Hands holding a briefcase of money by Efired via Shutterstock
Artificial intelligence is becoming as much a financing challenge as a technology challenge. The biggest AI developers are racing to deploy gigawatts of computing capacity, but the chips, networking equipment, and data centers required to support that expansion cost tens of billions of dollars. That is creating an opening for companies that can do more than manufacture the hardware.
Broadcom (AVGO) appears to be exploiting that opening. It is reportedly negotiating a financing package that could exceed $60 billion in senior debt and approach $100 billion when a potential $30 billion junior tranche is included. Apollo Global Management (APO) and Blackstone (BX) are considering participation.
If completed, this could become one of Broadcom’s most important competitive advantages over Nvidia (NVDA)—not because Broadcom has found a better GPU, but because it is helping customers figure out how to pay for the hardware.
This builds directly on the $35 billion AI XPV Platform Broadcom, Apollo, and Blackstone launched in June. That transaction was designed to finance more than 1 gigawatt of Anthropic’s compute capacity, with the broader platform targeting more than 20 GW through 2028. Broadcom said the platform combines its XPUs and networking technology with institutional capital to accelerate deployments for frontier AI companies.
The proposed $100 billion package would therefore represent a major expansion rather than a wholly new strategy.
The Real Prize Is Revenue Conversion
Broadcom already has the AI demand. The problem is converting that demand into shipments and revenue quickly enough.
Its fiscal second-quarter AI semiconductor revenue reached $10.8 billion, up 143% year-over-year (YoY). Management expects $56 billion of AI semiconductor revenue in fiscal 2026 and reiterated its forecast for more than $100 billion in fiscal 2027. Those numbers require customers to keep spending, and financing can remove a bottleneck.
Anthropic, for example, can lease compute without borrowing the entire purchase price onto its own balance sheet. The special purpose vehicle (SPV) owns the equipment, while Broadcom gets another path to sell XPUs, networking equipment, and related infrastructure.
That creates a powerful feedback loop: more financing capacity leads to more deployed compute, meaning more Broadcom hardware sales and greater revenue visibility.
It also helps Broadcom compete where Nvidia is less dominant. Google (GOOG) (GOOGL) TPUs, developed with Broadcom, represent a custom-accelerator alternative to Nvidia’s GPUs. Broadcom is supplying the silicon and networking while simultaneously helping create the financial infrastructure that gets those systems into AI data centers.
That is a different competitive weapon.
The Risk Is Moving Onto AVGO’s Contingent Liability Ledger
Granted, “off-balance sheet” does not mean “risk-free.” Broadcom isn’t taking on $100 billion of conventional corporate debt if the proposed structure proceeds. But guarantees can become liabilities when things go wrong.
Bank of America has modeled how Broadcom’s residual-value guarantees could grow if the AI XPV platform expands toward its 20 GW target. One scenario put maximum residual-value guarantee exposure at roughly $370 billion by mid-2029. However, estimated maximum loss exposure was far lower at about $42 billion under a 100% default assumption and $10.5 billion under a 25% default scenario. The divergent forecasts explain why investors should pay attention to the fine print.
If AI demand remains strong and chip values hold up, the structure could accelerate Broadcom’s growth while requiring relatively little balance-sheet capital. If demand weakens, leases default, or AI hardware depreciates faster than expected, Broadcom could face meaningful payouts. That is the trade-off.
Key Takeaway
In short, this is a net positive for Broadcom shareholders if the proposed financing is structured prudently. Broadcom is not merely selling chips. It is helping finance the infrastructure required to deploy them, giving customers a way around one of AI’s biggest constraints: access to capital.
The opportunity is enormous. Broadcom’s AI semiconductor business is expected to grow from $56 billion in fiscal 2026 to more than $100 billion in fiscal 2027, while its AI XPV platform targets more than 20 GW of compute.
That makes the potential $100 billion financing package more than a financial engineering exercise. It could become a demand-generation machine for Broadcom’s custom silicon and networking business.
Investors should watch three things: the final guarantee size, actual SPV utilization, and credit-rating commentary. If Broadcom can scale the platform without taking disproportionate credit risk, its financing expertise could prove nearly as valuable as its chip-design expertise.
The talks are still ongoing, so the final structure matters. But the strategic thesis is already clear: Nvidia may own the GPU ecosystem, while Broadcom is trying to own the financing ecosystem around custom AI compute
On the date of publication, Rich Duprey did not have (either directly or indirectly) positions in any of the securities mentioned in this article. All information and data in this article is solely for informational purposes. For more information please view the Barchart Disclosure Policy here.