Third-party capital leverages AI infrastructure investment, opening new earnings opportunities for NVIDIA through revenue sharing

Overview

The report assesses the impact of NVIDIA working with six institutional investment partners to promote “independent compute financing platforms.” These platforms intend to evaluate AI factory investment opportunities based on the economics of individual projects, with potential investment scale of $500 billion; NVIDIA provides ecosystem support and may provide limited residual value support of no more than 25% for some investments. Morgan Stanley believes this structure is primarily driven by third-party capital, underwriting, and decision-making, helping alleviate market concerns about circular financing, while supporting the expansion of sovereign AI and neo-cloud customers and potentially creating long-term high-margin revenue through usage-linked revenue sharing.

Core views

The core judgment is that the positives significantly outweigh the negatives. As AI compute is expected to remain supply-constrained for many years, new capital can promote AI factory construction and expand demand for NVIDIA hardware. If NVIDIA obtains a share of project profits or cloud revenue in exchange for limited support, its returns could translate into high-margin, repeatable annuity-like revenue. Project selection led by professional third-party investors is also more favorable than NVIDIA directly providing large-scale credit support in reducing circular-financing concerns. However, the $500 billion is only a potential investment amount, with no guarantee that capital will necessarily be deployed; NVIDIA may still bear credit risk, and rising ecosystem leverage will continue to trigger investor debate.

Analysis framework

The report first reviews NVIDIA’s existing exposure to credit guarantees and equity investments, then compares the risk differences between third-party-led financing and direct corporate credit support; it then conducts sensitivity analysis based on Morgan Stanley ModelWare forecasts, AI data center cost assumptions, GPU hourly pricing, minimum guaranteed pricing, and a 35% revenue-sharing ratio, and forms the rating and price target by combining bull, base, and bear case scenarios with relative P/E valuation.

Methodology notes

• Earnings ForecastMorgan Stanley ModelWare

  Standardized financial forecasting model

  Unless otherwise specified, the revenue, margins, EPS, and valuation metrics in the report are based on the Morgan Stanley ModelWare framework; data marked as estimates are Morgan Stanley Research forecasts.

• Sensitivity AnalysisAI Factory Revenue-Sharing Sensitivity Analysis

  Measures the impact of changes in GPU pricing, project scale, and sharing ratio on earnings

  The model uses assumptions including GB300 data center costs, market prices for GPU hours, cost-recovery minimum guaranteed prices, 75% annual utilization hours, and 35% excess revenue sharing to estimate incremental revenue, gross margin, and FY29 EPS uplift.

• Scenario ValuationBull-Base-Bear Case Scenario Analysis

  Builds a price range based on earnings and valuation multiples

  The report presents bull, base, and bear case scenarios of $330, $288, and $160, respectively, reflecting variables such as data center growth, full-stack monetization, competition, customer in-house chips, and export restrictions.

• Relative ValuationForward P/E Comparison

  Determines target price based on forecast EPS and peer valuations

  The base target price corresponds to approximately 22x Morgan Stanley’s CY27 EPS forecast of $13.08, close to the broader market and below the valuation levels of compute semiconductor peers such as AMD, AVGO, and INTC.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• NVIDIA Corp. (NVDA.US)

  Directly covered company, the main beneficiary of the AI factory financing and revenue-sharing mechanism as well as a potential bearer of credit risk.

  Strengths

  High market share in AI accelerated computing, strong Blackwell and Rubin product cycles, and system-level solutions that enhance long-term monetization capability; third-party capital can expand GPU demand, and revenue sharing may form high-margin recurring revenue.

  Weaknesses

  High market share and gross margin limit further near-term expansion of valuation multiples; the company’s investments, guarantees, or residual value support for partners make credit risk assessment more complex.

  Comparison

  The base valuation is approximately 22x CY27 EPS, close to the broader market and at a discount to compute semiconductor peers such as AMD, AVGO, and INTC.

