Nvidia takes on a new role: financier, not just supplier

Nvidia is taking a remarkable step beyond its traditional role as a semiconductor manufacturer. According to Nasdaq Markets, the company is in the process of providing financing guarantees of up to $105 billion for a large-scale OpenAI data center in Ohio — and expects to generate around $200 billion in chip revenue from the first construction phase alone.

This is not merely a business deal — it is a strategic move that ties Nvidia's own success even more closely to OpenAI's growth and the global AI race.

Nvidia expects to sell chips worth $200 billion in the first phase of a single project alone.
Nvidia backs OpenAI giant with $105 billion - Bilde 1

What it means for investors

For investors, the transaction sends several signals at once. On one hand, it confirms that demand for Nvidia's AI chips is so strong that the company is now willing to finance the infrastructure that buys them. On the other hand, it exposes Nvidia to credit and project risk in a way the company has historically avoided.

Nvidia's data center revenue for fiscal year 2026 already grew 92 percent to $75.2 billion, according to the company's own reports. A successful Ohio project could potentially double or triple that figure on a single contract.

$105B
Financing guarantee for the Ohio project
$200B
Expected chip deliveries, phase 1
Nvidia backs OpenAI giant with $105 billion - Bilde 2

Nvidia dominates — but competition is intensifying

Nvidia's GPU market share for hyperscale data centers exceeded 80 percent in 2024, and the company controlled around 65 percent of the overall AI chip market for data centers in 2023, according to industry research cited by multiple analysts. The CUDA software platform, which has been on the market for over a decade, is considered an industry standard and serves as a significant barrier to entry for competitors.

Nevertheless, the market landscape is shifting:

  • AMD is promoting its Instinct MI300X chip, which the company claims can be up to five times faster than Nvidia's H100 in certain operations and offers twice the memory (192 GB vs. 80 GB HBM3). AMD has, however, acknowledged that its ROCm software ecosystem is still less mature than CUDA.
  • Intel is positioning Gaudi3 as a cost-effective alternative with a list price around half that of the H100, according to available product information.
  • Google, AWS, and Microsoft are all developing their own AI chips (ASICs) to reduce dependence on external suppliers. Morgan Stanley analysts have estimated that the ASIC market could double in 2025, and Omdia forecasts that AI ASIC revenue could reach $84.5 billion by 2030.

Hyperscalers are building their own chips — Nvidia's dominance is real, but not unchallenged

The risk picture investors should keep in mind

The direct exposure from the guarantee deal in question is fundamentally about concentration risk: Nvidia is committing a significant portion of its financial capacity to a single customer and a single megaproject. If OpenAI's growth rate were to slow, the financing model were to be reconsidered, or regulatory obstacles were to emerge, the consequences could hit Nvidia directly.

More distant, but not irrelevant, is the geopolitical risk surrounding semiconductor manufacturing and export controls — a topic that has already affected Nvidia's ability to deliver to certain markets.

What happens next?

The Ohio project represents a new form of capital allocation in the AI industry, where chip manufacturers no longer wait passively for orders but actively finance the infrastructure that generates them. If the model proves successful, it could become a template for future megaprojects. Analysts will likely watch closely to see whether Nvidia chooses to report the risk from such guarantees more transparently in upcoming quarterly earnings reports.