How are Apollo and Blackstone betting on the next AI infrastructure supercycle? The computing power battle behind the $35 billion wager

On June 9, 2026, Broadcom, Apollo Global Management, and Blackstone jointly announced the establishment of the AI XPV Platform. This is a strategic attempt to deeply integrate the world's top chip design capabilities with the largest-scale private equity capital—aiming to deploy over 20GW of AI computing capacity through Broadcom's XPU chips and networking solutions before 2028.

The platform's initial transaction size reached $35 billion, led by Apollo, with Blackstone's credit and insurance business participating as the initial anchor investor. This funding will first support the expansion of Anthropic's previously announced over 1GW computing infrastructure, expected to begin deployment at Fluidstack data center sites in mid-2026. Anthropic and OpenAI are listed as the platform's initial clients.

$35 Billion in Computing Power Expansion: From 1GW to 20GW Roadmap

The first $35 billion transaction is the first pillar of the entire AI XPV Platform. Apollo stated that this financing adopts a multi-year phased capital commitment structure, designed to provide certain and executable funding for Anthropic's model training and inference.

In terms of scale, the initial 1GW computing expansion itself is quite substantial—by way of comparison, 1GW of power can supply electricity to approximately 750k households. The platform's long-term goal of 20GW means that by 2028, the total supported computing capacity will be 20 times the initial scale.

At the hardware level, Broadcom will provide customized XPU chips and networking solutions. Notably, Broadcom also bears residual value guarantees for the chips—covering related losses if Anthropic defaults. This arrangement somewhat reduces downside risk for capital providers and is one of the key structural designs enabling such large-scale private placements.

Broadcom CEO Hock Tan described this moment as a “historic inflection point where AI computing demand is fundamentally reshaping the global economic landscape.” Apollo President Jim Zelter emphasized that the platform's establishment reflects Apollo's confidence in Broadcom's technological leadership and Anthropic's cutting-edge roadmap. Blackstone President Jon Gray added that the demand for computing power has created an unprecedented opportunity for large-scale investment in AI infrastructure.

Why Anthropic? An AI Company Valued Near a Trillion Dollars

Anthropic's role as the recipient of the platform's first funding was no accident. On May 28, 2026, Anthropic announced the completion of a $65 billion Series H financing round, with a post-investment valuation of $965 billion, surpassing OpenAI's $852 billion valuation, making it the world's highest-valued AI large model startup. Based on the S&P 500 enterprise market cap ranking, Anthropic's valuation ranks as the 13th largest company in the U.S.

This valuation is supported by solid revenue growth. As of April 2026, Anthropic's annualized operating revenue exceeded $47 billion, more than 50% growth from $30 billion at the end of February during the Series G funding, with monthly new revenue exceeding $5 billion. Its revenue structure mainly relies on enterprise API calls, covering finance, legal, technology, and other fields.

In June 2026, Anthropic submitted a confidential S-1 registration statement to the U.S. SEC. Against this backdrop, using SPV (Special Purpose Vehicle) structures to place computing hardware off the balance sheet is an important financial arrangement for Anthropic, which is about to go public.

Anthropic Annual Revenue Growth Curve (2025-2026)

Why Is Private Capital Making a Major Bet on AI Infrastructure?

The participation of Apollo and Blackstone reflects a deeper structural change: AI infrastructure financing is shifting massively from traditional capital markets to private equity.

As of May 2026, there are 695 infrastructure funds worldwide in fundraising stages, with total target capital reaching $555 billion. Goldman Sachs estimates that over the next five years, total capital inflows into AI infrastructure could reach $4 trillion to $8 trillion. In 2026 alone, the AI infrastructure expenditure of the four major tech giants is expected to exceed $670 billion—this investment proportion even surpasses the U.S. railroad expansion period of the 1850s.

The core logic driving this trend is supply-demand imbalance. Capital expenditure by large cloud providers in 2026 has already risen to about $650 billion, potentially exceeding $1.1 trillion in 2027. However, relying solely on their operational cash flows, these companies cannot cover such extensive multi-year construction needs, making external financing the norm.

Apollo describes AI compute as “becoming one of the most attractive new asset classes in finance,” characterized by contracted cash flows, critical infrastructure attributes, and an increasingly severe supply-demand mismatch. Apollo Partner Jamshid Ehsani pointed out that this transaction is “the largest private equity financing ever executed.” From a macro perspective, private equity firms are becoming a key source of funding for AI companies—these firms are under pressure from high costs and limited supply of AI infrastructure.

