Can $100 Billion Change the Future of AI? The Inside Story of Nvidia and OpenAI’s Mega Partnership

TradingKey - Recently, headlines around the world spotlighted Nvidia’s deal to invest up to $100 billion in OpenAI, a move that’s already being called one of the boldest bets in the history of artificial intelligence. That sum represents roughly one-fifth of OpenAI’s $500 billion valuation at the time of the announcement. For Nvidia, whose free cash flow topped $70 billion in FY2025, this massive outlay is less a stretch than a strategic play for long-term AI dominance.

But the story goes far deeper. Throughout this report, we will show you what’s really happening behind the headlines, reveal how Nvidia is poised to anchor the most ambitious AI platform ever built, and invite you to consider what this game-changing partnership could mean for AI’s future, and for anyone investing in AI’s next chapter.

Partnership Structure and Scale

To put this in numbers, 10 GW equates to roughly 4–5 million GPUs: each GPU plus its rack draws about 2 kW–2.5 kW of power, so dividing the total 10 GW capacity by 2 kW yields 5 million GPUs and by 2.5 kW yields 4 million GPUs.

In 2025 Nvidia expects to ship approximately 4-5 million (double from last year) data-center–class, AI-optimized GPUs such as the H100 and GB200 series, exactly the volume OpenAI needs to fulfill its 10 GW roadmap. This means OpenAI alone will absorb a full year of Nvidia’s AI-class GPU output, roughly 10–12% of Nvidia’s total GPU production. 

Under the terms of the partnership, Nvidia’s $100 billion commitment is divided into ten $10 billion equity tranches, each released only when OpenAI completes one gigawatt of live, operational Nvidia-powered AI capacity. No funds are disbursed at signing; the first $10 billion tranche becomes payable only after OpenAI brings its inaugural 1 GW data center online on Nvidia’s next-generation Vera Rubin platform in the second half of 2026. Each additional $10 billion tranche is similarly contingent upon OpenAI certifying and activating another 1 GW milestone. This milestone-based structure aligns Nvidia’s capital outlays directly with real-world build progress, mitigating execution risks and ensuring that every dollar deployed corresponds to actual infrastructure on the ground.

Data Center Build Process

So how does this monumental infrastructure actually come together? Each 1 GW phase of the Stargate project follows a set sequence:

OpenAI defines the required specs and identifies potential sites, Oracle handles facility design and construction, and Nvidia delivers GPUs to the completed data centers. Oracle then installs Nvidia hardware while OpenAI deploys and tests AI workloads. Once a 1 GW site is fully operational and independently audited, Nvidia releases the next $10 billion investment tranche. This milestone-driven process ensures that capital aligns closely with completed infrastructure and is repeated for each successive gigawatt.

Source: TradingKey

Since early 2025 OpenAI has been actively scouting potential U.S. locations, evaluating hundreds of site proposals across more than 16 states and breaking ground on its flagship Abilene, Texas campus, to lock in the sites that will host each gigawatt tranche of capacity, beginning in the second half of 2026.

OpenAI’s Multi-Channel Financing

A project of this magnitude requires more creative financing. Beyond Nvidia's hardware commitment, OpenAI's broader capital requirements center on the ambitious Stargate initiative, which is a $500 billion, four-year program launched in January 2025 to build 10 GW of AI infrastructure across the United States. This massive project operates as a multi-partner effort where OpenAI handles site strategy and AI requirements, Oracle constructs and operates 4.5 GW of data centers, and SoftBank provides capital and expertise for another 1.5 GW.

The chart below illustrates OpenAI's projected compute spending trajectory, showing costs ramping from $16 billion in 2025 to a peak of $111 billion in 2028, encompassing inference, R&D, and monetizable compute components. 

Source: The Information

The financing backbone of Stargate is OpenAI's record-breaking $300 billion (4.5 GW) long-term cloud contract with Oracle, the largest AI infrastructure deal in history, which guarantees Oracle revenue over five years and enables Oracle to secure external funding. Oracle leverages this contracted demand to borrow extensively in bond markets, recently raising $18 billion at 5–6% interest rates to finance rapid construction and expansion. Under this structure, OpenAI doesn't provide Oracle with upfront cash; instead, Oracle builds the facilities using borrowed capital and gets repaid as OpenAI consumes the contracted compute services.

