NVIDIA is reshaping its business model in an unprecedented way, transitioning from a chip manufacturer to an "AI Infrastructure Financial Partner," with its core being a new cloud service revenue-sharing model based on a "guarantee for profit-share" logic.
NVIDIA is actively advancing revenue-sharing agreements with emerging AI cloud service providers (neo-clouds). Under these agreements, NVIDIA provides a credit backstop for the initial construction of these cloud providers in exchange for a percentage of their cloud revenue. According to the desk, Morgan Stanley analyst Joseph Moore likened this mechanism to "AI financial innovation," suggesting its potential scale is enormous. If the neo-cloud ecosystem develops to the equivalent of 25 gigawatts of computing power, comparable to the largest hyperscale cloud providers, it could generate a $500 billion annual revenue pool, based on a leasing price of $20 million per megawatt. If NVIDIA were to capture a quarter of the profit share, at a 100% gross margin, this could provide up to a 60% upside to its FY28 earnings before interest and taxes (EBIT).
Morgan Stanley maintains an "Overweight" rating on NVIDIA, designating it as a top pick in the semiconductor sector with a price target of $288, compared to the current stock price of $225.30. The analyst pointed out that this new business model offers both high upside potential and limited downside risk, but simultaneously acknowledged that market skepticism regarding its recurring revenue structure will further intensify the bull-bear divide.
Guarantee for Profit-Share: Creating Annuity-Style Income
According to Morgan Stanley's report, NVIDIA's new neo-cloud partnership model operates around a core mechanism. NVIDIA provides a credit backstop to AI cloud providers by committing to pay a specific hourly rental price for GPUs, enabling these providers to secure financing at near-investment-grade costs for data center construction. In return, NVIDIA receives a share of the revenue that exceeds the guaranteed price. Citing tracking of NVIDIA's CEO roadshow, Morgan Stanley notes that the two parties negotiate the guaranteed hourly GPU price. The AI cloud provider uses this guarantee to obtain financing, while NVIDIA receives a revenue share on the portion of the leasing price that exceeds the guarantee, with the exact percentage varying by agreement. The report also notes that in some cases, NVIDIA's backstop is not a pure cash credit guarantee but rather a consideration for securing GPU usage rights from the cloud provider for its own R&D—similar to its previous contract arrangements with CoreWeave, but with greater upside potential for NVIDIA.
In its sensitivity analysis, Morgan Stanley assumes NVIDIA receives a 35% revenue share on the portion exceeding the guaranteed price. Based on a $1 billion GB300 data center unit, under scenarios with GPU rental prices ranging from $10 to $16 per hour, every 5 gigawatts of computing power could generate incremental annual revenue of $37 billion to $65 billion for NVIDIA, boosting FY29 earnings per share (EPS) by approximately 7% to 13%. Even in a more conservative scenario of 2 to 5 gigawatts, the EPS uplift could be around 10%.
Strategic Rationale: Hyperscaler Underinvestment, NVIDIA Seeks New Growth Engine
The Morgan Stanley report reveals the strategic logic behind NVIDIA's move. NVIDIA management believes that traditional hyperscale cloud providers, constrained by free cash flow pressures and increasingly scarce land, power, and facility resources in the US, are unable to expand capital expenditure fast enough to match the growth in token demand. They are even being forced to divert computing power originally intended for model training to inference scenarios. Concurrently, geopolitical factors are accelerating this shift. As US cloud providers face policy pressure when building data centers in other countries, some nations are beginning to treat their local land and power resources as strategic AI assets, fostering local sovereign clouds or regional neo-cloud providers. Morgan Stanley cites Yotta, YTL, and SK Telecom as typical examples.
Against this backdrop, NVIDIA is choosing to intervene proactively, cultivating a new group of computing power buyers through capital backstopping. This allows NVIDIA to bypass the capital expenditure constraints of hyperscale cloud providers while strengthening its dominant position in the global computing power landscape. The report points out that the capital expenditure of publicly listed neo-clouds currently accounts for approximately 6% of global cloud capex, and this share is growing rapidly, not yet inclusive of sovereign and private cloud players.
