Nvidia, OpenAI, Oracle, Meta, Amazon and Anthropic have built a complex web of agreements worth an estimated $1.8 trillion, spanning chip supplies, investments, leasing deals and financial guarantees. The network is raising questions over how solid the financial figures underpinning the artificial intelligence boom really are.
Leading US AI companies have signed a growing number of interconnected agreements in recent months, ranging from chip supply and equity investments to data-center leases and loan guarantees. Valued at least $1.8 trillion in total, the web of deals is blurring the lines between companies and prompting investors to take a closer look at their balance sheets.
Nvidia’s latest move
The latest development to fuel the debate is a deal reportedly being considered by Nvidia.
According to The Wall Street Journal, the chipmaker is preparing to provide a $250 billion loan guarantee to help finance a massive data center planned in Ohio. The facility is expected to be leased by OpenAI, the developer of ChatGPT, in which Nvidia also holds a stake. Financing for the project is reportedly being arranged by Japanese technology group SoftBank, another OpenAI investor.
The data center is expected to cost more than $500 billion, with Nvidia also set to supply the chips needed for the facility. The guarantee from Nvidia, which has a market capitalization of around $4.76 trillion, is intended to lower the project’s borrowing costs.
Markets grow cautious
The increasingly intertwined financial relationships between the companies are nevertheless raising concerns among investors.
Following reports of the $250 billion guarantee, Nvidia’s five-year credit default swap (CDS) premium reached a record high. The cost of insuring against the company’s default rose from 0.14% a year to 0.82%.
While the figure remains low compared with many other sectors, the increase suggests that investors’ confidence in the AI industry is not unlimited. Markets are now looking beyond promises of hundreds of billions of dollars in investment and revenue to assess how quickly AI is actually being adopted and whether the industry can generate enough profits to justify its enormous spending.
All eyes on earnings
The financial results of major technology companies due in the coming months will provide an important test for the industry.
Raphael Thuin, head of capital markets strategies at French investment firm Tikehau, said investors are focusing on two key questions: Is the pace of AI infrastructure spending beginning to slow? And can companies provide tangible evidence that their investments are generating returns?
Alphabet is so far the only major technology company to have reported its results. The company said its Gemini AI platform has reached around 1 billion users. At the same time, however, Google’s parent company recorded more than $5 billion in cash outflows for the first time in its history.
The investment race continues
Alphabet has said it plans to spend more than $200 billion in 2026 alone on chips, data centers, servers and personnel as it seeks to maintain its growth in AI.
As part of the effort, the company has raised $80 billion in capital in a short period and has also taken on between $50 billion and $70 billion in debt.
Alphabet is not alone. Other major technology companies are making investments on a similar scale. The sector invested around $400 billion in data centers and technology infrastructure in 2025, a figure expected to rise to at least $800 billion in 2026.
Investors, however, are increasingly questioning whether spending on such a scale is truly necessary.
Chinese pressure intensifies cost debate
Another factor adding to the debate is the emergence of lower-cost AI models from Chinese companies.
Moonshot AI’s Kimi K2.6 model has been priced at $1.71 per million tokens, while OpenAI’s most advanced model, GPT-5.5, costs around $11.25 per million tokens.
According to BlackRock analysts, the cost of AI usage has become one of the industry’s biggest concerns. Gartner estimates that global spending on AI models and platforms will reach $64 billion in 2026, representing a 63% increase from the previous year.
As costs rise, companies are expected to shift toward cheaper models, potentially forcing US AI companies to cut their prices. For OpenAI, Anthropic and other companies that need to generate returns on hundreds of billions of dollars in investment, however, lowering prices may prove difficult.
Who will emerge as the winner?
According to BlackRock, the competition in AI is no longer simply about developing the “best model”. The real battle is increasingly about who will capture the largest share of the economic value created by the technology.
Analysts say falling AI model prices could change the industry’s winners, potentially making infrastructure providers such as Nvidia a safer investment than model developers such as OpenAI and Anthropic.
But if low-cost Chinese models significantly erode the market share of US developers, an even bigger question will emerge: how resilient is the roughly $1.8 trillion financial network currently supporting the AI industry?
For investors, the fundamental question is whether this enormous structure has been built on solid foundations — or whether it is a house of cards waiting to collapse under the first major shock.
Be First to Comment