Apart from this, in September 2024, SoftBank’s Vision Fund 2 injected $2.2 billion into OpenAI, with additional scheduled direct payments of $10 billion to OpenAI tied to the rollout of major data center sites in 2025. OpenAI also secured a $4 billion revolving credit facility in 2024 with major banks, carrying an interest rate of approximately 6% (SOFR + 100 basis points) and a three-year term with an option to extend. Combined with Nvidia’s $100 billion and Microsoft’s $10 billion equity investment in 2023 (holding 20-30% of shares), this financing structure allows OpenAI to avoid higher-cost borrowing common among AI startups.

Collectively, the various financing commitments from partners represent a significant portion of the Stargate program, but substantial additional capital will still need to be raised through future debt facilities, new equity partners, or alternative financing structures as construction milestones are achieved. 

Revenue Recognition & Valuation for Nvidia

Each gigawatt milestone is a financial windfall for Nvidia. Nvidia CEO Jensen estimates that each GW of AI data center capacity entails total capital expenditure between $50 billion and $60 billion, of which approximately $35 billion is dedicated to Nvidia chips and system integration. The remaining $15 billion–$25 billion covers all other site and facility costs. The following table shows a representative breakdown of cost structure:

Source: Jensen Huang, SemiAnalysis, Deloitte, TradingKey

Although only the first GW of Nvidia systems, equating to roughly $35 billion in chip and integration sales, has been contractually confirmed for deployment in the second half of 2026, this incremental revenue will be fully additive to Nvidia’s existing fiscal 2027 revenue base.

Scenario 1: Outright Purchases Hardware 

Nvidia will recognize the full $35 billion in hardware sales as soon as 1 GW goes live in H2 2026. Assuming a 50% net margin, justified by the premium pricing and steep operating leverage of AI infrastructure, conservatively aligning with last few quarters’ numbers. The current forward P/E of approximately 35x reflects Nvidia’s strong AI leadership and expanding our valuation multiple to 40x to show the intensifying investments by partners like Oracle that will accelerate the AI infrastructure boom, further driving Nvidia’s growth.

Scenario 2: OpenAI Leases GPUs

There is active discussion of OpenAI leasing GPUs, which would lower OpenAI’s upfront hardware costs by 10–15%. But in both scenarios, Nvidia receives full $35 billon to build data centers.

Note: FY2026 ended in Jan 2026, and FY2027 ended in Jan 2027.

Therefore, based on the first GW milestone, FY2027 target price for Nvidia is $207–$237, underpinned by accelerating AI infrastructure demand and sustainable high-margin. 

Nvidia’s Central Role in the Expanding AI Infrastructure Ecosystem

Nvidia’s AI ecosystem goes beyond chip design. This network strengthens Nvidia’s hold on cloud AI infrastructure and raises competition among AI chipmakers and cloud providers. Microsoft remains a major investor and Azure partner for OpenAI, but OpenAI has diversified compute with Oracle, SoftBank, and others. Meanwhile, Google, Amazon, and Meta invest heavily in their own AI chips and infrastructure. This map highlights Nvidia’s key role across chips, cloud, and data centers.

Source: TradingKey

Conclusion: Nvidia’s Strategic Win and Industry Impact

Concerns about a circular investment loop, where Nvidia invests heavily in OpenAI, which then purchases Nvidia hardware, raise questions about sustainability. However, this partnership secures Nvidia’s demand during a lasting AI supercycle, with phased, infrastructure-linked investments that mitigate risk.

Nvidia’s $100 billion investment reinforces its dominance by making its chips central to the largest AI infrastructure buildout in history. Nvidia’s revenue from this deal could reach $350-$500 billion over several years, boosting Nvidia’s leadership and accelerating consolidation across chipmakers, cloud providers, and data center operators.

While risks exist, Nvidia’s strategy is backed by expanding compute demand, strong financials, and extensive ecosystem control, suggesting this is far from a bubble now. But with new competitors on the horizon, massive capital still to be raised, and rapid shifts in global demand, everything could change. The true test will be whether this unprecedented investment can result in killer AI applications that deliver broad, sustained profitability, and if the future of AI lives up to its potential as a transformative growth engine for the entire industry.

Get Started