Apollo, Blackstone Eye $35B Broadcom AI Chip Deal

Apollo Global Management and Blackstone are reportedly in discussions to participate in a massive $35 billion financing package for Broadcom, one of the largest private credit deals in recent memory. The funding would support Broadcom's accelerating push into custom artificial intelligence chips, a market segment that has become one of the hottest battlegrounds in the semiconductor industry.

Negotiations remain ongoing and terms could still shift, according to people familiar with the matter. The sheer scale of the deal underscores how deeply intertwined the worlds of private credit and AI infrastructure have become in 2025.

Key Takeaways at a Glance

• Apollo and Blackstone are among private credit firms in talks with Broadcom over roughly $35 billion in financing
• The capital would fund Broadcom's development of custom AI task-specific chips (ASICs)
• Deal terms are still being negotiated and could change
• The financing would rank among the largest private credit transactions ever arranged
• Broadcom's AI chip business has grown explosively, driven by hyperscaler demand
• The move signals private credit's deepening role in funding AI infrastructure buildouts

Why Broadcom Needs $35 Billion for AI Chips

Broadcom has rapidly emerged as one of the most important players in the AI chip ecosystem, but through a different path than Nvidia. While Nvidia dominates the market for general-purpose AI accelerators (GPUs), Broadcom designs custom application-specific integrated circuits (ASICs) tailored to the unique workloads of major cloud and AI companies.

The company's largest customers — widely reported to include Google, Meta, and ByteDance — commission these bespoke chips to optimize performance and reduce costs for their specific AI training and inference tasks. This custom silicon approach has proven enormously appealing to hyperscalers looking to reduce their dependence on Nvidia's expensive GPU ecosystem.

Broadcom's AI-related revenue has surged as a result. In its most recent fiscal year, the company reported that its AI semiconductor revenue exceeded $12 billion, a figure that CEO Hok Tan has projected could grow substantially in the years ahead. A $35 billion financing package would give the company significant firepower to expand fabrication capacity, invest in next-generation chip architectures, and deepen its engineering partnerships with key clients.

Apollo and Blackstone Deepen Their AI Bets

The involvement of Apollo Global Management and Blackstone highlights a broader trend: the world's largest alternative asset managers are aggressively positioning themselves as financiers of the AI revolution. Both firms have built massive private credit operations in recent years, and AI infrastructure has become a primary deployment target.

Apollo, which manages over $700 billion in assets, has been particularly vocal about the opportunity in AI-related lending. The firm has funded data center construction, semiconductor supply chain expansions, and energy infrastructure projects tied to AI compute demand. A deal of this magnitude with Broadcom would represent one of Apollo's most significant AI-adjacent commitments to date.

Blackstone, the world's largest alternative asset manager with over $1 trillion under management, has similarly leaned heavily into AI infrastructure. The firm has committed tens of billions to data center development through partnerships and direct investments. Participating in Broadcom's financing would extend Blackstone's AI exposure deeper into the semiconductor layer of the technology stack.

• Apollo: Over $700 billion in AUM, active in AI infrastructure lending
• Blackstone: Over $1 trillion in AUM, major data center investor
• Private credit market: Has grown to over $1.7 trillion globally
• AI infrastructure spending: Expected to exceed $300 billion annually by 2026

Private Credit Reshapes How AI Gets Funded

Traditionally, a company like Broadcom would turn to the public bond market or syndicated bank loans for a financing of this scale. The fact that private credit lenders are being tapped for a roughly $35 billion package reflects a fundamental shift in how large-scale technology investments get funded.

Private credit offers several advantages for borrowers like Broadcom. Deals can be structured more flexibly, executed more quickly, and kept more confidential than public market alternatives. For lenders like Apollo and Blackstone, these transactions offer attractive yields — often in the range of 8% to 12% — on credits backed by companies with strong cash flows and strategic importance.

The AI boom has supercharged this dynamic. Companies across the AI value chain — from chip designers to data center operators to cloud providers — need enormous amounts of capital deployed on compressed timelines. Private credit funds, which have raised record amounts of dry powder in recent years, are uniquely positioned to meet this demand.

Compared to Nvidia's approach of self-funding through its enormous profit margins (the company generated over $60 billion in net income in fiscal 2025), Broadcom's decision to seek external financing reflects the different economics of the custom chip business. ASIC design requires significant upfront investment in engineering and fabrication partnerships before revenue materializes, making external capital a logical choice.

Broadcom vs. Nvidia: The Custom Chip Battleground

The financing underscores the intensifying competition between Broadcom's custom ASIC model and Nvidia's GPU-centric approach. Both companies are benefiting from the AI spending boom, but they serve different needs.

Nvidia's H100, H200, and Blackwell GPU families offer versatile, off-the-shelf AI acceleration. They are the default choice for most AI workloads and dominate the market with an estimated 80%+ share of AI training chips. However, their high cost and limited availability have pushed the largest technology companies to explore alternatives.

Broadcom fills that gap with custom-designed chips that can deliver superior performance-per-watt for specific workloads at lower per-unit costs — once the upfront design investment is amortized across large production runs. Google's TPU (Tensor Processing Unit), which Broadcom helps design and manufacture, is the most prominent example of this approach.

The $35 billion financing would allow Broadcom to:

• Scale its custom chip design teams and support more simultaneous customer programs
• Invest in advanced packaging and fabrication technologies at partners like TSMC
• Develop next-generation interconnect and networking silicon optimized for AI clusters
• Build out testing and validation infrastructure for increasingly complex chip designs
• Potentially pursue strategic acquisitions in adjacent technology areas

What This Means for the AI Industry

For the broader AI ecosystem, a deal of this magnitude sends several important signals. First, it confirms that the AI infrastructure buildout is far from peaking. If private credit firms are willing to commit $35 billion to a single semiconductor company, they clearly see years of sustained demand ahead.

Second, it validates the custom chip model as a durable alternative to the Nvidia GPU monopoly. Broadcom would not seek — and lenders would not provide — this level of financing if the custom ASIC opportunity were not substantial and growing.

Third, the deal highlights the growing importance of financial engineering in the AI race. The companies that can access the most capital on the best terms will have a significant advantage in building out next-generation AI infrastructure. Broadcom's ability to attract financing from the world's most sophisticated private credit firms is itself a competitive moat.

For developers and businesses building on AI, the implications are positive. More investment in custom chip development means more competition, which should eventually translate into lower compute costs and more diverse hardware options for running AI workloads.

Looking Ahead: Timeline and Next Steps

The deal remains in negotiation, and several key variables could influence its final structure. The total financing amount could shift from the roughly $35 billion figure currently reported. Additional private credit firms beyond Apollo and Blackstone may join the lending syndicate. And the specific terms — interest rates, covenants, maturity dates — are still being worked out.

If finalized, the transaction would likely close in the second half of 2025, giving Broadcom access to capital that could fuel its AI chip roadmap through the end of the decade. The deal would also set a new benchmark for private credit's role in technology financing, potentially opening the door for similar mega-transactions involving other semiconductor and AI infrastructure companies.

Industry observers will be watching closely for any formal announcement. In the meantime, the mere fact that discussions of this scale are taking place confirms a simple truth: in 2025, the AI chip race is not just a technology competition — it is a capital competition. And Broadcom, backed by some of the world's deepest pockets, intends to compete at the highest level.