Key Takeaways
- • Broadcom, Apollo Global Management, and Blackstone's Credit & Insurance Business jointly launched the AI XPV Platform designed to enable over 20 gigawatts of global AI compute capacity through 2028, targeting frontier AI labs including Anthropic and OpenAI.
- • The platform launches with an initial $35 billion transaction led by Apollo in partnership with Blackstone to support Anthropic's capacity expansion of more than 1 gigawatt of compute infrastructure, with deployment expected to begin mid-2026 on Fluidstack-based sites.
- • The strategic initiative leverages Broadcom's XPUs (custom processors) and networking solutions, with the explicit goal of reducing per-token delivery costs and enabling frontier model training and inference at lower power consumption and capital efficiency.
- • This partnership establishes a scalable financing and deployment framework for future XPU-based compute infrastructure, positioning the three firms as primary architects of critical AI supply chain infrastructure alongside leading AI research organizations.
Summary
Broadcom, Apollo Global Management, and Blackstone announced on June 9, 2026, the establishment of the AI XPV Platform, a landmark strategic initiative designed to accelerate the deployment of over 20 gigawatts of global AI compute capacity through 2028. The platform represents a significant alignment between semiconductor infrastructure, institutional capital, and frontier AI development, addressing the acute demand for compute resources among leading artificial intelligence research organizations.
The partnership launches with an immediate $35 billion initial transaction led by Apollo in partnership with Blackstone's Credit & Insurance Business. This inaugural tranche is allocated to support Anthropic's previously announced capacity expansion of more than 1 gigawatt of compute infrastructure, with deployment expected to commence mid-2026 through Fluidstack-based infrastructure sites. According to Broadcom, this deployment builds upon the existing strategic relationship between Broadcom and Anthropic and demonstrates both the scale and near-term execution capability of the platform.
The AI XPV Platform is architected around Broadcom's XPU technology and customized networking solutions, specifically optimized for the computational demands of frontier model training and inference. The framework is explicitly designed to reduce per-token delivery costs and minimize power consumption, addressing two critical cost vectors for large-scale AI deployment. By establishing a scalable financing and deployment structure, the platform creates a repeatable model for future capacity expansions beyond the initial Anthropic deployment, with named targets including OpenAI among frontier AI labs.
Hock Tan, President and CEO of Broadcom, characterized the initiative as positioned at "a historic inflection point where the demand for AI compute is fundamentally reshaping the global economic landscape," underscoring the strategic significance of the infrastructure buildout. The partnership demonstrates how semiconductor manufacturers, alternative asset managers with significant capital deployment capacity, and credit-focused investment platforms can coordinate to address the capital intensity and operational complexity of global AI infrastructure deployment.
The partnership launches with an immediate $35 billion initial transaction led by Apollo in partnership with Blackstone's Credit & Insurance Business. This inaugural tranche is allocated to support Anthropic's previously announced capacity expansion of more than 1 gigawatt of compute infrastructure, with deployment expected to commence mid-2026 through Fluidstack-based infrastructure sites. According to Broadcom, this deployment builds upon the existing strategic relationship between Broadcom and Anthropic and demonstrates both the scale and near-term execution capability of the platform.
The AI XPV Platform is architected around Broadcom's XPU technology and customized networking solutions, specifically optimized for the computational demands of frontier model training and inference. The framework is explicitly designed to reduce per-token delivery costs and minimize power consumption, addressing two critical cost vectors for large-scale AI deployment. By establishing a scalable financing and deployment structure, the platform creates a repeatable model for future capacity expansions beyond the initial Anthropic deployment, with named targets including OpenAI among frontier AI labs.
Hock Tan, President and CEO of Broadcom, characterized the initiative as positioned at "a historic inflection point where the demand for AI compute is fundamentally reshaping the global economic landscape," underscoring the strategic significance of the infrastructure buildout. The partnership demonstrates how semiconductor manufacturers, alternative asset managers with significant capital deployment capacity, and credit-focused investment platforms can coordinate to address the capital intensity and operational complexity of global AI infrastructure deployment.
Go deeper than the headline
Research every firm in this story, track the ones that matter and get the daily briefing — free for 7 days.