Detailed Analysis
Apollo Global Management and Blackstone have jointly arranged a $35 billion chip financing package for Anthropic, the AI safety company and maker of the Claude family of large language models, in what represents one of the largest infrastructure financing deals in the history of artificial intelligence. The arrangement, reported by the Financial Times, involves two of the world's most prominent alternative asset managers structuring what is effectively a debt or structured financing facility designed to fund the acquisition of semiconductor hardware — most likely Nvidia GPUs and related accelerator chips — that power the training and inference workloads underpinning Anthropic's AI systems.
The scale of the deal underscores the extraordinary capital intensity that frontier AI development has come to demand. Training and running state-of-the-art large language models requires vast clusters of specialized chips, data center infrastructure, and power supply commitments that can stretch into the tens of billions of dollars over multi-year horizons. Anthropic, having already secured substantial equity investment from Amazon and Google, appears to be diversifying its capital structure by turning to private credit and alternative asset markets to finance hardware specifically, rather than diluting equity holders further. This approach — separating infrastructure financing from operational equity — has become an increasingly common structure in the AI industry.
The involvement of Apollo and Blackstone signals a maturation of the AI investment ecosystem, in which traditional private equity and credit giants are now competing aggressively with sovereign wealth funds, strategic corporate investors, and venture capital for exposure to AI infrastructure. Both firms have been building out their private credit and infrastructure lending capabilities, and a deal of this magnitude with Anthropic represents a marquee transaction that validates the creditworthiness and strategic importance of frontier AI labs as borrowers. It also reflects the broader financialization of AI compute, where chips and data center capacity are increasingly treated as financeable hard assets rather than purely operational expenses.
This transaction connects to a sweeping trend in which the compute layer of AI — the physical chips, the power contracts, and the cooling infrastructure — has become a distinct asset class attracting specialized financing structures. Competitors such as OpenAI, xAI, and Microsoft-backed ventures have similarly pursued large-scale infrastructure financing arrangements, often backed by sovereign wealth or large institutional capital. The $35 billion figure for Anthropic alone suggests that the total capital mobilizing around AI chip acquisition globally is measured in the hundreds of billions, placing chip financing at the center of geopolitical, financial, and technological competition over AI leadership in the mid-2020s.
Read original article →