To The Moon

NVIDIA’s announcement to mobilize up to $500 billion in third-party capital for AI infrastructure has been met with a strong defense from Morgan Stanley, which views the structural arrangement as fundamentally resolving market concerns about "circular financing" while opening a high-margin, recurring revenue stream for the chipmaker.

On August 10, NVIDIA NVIDIA revealed a partnership with six institutional investors to form a so-called "independent computing capacity financing platform," with a potential total investment scale of $500 billion dedicated to AI factory construction. Within this structure, NVIDIA’s role is limited to providing a "residual value support mechanism" of no more than 25% of any single investment, while the decision-making authority and capital sourcing are entirely managed by independent third parties.

In a research note released on August 11, Morgan Stanley highlighted that the core value of this arrangement is twofold: the third-party-led decision-making process effectively reduces skepticism about NVIDIA’s potential "self-reinforcing" investment, and the revenue-sharing agreements create a repeatable, near-100% gross margin annuity-like income stream for the company.

Morgan Stanley analyst Joseph Moore estimates that if NVIDIA captures a 35% share of GPU usage revenue above the breakeven point, the annual revenue from per-watt computing capacity could exceed $100 billion by fiscal year 2029, translating to an earnings per share upside of more than 10%.

Morgan Stanley maintains an overweight rating on NVIDIA with a price target of $288, implying about 32% upside from the current share price of $217.55.

Structure Design is Key to the $500 Billion Mobilization

Morgan Stanley emphasizes that NVIDIA’s announcement is not a committed investment but rather the formation of a platform with six institutional partners, with a potential total investment scale of $500 billion. Each partner will independently evaluate projects before deciding on capital deployment, and no specific investment decisions have been made yet.

Within this structure, NVIDIA is defined as an "ecosystem supporter" and provides a "residual value support mechanism" capped at 25% of each investment. NVIDIA describes this mechanism as "limited, residual-value-based support designed to supplement, not replace, independent underwriting."

The key significance of this design, according to Morgan Stanley, is that NVIDIA’s participation is strictly limited to a minority stake, with decision-making authority resting with professional third-party investors. This structurally cuts off the circular logic of "NVIDIA lending to customers, who then buy NVIDIA chips."

NVIDIA CEO Jensen Huang previously stated on social media that this move was "specifically designed to address concerns about circular financing."

Circular Financing Concerns: A Risk That May Be Overstated

Morgan Stanley addressed and clarified the market narrative around "circular financing" in its report, noting that NVIDIA’s direct credit exposure to date is actually quite limited, primarily concentrated in credit guarantees for a small number of emerging cloud service providers (neoclouds). According to NVIDIA’s latest 10-Q filing, the total exposure from lease guarantees to partners is approximately $3.5 billion.

Additionally, while NVIDIA has made equity investments in some customers and ecosystem partners, Morgan Stanley emphasizes these are equity investments, not grants or credit guarantees. For publicly traded investments, the portfolio is performing well overall; returns on private company investments are difficult to assess precisely but remain equity-like in nature.

Morgan Stanley also notes that some media reports tend to exaggerate NVIDIA’s off-balance-sheet risks, citing large credit arrangements that have not yet materialized and are planned to unfold over the next decade. The report concludes that NVIDIA’s overall credit exposure management remains prudent.

Revenue Sharing: Creating a High-Margin Annuity Channel

Another strategic layer of this financing structure is its deep integration with NVIDIA’s revenue-sharing plan.

Morgan Stanley points out that NVIDIA’s press release explicitly states the platform will "enable long-term usage-based revenue," meaning relevant investment projects will be incorporated into NVIDIA’s disclosed revenue-sharing framework.

According to Morgan Stanley’s model, if NVIDIA captures a 35% share of GPU usage revenue above the breakeven point, assuming a market GPU rental price of $13 per hour and 5 GW of new computing capacity added annually, NVIDIA could generate approximately $51 billion in incremental revenue, corresponding to a 9.83% EPS upside for fiscal year 2029.

The key characteristic of this revenue stream is its gross margin structure. Since NVIDIA does not bear hardware depreciation or operating costs in the revenue-sharing arrangement, the associated revenue is near 100% gross margin and is recurring, similar to the annuity nature of software subscriptions or cloud services.

Morgan Stanley views this as a significant step in NVIDIA’s transition from a "hardware seller" to an "AI infrastructure ecosystem operator."

Potential Risks: Uncertainties Remain

While acknowledging the positive aspects of the structure, Morgan Stanley also lists several counterarguments worth noting.

Capital deployment uncertainty is a key factor. The $500 billion figure is a nominal potential investment scale, and each partner will independently evaluate projects on a case-by-case basis, with no mandatory capital deployment mechanism. Actual deployment could vary significantly.

NVIDIA still bears some credit risk. Even with the 25% participation cap, NVIDIA will retain credit exposure in some projects. Morgan Stanley expresses a preference for NVIDIA to provide support in the form of GPU usage rights rather than pure credit guarantees, as the former has a lower actual cost, given NVIDIA’s own need for GPUs in developing its open-source models.

Rising ecosystem leverage could spark debate. Even if NVIDIA itself does not provide leverage, the overall debt level of the AI ecosystem will rise as sovereign funds and emerging cloud providers make large-scale investments. Morgan Stanley notes that some major hyperscale cloud providers are currently in a cash-flow-neutral or even negative state, and there is disagreement over whether the AI ecosystem can sustain increasing leverage.

Morgan Stanley Maintains Top Pick: Valuation Remains Attractive

Overall, Morgan Stanley believes the positive implications of this financing structure significantly outweigh the potential risks, maintaining an overweight rating on NVIDIA with a price target of $288 and naming it a top pick in the semiconductor sector.

Morgan Stanley forecasts that NVIDIA’s revenue will grow 82% year-over-year in fiscal 2026 and 52.4% in fiscal 2027, corresponding to EPS of $4.77 and $8.98, respectively.

Based on Morgan Stanley’s ModelWare framework, the fiscal 2027 EPS estimate of $8.96 implies a price-to-earnings ratio of about 24 times, which is lower than comparable computing semiconductor companies like AMD and Broadcom, making the valuation relatively attractive.

Meanwhile, Morgan Stanley notes that NVIDIA’s Blackwell series remains the mainstream solution for generative AI workloads, with computing demand continuing to outpace supply. The next-generation Rubin architecture is expected to extend its performance lead, supporting the logic that the product cycle underpins the valuation.