Nvidia bank or the AI future on credit

SP500

Key zone: 7,680 - 7,750

Buy: 7,800 (on a decisive break above 7,800); target 7,950; StopLoss 7,720

Sell: 7,650 (on strong negative fundamentals); target 7,500-7,350; StopLoss 7,720

Nvidia continues to post exceptionally strong financial results, and the AI boom still shows no signs of slowing down. But as the business scales, the company’s role is changing as well. From a manufacturer of accelerators, Nvidia is gradually turning into one of the key financiers of AI infrastructure.

The more aggressively AI infrastructure expands, the more Nvidia puts its own capital at risk.

Previously, the main question was simple: how many GPUs are customers willing to buy? Now a far more dangerous issue is emerging: can these customers earn enough to service the infrastructure built on Nvidia’s equipment?

Nvidia Is Selling More Than GPUs — It Is Selling the Entire System a Chance to Grow

A reminder:

Nvidia Is Becoming a Bank for Its Own Ecosystem

The latest report shows that Nvidia’s revenue reached a record $96.2 billion, up 18% quarter over quarter and 106% year over year. Data Center generated $89 billion (+117%), net income totaled $59.7 billion, and GAAP operating income reached $63.7 billion. Gross margin remained at 75%. The company expects $108 billion in revenue next quarter and forecasts sales growth of approximately 70% in the next fiscal year.

The numbers are impressive. But they are also creating a new problem: for Nvidia to continue growing at this pace, its customers will have to build and finance a colossal amount of new infrastructure.

The company is effectively becoming the bank of its own ecosystem.

• Customer financing is evolving from a collection of individual deals into a standalone element of Nvidia’s strategy.
• The company provides guarantees, supports data center financing, makes commitments related to compute capacity utilization, invests in neocloud companies, and works with financial groups to create mechanisms for raising enormous amounts of capital.
• Nvidia announced the creation, together with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR, of financial platforms capable of mobilizing more than $500 billion in private capital to build AI infrastructure.

Nvidia is now interested not only in whether a customer wants to buy GPUs. It also needs a bank, infrastructure fund, or private equity investor to agree to finance that purchase.

The biggest bet has already been made.

• On August 17, Nvidia announced its participation in the PORTS-Pike project in Ohio together with SB Energy and OpenAI.
• Nvidia is providing credit support for land, energy infrastructure, and construction of the initial 4.25 GW, has the option to expand the project by another 3.75 GW, and is investing $1.5 billion in SB Energy.
• The first capacity is expected to come online starting in 2028. Nvidia’s potential support for the project could reach approximately $105 billion.

The economic logic of this model looks flawless:

more demand → more data centers → more financing → more GPUs → more cloud capacity → cheaper AI compute → more AI applications → even more demand.

For lenders, Nvidia’s support reduces risk. OpenAI gets infrastructure, SB Energy gets capital. And Nvidia gains a potential market for enormous quantities of its own accelerators, processors, and networking equipment.

So, what does this mean?

AI is becoming too expensive even for Big Tech. Nvidia’s investments are already beginning to have a significant impact on profits, but at the same time, operational risk is increasing. The next stage of the AI revolution requires such massive investment that Microsoft, Amazon, Meta, or Alphabet alone are no longer enough.

Nvidia plans to bring pension funds, insurance companies, infrastructure funds, private equity, and debt investors into this system. And the more third-party money flows into AI infrastructure, the more closely Nvidia becomes tied to its financial performance.

The main risk is a domino effect. If returns on AI projects fall short of expectations, problems will begin spreading throughout the entire chain.

That is why investors can no longer afford to monitor only Nvidia’s revenue, Blackwell, and Rubin. The debt burden of neocloud companies, GPU utilization, refinancing costs, the residual value of previous generations of accelerators, and the volume of guarantees Nvidia is willing to take onto its own balance sheet are becoming far more important.

It remains unclear what internal limits the company has established for such deals and whether it is creating separate cash reserves for potential guarantee payments.

If the economics of AI projects begin to break down, investors will have to evaluate Nvidia not only as a hardware manufacturer, but also as a lender to its own ecosystem.

Let’s see how this plays out.

So we act wisely and avoid unnecessary risks.

So we act wisely and avoid unnecessary risks.