Asset ownership
Real hardware, on the balance sheet. Investors own the underlying GPUs, not equity in an operating company.
Stable cash flows from long-term GPU leases, secured by owned infrastructure. A new asset class built on the same financial architecture that has financed aircraft, telecom, and shipping for decades.
The AI build-out is the largest infrastructure deployment of our generation. By 2030, annual global spend on AI compute infrastructure is projected to exceed $7 trillion, driven by training frontier models, serving inference at scale, and the buildout of sovereign AI capacity.
GPUs are the core asset of this build-out. They’re physical, contractable, and depreciable — making them a natural fit for structured asset financing rather than speculative equity investment.
Zuba gives institutional capital direct exposure to this build-out. We own the hardware, lease it to operators under long-term contracts, and pass through the cash flows to our investors. The structure is designed for predictable, contracted returns — not venture-style upside.
We are not betting on the equity of any AI company. We are financing the silicon that the AI economy runs on. That distinction — asset ownership rather than equity speculation — is what makes this an institutional-grade opportunity.
Real hardware, on the balance sheet. Investors own the underlying GPUs, not equity in an operating company.
Long-term lease payments create predictable, recurring revenue streams backed by operator commitments.
Redeployment and resale upside when leases end. Hardware retains economic life beyond the initial contract.
Senior to equity in default scenarios. Asset-backed structure provides stronger protections than unsecured positions.
A typical Zuba investment in GPUs is structured as a senior-secured debt facility with mezzanine and equity tranches, optimized for risk-adjusted returns.
Indicative target returns · Actual terms per fund
Lease payments are distributed down the capital stack, with senior tranches paid first and equity capturing residual upside.
Projected global spend on AI compute infrastructure by 2030 — a sustained multi-year build-out, not a cycle.
Inference workloads shift from one-time training events to ongoing operating cost — durable, contracted revenue.
Nations and regions building domestic AI capacity — long-term demand floor for compute infrastructure.
Every investment has risk. Here’s how they’re allocated in our structure.
“This profile is attractive to private credit investors seeking predictable cash flows. It is equivalent to financing long-lived, asset-backed investments like aircraft, railcars, power plants, and telecom towers.”
— Zuba Investor MaterialsRequest our investor deck or schedule a confidential conversation with our capital markets team.