OpenAI, the leading company in the development of artificial intelligence, is facing enormous infrastructure needs – from data centers to specialized chips – in order to maintain the pace of development of its technologies. However, instead of taking on the financial burden of this expensive construction itself, OpenAI has managed to finance its ambitious infrastructure projects through the balance sheets of its partners, who are rapidly taking on debt on its behalf. It is estimated that the total loans and bonds connected to OpenAI-related projects are approaching nearly USD 100 billion, with the debts nominally borne by the partners, while OpenAI benefits without direct financial risk.

Partners take on billions in debt for AI infrastructure

To secure the computing power required by OpenAI, large partners and investors are investing tens of billions of dollars through loans and joint projects. Some of the key actors and their contributions are:

SoftBank:
This Japanese investment group raised about USD 20 billion during 2023 for investments in AI – mostly directed toward OpenAI. SoftBank also took out a USD 8.5 billion bridge loan for recapitalizing OpenAI (of which USD 1 billion has already been repaid). Through a partnership called Stargate, SoftBank together with OpenAI plans to develop several large AI data centers in the U.S., as part of an infrastructure venture worth up to USD 500 billion for about 10 gigawatts of capacity by the end of 2025.

Oracle:
Technology giant Oracle has become a key cloud partner of OpenAI. It has already issued USD 18 billion in corporate bonds to finance the construction of data centers for OpenAI. Analysts estimate that Oracle may have to borrow a total of around USD 100 billion over the next four years to fulfill its obligations toward OpenAI. Through the Stargate project, Oracle is building several hyperscale data centers (e.g., in Texas, New Mexico, and Wisconsin) that will provide thousands of NVIDIA GPUs and up to 4.5 GW of new computing power for OpenAI.

CoreWeave:
This is a specialized cloud infrastructure provider based on GPU technology. CoreWeave has taken on more than USD 10 billion in debt to expand capacity – leasing data centers and purchasing graphics chips. In return, CoreWeave has large contracts to supply computing power to Microsoft (Azure), and part of those resources is indirectly used by OpenAI through Microsoft’s cloud services. OpenAI has reportedly signed contracts worth more than USD 22 billion with CoreWeave for training AI models on its infrastructure.

Other infrastructure partners:
An entire ecosystem of specialized companies participates in financing OpenAI’s infrastructure. Investment fund Blue Owl Capital and Crusoe Energy formed a joint venture that took around USD 10 billion in loans from JPMorgan to build the first OpenAI data center in Texas – the loan will be repaid through Oracle’s long-term lease of that center. Blue Owl separately took an USD 18 billion loan (mostly from Japanese banks) for another large center in New Mexico, which Oracle leases for OpenAI’s needs. In addition, a consortium of banks led by Oracle’s partner Vantage is negotiating a syndicated loan of USD 38 billion for new facilities, bringing total indebtedness associated with OpenAI projects close to USD 100 billion.

Using other people's balance sheets as a growth strategy

This financing model has allowed OpenAI to expand globally at lightning speed without taking on debt on its own balance sheet. OpenAI as a company currently has almost no debt – sources state that it has an unused credit line of about USD 4 billion, but it has not yet been tapped. While OpenAI operates at a loss and aggressively reinvests in development, the huge costs of building data centers, acquiring equipment, and interest on loans remain borne by its partners and financial institutions, in line with the informal principle: “OpenAI orders, partners pay.”
In this way, OpenAI keeps its balance sheet “light,” while at the same time securing long-term contracts guaranteeing access to necessary infrastructure.

The motivation for this approach lies in the incredible demand for computing power. This year, OpenAI committed to using as much as USD 1.4 trillion in cloud and hardware resources over the next eight years, corresponding to capacity greater than 36 gigawatts – roughly equal to the combined output of dozens of the world’s largest data centers. This aggressive pre-ordering of infrastructure services is intended to prevent bottlenecks such as chip and server shortages that currently limit user growth and the pace of advanced-model training. In other words, OpenAI is using the financial strength and balance sheets of its partners to accelerate its development, while promising partners a large volume of future business in return.

Advantages and challenges of this approach

The strategic “reliance on other people’s balance sheets” brings many advantages for OpenAI. First of all, the company has managed to avoid enormous upfront investment in assets – data centers, electrical infrastructure, and expensive chips – which it would struggle to finance as a young firm. Instead of taking on debt or spending its own capital, OpenAI has gained access to world-class infrastructure through partnerships, at the expense of others. The risk of financing and amortizing the equipment has been transferred to larger companies (such as Oracle) or investors who have the capacity to carry it.
At the same time, OpenAI gains speed – it can increase global computing capacity faster than if it were constrained by its own budget. This model has also encouraged creative partnerships: for example, the Stargate project combines OpenAI’s expertise in AI software with Oracle’s cloud infrastructure and SoftBank’s capital, creating synergy for accelerated development of the next generation of AI data centers.

However, there are also clear challenges and risks. The enormous debt accumulating around OpenAI means a significant dependence of many actors on the success of this one company. Financial experts on Wall Street warn that concentrating so much leverage on a single client carries systemic risk. If OpenAI’s revenues fall short of optimistic forecasts, the partners who have taken on the most debt may be the first to face liquidity pressure.
Specifically, relatively smaller providers such as CoreWeave, which have taken out large loans because of contracts with OpenAI, could encounter problems if demand for services does not grow at the expected pace. Similarly, even technology giants like Oracle rely on continued cooperation with OpenAI to justify their large investments – any slowdown in OpenAI’s development or a move to another platform could leave those capacities underutilized.

In addition, high interest rates in the current environment make these loans more expensive, increasing pressure on partners to deliver results quickly. It is also important to remember that although OpenAI does not directly owe these loans, it bears obligations through contracts to use and pay for this infrastructure. If market conditions were to change (e.g., saturation of AI services or regulatory slowdowns), the cost burden could indirectly affect OpenAI through contract renegotiations or investor pressure.

Despite these risks, this debt-driven undertaking has so far proven to be a successful engine of growth. Thanks to it, OpenAI has become the world’s most valuable private tech company (valued at around USD 500 billion) and continues to attract capital and partners willing to take on shared risk in exchange for a stake in the AI revolution.

In short, OpenAI has managed to build planetary-scale infrastructure without significant indebtedness by shifting that burden onto the shoulders of its powerful allies. This innovative financial model has allowed the company to focus on its mission of developing advanced intelligence, while continuing global expansion powered by partnerships and debt – an approach that could yield enormous rewards, but one that all participants will watch closely due to its potential risks.

/ / /

"Standard Prva" LLC Bijeljina is a company registered in Bijeljina at the District Commercial Court in Bijeljina. Company’s activities are accountancy, repurchases of receivables, angel investing and other related services. Distressed debt is a part of the Group within which the company repurchases the receivables, which function and are not returned regularly.

Lawyer’s Office Stevanović is the leading lawyer’s office in the region with the seat in Bijeljina. The LO abbreviation represents Lawyer’s Office of Vesna Stevanović and Lawyer’s Office of Miloš Stevanović.

Contact for media press@advokati-stevanovic.com or via telephone 00 387 55 230 000 or 00387 55 22 4444.