The biggest challenge to the further expansion of AI infrastructure is no longer the availability of chips or data centers, but rather the financing of these projects. This is the conclusion reached by analysts at SemiAnalysis in their report, “Nvidia GPU debt backstop unleashes the AI project trinity.” According to the research firm, the AI sector is evolving into a credit market with more than $7 trillion in outstanding debt by 2029, while total investment in AI infrastructure will reach approximately $11.1 trillion between 2024 and 2029.
According to SemiAnalysis, this will shift the bottleneck once again. In 2025, data centers limited AI growth; by early 2026, GPU availability was the main constraint, but the next phase will revolve around access to capital. The researchers expect annual investments in GPUs, storage, networks, CPUs, and data centers to exceed $2 trillion by 2028. As a result, credit markets will play a central role in financing new AI projects.
AI project trinity
SemiAnalysis describes the challenge as “AI Project Trinity.” For a new AI cluster to be established, three conditions must be met simultaneously: financing, a purchase guarantee for computing power, and available data center capacity. In practice, however, these three factors tend to block one another. Banks are unwilling to provide financing until there is a buyer, buyers want assurance that the infrastructure will be built, and data center operators demand financial guarantees before making capacity available.
According to SemiAnalysis, Nvidia is attempting to break this deadlock by acting as its own financial guarantor. The company is using its strong balance sheet to offer lenders greater certainty, enabling even smaller GPU cloud providers to secure financing. The researchers, therefore, even describe Nvidia as a future “central bank of AI.”
Network World supplements that analysis with details about the arrangement. Nvidia offers so-called “neoclouds” a minimum revenue guarantee: if an operator is unable to fully lease out its GPU capacity, Nvidia will purchase a portion of that capacity itself. If revenue exceeds the guaranteed level, Nvidia shares in those additional earnings. Thanks to this guarantee, banks are more willing to provide billion-dollar loans for new GPU clusters.
More room for smaller AI players
According to SemiAnalysis, this approach could significantly expand the AI market. Until now, most large-scale GPU projects have been financed only through five-year purchase agreements with hyperscalers. Thanks to Nvidia’s guarantee, cloud providers can also offer shorter-term contracts to AI startups and other customers who need capacity for only one to three years. This makes the market more accessible to players that lack the financial clout of AWS, Microsoft, or Google.
The researchers view this development as a fundamental shift. Nvidia is no longer limiting itself to supplying chips and AI software but is also taking on a role traditionally reserved for banks and other financiers. According to SemiAnalysis, this shifts the competitive landscape in AI from production capacity to the question of who can finance the massive investments in next-generation AI infrastructure.