Artificial intelligence leaders including OpenAI and Anthropic are deploying billions in their own capital to subsidize and guarantee returns for partners using their products, a form of financial engineering that echoes the dot-com bubble and obscures true market demand.
A new trend is quietly reshaping the booming artificial intelligence industry as leading firms, including OpenAI and Anthropic, are deploying billions of dollars to finance the adoption of their own products, raising concerns among analysts about the sustainability of their reported growth.
"When a seller pays customers to buy its products, it is unclear if its revenue growth reflects vibrant demand or a willingness to accept subsidies," Robert Pozen, a Senior Lecturer at MIT Sloan School of Management and a former president of Fidelity Investments, said in a recent analysis. He argues these deals muddy the distinction between sound growth and artificial financial engineering.
The scale of these arrangements is substantial. OpenAI is contributing up to $1.5 billion to a joint venture known as DeployCo, which aims to distribute its enterprise tools among companies owned by private-equity firms like TPG and Bain Capital. According to reports, OpenAI has guaranteed these partners a minimum annual return of 17.5 percent. Similarly, Anthropic is establishing a $1 billion joint venture with firms including Blackstone, contributing $200 million of its own cash. Alphabet's Google has also created a $750 million fund to subsidize the use of its Gemini models by major consulting firms.
This strategy of financing your own sales carries significant risks and draws parallels to the telecom equipment bust of the late 1990s. Companies like Lucent and Nortel lent billions to customers to purchase their equipment, only to face massive defaults and collapses when the financial environment soured. While the current AI deals are not structured as loans, they create a similar dilemma by potentially inflating revenue figures with subsidized, rather than organic, demand. For investors, this poses a critical challenge ahead of anticipated initial public offerings from these AI giants.
The current situation is reminiscent of the late 1990s telecom bubble, where equipment makers like Lucent and Nortel offered extensive financing to their customers. Lucent extended between $7 billion and $8 billion in financing, while Nortel provided over $3 billion. The strategy initially appeared successful, booking sales as revenue and loans as assets. However, when the market turned in 2000-2001, customers defaulted, leading to catastrophic losses. Lucent posted a $16 billion loss in 2001, and its stock price collapsed from a high of $84 to just 76 cents. Nortel faced a similar fate, writing down nearly $16 billion and facing SEC charges for improperly recognized revenue. The episode serves as a cautionary tale about the dangers of a company financing its own growth.
These financing deals create distorted incentives that may not serve the market in the long run. Private-equity firms, attracted by guaranteed returns, may mandate the rapid rollout of AI tools within their portfolio companies, irrespective of genuine need or fit. A recent MIT study highlighted that 95 percent of generative AI projects at surveyed companies are failing to deliver significant value, suggesting a deep disconnect between top-down mandates and on-the-ground utility. As one industry executive noted, true adoption requires letting employees become part of the process, not just secretly using their own preferred tools while the company plan falters. When consultants from firms like McKinsey or Deloitte are incentivized by a $750 million Google fund, their recommendations may be swayed by the subsidy rather than a neutral assessment of the best AI model—be it from Anthropic, Google, or OpenAI—for the client's specific problem.
As AI companies march toward potential IPOs, these practices demand intense scrutiny from investors. OpenAI is reportedly projected to lose nearly $14 billion in 2026, even with surging annualized revenue. The guaranteed 17.5% return to its private-equity partners could expose OpenAI to losses of up to $700 million a year if the joint venture underperforms. Investors must demand transparency and ask critical questions. What percentage of revenue comes from these subsidized channels? What are the renewal rates for customers not supported by these financial incentives? Are contracts tied to delivered outcomes or just traditional software pricing? Without clear answers, the market risks rewarding financial engineering over the creation of sustainable, long-term value based on genuine customer demand.
This article is for informational purposes only and does not constitute investment advice.