Existing Footprint: From CoreWeave to a $500 Billion Industrial Fund
NVIDIA's presence in the neo-cloud space has been established for some time. According to the Morgan Stanley report, since investing in CoreWeave in 2023, NVIDIA has extended this strategy to several other cloud providers. It currently holds over 10% equity in CoreWeave and has committed to purchasing all of its unsold computing power until April 2032. NVIDIA also holds approximately 10% equity in Nebius, has signed a $3.4 billion, 5-year cloud services contract with IREN, including $2.1 billion in warrants, and has made private equity investments in Nscale, Crusoe, Lambda, and Firmus.
On the software front, NVIDIA has launched the DGX Cloud Lepton platform, a marketplace that connects developers with GPU computing power from its cloud partners, dynamically routing computing demand to cloud providers with available capacity. On the hardware infrastructure side, NVIDIA has also taken concrete steps recently. Reports indicate that NVIDIA, in partnership with others, has announced the construction of 16 new fiber optic routes in the US, spanning 8,000 miles of long-haul fiber, with neo-cloud support explicitly cited as a project driver. Other media reports suggest NVIDIA is signing a 15-year, $20 billion lease for HUT 8's 1-gigawatt Beacon Point campus. On the capital operations front, NVIDIA recently announced the mobilization of a $50 billion investment fund, explicitly stating in its official announcement that the fund will be used for "supporting long-term usage-linked revenue." Morgan Stanley notes that the fund's mechanism limits NVIDIA's maximum direct exposure to any single investment to 25%, which mitigates cyclical risk to some extent.
Bull vs. Bear: High Growth Potential and Cyclical Risk Coexist
Morgan Stanley acknowledges that the new model has sparked significant divergence in the market. On the bullish side, the argument is that, firstly, computing power supply continues to lag behind token demand growth, and new computing power has genuine demand support. Secondly, the annuity-style income from holding minority stakes in multiple cloud providers will enhance the predictability of NVIDIA's long-term earnings. Thirdly, this move helps NVIDIA solidify its chips as the industry standard ahead of the new Vera Rubin product cycle—for smaller cloud providers, NVIDIA's chips, with their approximately 85% market share and most comprehensive ecosystem, remain the optimal choice.
The bearish side focuses on the following risks: the model could create "artificial demand"—meaning that without NVIDIA's credit backstop, some computing power investments would not be viable. Additionally, the capital strength of participants varies, and the overall model could face significant pressure in a scenario of computing power oversupply. Morgan Stanley comments that high transparency in disclosures is a key prerequisite for maintaining market trust in this model. In a downside scenario, Morgan Stanley also points out a potential hedging path: if computing power becomes oversupplied, NVIDIA could use the acquired GPU usage rights to advance its own model development. NVIDIA's current flagship open-source model, Nemotron 3 Ultra, with 550 billion parameters, is already a leader in the open-source model market outside of China. The next-generation Nemotron 4 is expected to surpass the trillion-parameter threshold, further solidifying its competitiveness in model development.
Impact on Existing Neo-Clouds: Increased Competition, Execution is Key
Morgan Stanley notes that NVIDIA's move is a double-edged sword for existing neo-cloud players. Over the long term, the increase in potential competitors creates pressure. However, in the short term, cloud providers with a reliable track record of operations have the opportunity to secure more new computing power investments. The report emphasizes that NVIDIA's credit backstop only solves the financing problem. Cloud providers still need to complete construction on time and convert it into revenue within a complex geopolitical and supply chain environment—this places a very high demand on operational execution.
Overall, Morgan Stanley maintains NVIDIA as its top pick in the semiconductor sector, with a price target of $288, based on approximately 22 times CY27 P/E valuation. The analyst expects NVIDIA's revenue to grow 82% year-over-year in FY2026, followed by a further 52.4% growth in FY2027. The FY29 EPS forecast is $17.63.
Comments