  Risks

  Potential capital may not be deployed, project returns may be insufficient, credit support losses, rising AI ecosystem leverage, cooling data center demand, competition from AMD or customers’ in-house chips, tariffs and export controls, and weaker-than-expected AI end-device demand.

Key data

• Potential investment mobilization scale$500bnTo be evaluated by six institutional partners for independent projects, and not committed or guaranteed deployment capital.
• Potential NVIDIA support ratioUp to 25%May take the form of co-investment or limited residual value support, with the specific structure to be evaluated project by project.
• Existing partner lease guarantee exposure$3.5bnThe report cites data disclosed in NVIDIA’s most recent 10-Q.
• Revenue-sharing assumption35%Refers to the revenue-sharing ratio after exceeding an approximate breakeven level, used for sensitivity analysis.
• Potential FY29 EPS upliftMore than 10%This is a scenario result under specific cost, pricing, and project scale assumptions, not a confirmed benefit already incorporated.
• FY29 EPS sensitivity range4.41%—12.54%Shows the impact of different revenue-sharing scenarios on Morgan Stanley’s FY29 EPS forecast.
• Price target$288.0012-month base price target; the bull case is $330 and the bear case is $160.
• Reference closing price$217.55As of the close on 2026-08-10, corresponding to approximately 32.4% potential upside to the target price.
• CY27 EPS forecast$13.08The base valuation uses approximately 22x forward P/E.
• Revenue growth forecast82.0% in 2026; 52.4% in 2027Revenue growth assumptions in the report’s base case.

Impact & implications

For NVIDIA, third-party capital can expand AI infrastructure construction and GPU procurement while reducing the company’s direct capital burden, and may extend one-time hardware sales into long-term usage-based revenue sharing. For valuation, if this model is successfully implemented, the incremental revenue will have high gross margins and strong repeatability, potentially driving upward revisions to earnings forecasts; however, the market will also pay closer attention to credit support terms, off-balance-sheet risks, and the true economic returns of partnership projects. For the AI value chain, this mechanism helps sovereign institutions and neo-cloud companies supplement hyperscaler capital expenditure, but it may also increase leverage and financing dependence across the ecosystem.

Risks

• The $500 billion is a nominal potential investment scale; each project must be independently reviewed, and there is no guarantee that capital will ultimately be deployed.
• NVIDIA may bear up to 25% co-investment or residual value support, and specific transactions may still bring credit losses and off-balance-sheet risk concerns.
• If the form of support is a pure credit guarantee rather than a GPU usage commitment, NVIDIA’s economic cost and tail risk may be higher.
• Financing for sovereign AI and neo-cloud projects will raise overall leverage in the AI ecosystem, and some hyperscalers have approached or entered negative free cash flow due to upfront investment.
• If compute supply catches up with demand more quickly, the AI end market fails to materialize, or customers reduce GPU purchases, data center growth may slow significantly.
• A recovery in AMD’s competitiveness, cloud customers expanding in-house chips, and declining AI development costs may erode NVIDIA’s market share and pricing power.
• Tariffs, export controls, and weaker-than-expected recovery in China market revenue may suppress growth.
• Morgan Stanley has investment banking and other business relationships with NVIDIA, and the report discloses related potential conflicts of interest.

What to watch

• The identities of the six institutional partners, fundraising progress, and the actual deployment ratio of the $500 billion potential capacity.
• NVIDIA’s capital contribution ratio, form of credit support, priority, term, and loss-bearing mechanism in individual projects.
• Whether cooperation agreements explicitly include usage-linked revenue sharing, as well as the sharing ratio and starting threshold.
• GPU utilization, customer acquisition capability, cash flow, and refinancing conditions of neo-cloud and sovereign AI projects.
• Whether partner lease guarantee exposure continues to rise from the currently disclosed $3.5bn.
• Whether AI token demand can maintain approximately 10x growth, and whether compute remains in sustained short supply over the next several years.
• Blackwell and Rubin product cycles, customer capital expenditure commentary, and data center revenue growth.
• The impact of AMD, customers’ in-house accelerators, export restrictions, and tariff changes on NVIDIA’s share and China business.