Broadcom, Apollo and Blackstone launch AI infrastructure financing platform starting with Anthropic compute

Broadcom is teaming with Apollo Global Management and Blackstone on a new financing platform aimed at underwriting the buildout of AI infrastructure, with the effort set to begin by supporting Anthropic’s next round of compute capacity. The partners described the initiative as a way to mobilize large amounts of institutional capital to match the scale and timing requirements of AI chip deployments and data-center infrastructure.

Apollo said it would lead an initial $35 billion capital solution as part of Broadcom’s “AI XPV Platform,” in partnership with Blackstone and “leading global banks.” The Apollo announcement ties the first phase to Anthropic’s previously stated compute expansion of more than 1 gigawatt, framed as part of a broader effort intended to enable over 20 gigawatts of compute capacity for frontier AI labs through 2028.

While the announcement centers on the financing structure, it also highlights the industrial and technology stack the platform is meant to enable. Apollo described the model as pairing “some of the world’s most advanced silicon and networking solutions” with long-term, flexible capital, with the objective of accelerating compute deployment across what it called the frontier AI ecosystem. Broadcom’s role in the platform is broadly characterized as supplying infrastructure technology that can be scaled, rather than as a pure financial intermediary.

Both Apollo and Blackstone have positioned the arrangement as a new approach to mobilizing private capital for digital infrastructure. In Apollo’s description, the initial transaction is designed to deliver “committed, certain capital” across a multi-year draw schedule, meaning cash funding is not solely dependent on short-term pricing or near-term deployment milestones. The company did not provide granular terms in the announcement beyond that high-level structure, including how draw triggers would operate in practice.

The market impact of deals like these is that major AI workloads are increasingly constrained by access to compute hardware, power and the network of data-center systems needed to run training and inference at scale. Financing structures can shift the bottleneck from demand for capital to the availability of chips, networking equipment, power capacity and construction lead times, all of which tend to move on different timelines than software roadmaps.

Sector context also matters for Broadcom because the company has been moving beyond selling components to supplying a more complete set of AI infrastructure offerings, including accelerators and related networking and systems. Initiatives that connect infrastructure procurement with long-horizon financing are consistent with that direction, especially as AI infrastructure spend becomes less like a one-off capex decision and more like an ongoing capacity planning exercise.

A key caveat is that the disclosed details remain high level. Apollo’s announcement does not spell out the exact legal or economic structure of the AI XPV Platform beyond the existence of the $35 billion “capital solution,” nor does it identify all participating banks or the specific financing instruments. It also does not clarify whether additional AI labs besides Anthropic are locked into capacity commitments at launch, or how Broadcom’s technology roadmap and supply chain assumptions are reflected in the commitments.

What to watch next is how the platform translates into signed procurement and deployment timelines. Investors and customers will likely look for disclosures on the compute capacity schedule toward the 20 gigawatt target, evidence of Anthropic buildout progress, and whether Broadcom further specifies how its hardware and systems offerings map to the financed capacity. Future updates from Broadcom, Apollo, and Blackstone would also be expected if the platform expands to other frontier AI labs or if the initial $35 billion tranche is followed by additional funding rounds.