AI Circular Deals

Photo by Steve Johnson on Unsplash

What is a “Circular” Deal? 

If you keep up with tech or finance news, you may have heard the word "circular” being used to describe a number of recent transactions between major AI and AI-adjacent companies. In a typical circular deal, one company sells a product/service to another, but instead of paying cash, the buyer finances the purchase by giving or promising an equity stake to the seller, resulting in the seller’s own capital effectively helping fund its sales. Moreover, the money exchanged essentially cycles between the two. Thus, a company can simultaneously project sales and investment, which in turn, inflates each company’s profits and valuation. Transactions can be further complicated when a company that wishes to project outside investment forms a separate legal entity, like a special purpose vehicle (“SPV”), to execute these transactions. 

Recently, OpenAI has paired up with Nvidia and Oracle in a mutually reinforcing ecosystem of computing power, hardware, and infrastructure. Nvidia is supplying massive quantities of its AI-optimized chips to OpenAI and has committed up to $100 billion in investment to support the deployment of at least 10 gigawatts of Nvidia systems for OpenAI’s data centers. Oracle is simultaneously expanding data center capacity through its Stargate project that will house the Nvidia chips that serve OpenAI’s LLMs and other products like Sora.

 In short, money is moving in a loop among the three: Nvidia invests in and sells its GPUs to OpenAI while OpenAI uses that hardware to build and train models hosted on Oracle’s cloud infrastructure. In turn, Oracle is spending billions on Nvidia chips to expand its data centers, which will host OpenAI’s software–an arrangement that OpenAI says could generate $300 billion in payments to Oracle over the next five years. 

Bubble Risk 

The primary fear among investors is that these convoluted deals are creating an “AI bubble”, but what exactly does this mean? A bubble occurs when an asset’s value rapidly increases above its intrinsic or fundamental value because speculation continues to inflate the price of the asset despite underlying economic realities. The dotcom bubble of the early 2000s and the housing bubble of 2008 are two examples of a boom-and-bust cycle in the last 25 years. 

Recently, the valuations of AI-related companies such as chipmakers, data center builders, and software developers have skyrocketed. Nvidia’s market cap surpassed $3trillion, while Oracle is now valued at over $800 billion. Private companies like OpenAI and Anthropic have been given multi-billion-dollar valuations despite limited revenue. These valuations resemble the dotcom bubble, where huge capital investments were made based on projected future market dominance rather than current profitability. Circular deals fuel the fear of a bubble because they tend to obscure true demand for equity in these companies, which may be artificially inflating their market valuations. 

AI-related companies like Nvidia are overwhelmingly responsible for recent stock market gains and record highs. In fact,  56.6% of S&P 500 gains over the last year have come from AI and account for all but 0.1% of GDP growth through the first half of 2025. Our current economy is floated by the success of AI stocks, which depend on rapid growth and the ability to build out AI infrastructure. The demand for rapid growth and infrastructure has motivated companies like Coreweave and Oracle to make opaque “circular” deals. As a result, many investors speculate that these joint ventures and intra-firm financing arrangements are dangerously underpinning the market. If the AI bubble were to burst as a result of circular deals creating liquidity issues or scaring off investors, the pop would have ripple effects across the market. 

Compliance and Anti-Trust Concerns

When entering these circular deals, companies and their legal departments have much at stake. Any time a company enters a deal, it must document and disclose to its investors and ensure policy compliance. Working within a small ecosystem can make doing so complicated.

Deals like Nvidia’s investment in OpenAI in exchange for a sales commitment open up new compliance requirements. These deals are Related Party Transactions, which occur when relationships form between two connected parties. These relationships can occur through family members, close associates, or–the most significant relationship in this case–significant shareholders. In these transactions, companies must disclose their relationships, the cost of their transactions, and even nominal exchanges in their financial reporting. 

Investors will be keen to know about these transactions and whether or not companies remain compliant in disclosing them. Parties that fail to properly disclose their related party transactions can face sanctions by multiple regulatory bodies. Failure to disclose can also mislead investors about the actual value of a company–a result that could lead to derivative actions by shareholders. Even when these transactions are disclosed, the sheer amount of circular deals can inflate market values and create misleading information. 

In addition to regulatory and derivative action risks, circular deals may invite securities fraud claims. Section 10(b) of the Securities Exchange Act of 1934 and SEC Rule 10(b)-5 make it unlawful to “omit to state to a material fact” concerning a publicly traded stock. If a firm reports strong revenues without disclosing that much of it comes from partners it also finances or invests in, the omission could be materially misleading under  SEC v. Texas Gulf Sulphur Co., because investors might otherwise assume the growth reflects genuine, external demand. 

Considering how much some companies are investing in AI vendors, stakeholders may expect strong promises of regulation and compliance. But the data shows that those expectations might be unrealistic. As few as 17% of AI vendors promise comprehensive regulatory compliance–and it is their buyers who may face the liability. 88% of AI vendors use liability caps with their buyers, putting the burden on the companies that purchase their products and invest in them. 

There are also antitrust concerns with these circular deals. Investments like those from Nvidia to OpenAI are being made at a massive scale. Massive enough that they may prevent any new competitors from working with AI vendors. These deals create closed ecosystems between buyers and vendors and establish exclusive dealing. Exclusive dealings as a whole are not illegal, but they can violate antitrust laws if they prevent competitors from competing effectively. Companies engaging in circular deals must ensure that they are not creating a foreclosure of their verticals, and vendors must ensure they are not locking customers in at the risk of closing off any competition.

*The views expressed in this article do not represent the views of Santa Clara University.