OpenAI has not completed raising about $18 billion in initial funding for a custom artificial intelligence chip project it is pursuing with Broadcom, a report said. With competition to invest in AI infrastructure overheating, the difficulty of raising capital to back mega-projects is increasingly seen as a signal.

Blockchain media outlet BeInCrypto reported on Wednesday local time that the funding gap emerged at an early stage of OpenAI’s in-house chip strategy. OpenAI is pushing to develop its own AI accelerator to reduce its reliance on Nvidia, with Broadcom taking part as a key partner.

The project is part of a plan to build a total of 10 gigawatts of AI accelerator infrastructure over several years. The two companies previously unveiled a roughly $500 billion hardware expansion plan in October 2025 that included securing chip, network, power and data centre capacity. The industry has viewed it as one of the largest infrastructure investments of the AI era.

The problem is that capital is not keeping pace with the speed at which the investment plans are growing. BeInCrypto analysed the case as the latest example of AI infrastructure deals exceeding the financial market’s capacity to absorb them.

OpenAI’s financial burden is also growing. Despite rising revenue, the company has been discussed as possibly burning through a cumulative about $115 billion through 2029. Reports have also said it recently failed to meet internal growth targets, adding to views that investor scrutiny has become tougher.

Funding pressure on partners is also becoming reality. Oracle issued about $18 billion of corporate bonds last September to support OpenAI-related projects. It is cited as an example showing that the cost of building AI infrastructure is growing beyond the level individual companies can bear.

Market research firms forecast global hyperscalers’ capital expenditures in 2026 to reach about $600 billion to $720 billion. About 75 percent is expected to be concentrated on AI infrastructure. But in the financial sector, there is concern about a gap between expectations for AI demand and the pace of actual cash recovery.

This trend is also being detected in Nvidia’s financial indicators. Nvidia said recently that customers’ unpaid balances are rising sharply, and accounts receivable are reported to have increased to about $33 billion. Some investors are also raising doubts about the process by which AI infrastructure contracts translate into actual cash flow.

Scepticism is also spreading about the mega AI deals being announced. Some investors said announcements of multi-billion-dollar deals among OpenAI and Broadcom, AMD and Nvidia are being used to build expectations rather than focused on whether they will actually be executed.

OpenAI has few solutions to choose from. Options being discussed include restructuring the initial fundraising, securing new investors, or reducing the scale of chip deployment. Any choice could affect the volume of 2026 AI capital spending forecasts that translates into actual execution.

As aggressive infrastructure investment across the AI industry continues, the ability to actually deploy capital is emerging as a key market variable.