffered significant losses when that bubble burst. Unlike fiber, AI chips typically have shorter useful lifespans of only a few years, increasing pressure on companies to meet revenue expectations in the near term.

If demand growth does not materialize as anticipated, shareholders and institutional investors could face substantial losses, underscoring the risks implicit in the current AI financing model.

This article was reported by synthesizing coverage from multiple independent news outlets to present a balanced, unbiased account. Read about our editorial standards. Spotted an error? Let us know.