Key Data on Private Capital Accelerating into AI Infrastructure

| Indicator | Data | Source | | --- | --- | --- | | First transaction of AI XPV Platform | $35 billion | Apollo/Broadcom announcement | | Platform long-term target | 20GW AI capacity (by 2028) | Broadcom announcement | | Anthropic Series H financing | $65 billion | Anthropic announcement | | Anthropic post-investment valuation | $965 billion | Anthropic announcement | | Global infrastructure fund fundraising | 695 funds / $555 billion (as of May 2026) | Goldman Sachs research | | AI capital expenditure by four tech giants | $5.3 trillion (2025-2030) | Goldman Sachs research |

Impact on Related Assets: Broadcom, NVIDIA, and the Broader AI Supply Chain

The platform's establishment has directly or indirectly affected the industry chain upstream and downstream.

As the core chip supplier of the platform, Broadcom (AVGO) is the biggest direct beneficiary. After the announcement, Blackstone's stock rose 5.34% in a day, Apollo increased by 1.49%. As of June 17, 2026, Broadcom's stock closed at approximately $396.94. JPMorgan raised Broadcom's target price from $500 to $580 on June 18, reaffirming an “overweight” rating. Broadcom's fiscal Q2 2026 revenue reached $22.19 billion, up 48% year-over-year.

For NVIDIA (NVDA), the impact is more complex. Broadcom's customized chip solutions offer an alternative path for tech companies seeking to reduce dependence on NVIDIA. However, the explosive growth in AI compute demand also benefits NVIDIA's GPU business in the long term. As of the close on June 17, 2026, NVIDIA's stock was $204.65, with a market cap of about $4.96 trillion. On a broader scale, high-bandwidth memory suppliers like Micron Technology will also benefit from the ongoing expansion of AI infrastructure.

AI Factory and Financing Innovation

The establishment of the AI XPV Platform marks the transition of the “AI Factory” concept from theory to practice. The AI Factory refers to an integrated production system combining large-scale compute, customized chips, long-term capital, and cutting-edge model capabilities. Broadcom's XPU and networking solutions, Apollo and Blackstone's long-term capital, and the model demands from Anthropic and OpenAI form a closed loop on this platform.

The core innovation of this model lies in its financing structure. Apollo and Blackstone are not simply providing a loan but have established a scalable capital framework—the initial $35 billion transaction is just the beginning, and the platform is designed to continuously provide follow-up financing as demand scales. Apollo stated they look forward to “expanding this platform together with Broadcom, Blackstone, and broader partners as AI infrastructure accelerates.”

This “chip manufacturer + private capital + model company” tripartite collaboration is becoming a new paradigm for AI infrastructure financing. Won Kim, Head of Corporate Development and AI Infrastructure at Broadcom, noted, “The growth rate of AI compute demand has already exceeded the capacity of traditional capital markets.”

Breakdown of the $35 Billion AI XPV Platform Financing Structure

| Level | Scale | Yield/Features | Risk Characteristics | | --- | --- | --- | --- | | Senior A1 Notes | $600 million | U.S. Treasury + 100bps | Broadcom gap fill guarantee, investment grade | | Senior A2 Notes | $24 billion | ~5.75% | Broadcom gap fill guarantee, investment grade | | Subordinated (Junior) Notes | $4.5 billion | ~8.5% | No Broadcom guarantee, issued at 98-99 cents discount | | Equity Layer | $800 million | SPV actual holder | Highest risk, highest potential return | | Total | ~$35 billion | — | — |

Capturing AI Infrastructure Investment Opportunities via Gate Trading Platform

For investors seeking to participate in AI infrastructure opportunities, the Gate platform offers a convenient access channel.

On June 1, 2026, Gate officially launched real stock trading services, allowing users to directly trade US stocks and ETFs using USDT within the platform. By mid-June, Gate had listed over 10,000 real stocks and ETFs, covering NYSE, Nasdaq, NYSE Arca, NYSE American, and BATS. On June 11, Gate further launched Hong Kong stock trading, initially covering over 1,500 Hong Kong-listed stocks.

Through the Gate platform, users can directly trade assets highly related to AI infrastructure themes, including core chip companies like Broadcom (AVGO), NVIDIA (NVDA), and broader tech sector ETFs. The Gate stock trading is spot trading, with no funding rates or overnight holding fees. Android users can update the Gate app to the latest version, and iOS users to version 8.21.5 to access these features.

Conclusion

The joint participation of Apollo and Blackstone in Broadcom's AI XPV Platform marks a milestone in the large-scale systemic entry of private capital into AI infrastructure. The $35 billion initial transaction, the 20GW long-term capacity target, and the “chip + capital + model” tripartite collaboration outline a new paradigm for AI infrastructure financing.

The core driver of this model is not short-term speculative sentiment but the structural gap between AI compute demand and capital supply that is widening. When the capital needed for ultra-large-scale compute construction exceeds traditional financing capacity, private capital's involvement becomes inevitable—and not accidental. For investors, understanding this structural shift is key to making rational allocations in the AI infrastructure supercycle. Through emerging trading platforms like Gate, investors are gaining unprecedented convenient pathways to turn this macro trend into concrete asset allocation actions.