Written by: Rita

NVIDIA has signed a memorandum of understanding with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR to mobilize over $500 billion in third-party capital through an independent platform. Bank of America noted in its report on August 10 that this financing structure shifts the capital burden from NVIDIA's balance sheet to the consortium, marking a significant shift from the previous vendor financing model that provided $100 billion to $250 billion to OpenAI. Bank of America estimates that NVIDIA's past equity investments in the ecosystem total approximately $70 billion, which accounts for only 15% of the expected free cash flow of about $470 billion from 2026 to 2027, and does not affect the company's previously committed ability to return 50% of free cash flow to shareholders.

Third-party financing accounts for only 15% of FCF, does not affect shareholder returns

Bank of America estimates that NVIDIA has committed approximately $70 billion in direct equity investments to ecosystem partners. This includes a recent $30 billion investment in OpenAI, up to $10 billion in Anthropic, and a $5 billion investment in Ilya Sutskever's Safe Superintelligence. Additionally, there is a $1.5 billion GPU sale-leaseback with Lambda, a plan to lease back unused GPUs to neocloud, and a $860 million leasing obligation for an undisclosed partner.

Bank of America believes that considering NVIDIA can generate about $470 billion in free cash flow over the next two years, the $70 billion investment scale is fully manageable and does not affect the company's previously committed promise to return 50% of free cash flow to shareholders.

GPU is an "investable asset class," with residual value supported by the CUDA moat

In order for $500 billion of capital to regard computing power as an "investable asset class," residual value must be guaranteed. NVIDIA precisely provides this guarantee. Computing power can be transferred across operators, CUDA continues to extend its lifespan, resale and leasing prices remain high, and the depreciation curve remains healthy.

Bank of America believes that NVIDIA guarantees asset quality, not the debt itself. The asset quality itself is combating depreciation, which is a core difference between NVIDIA and traditional hardware suppliers. The shorts' concern of "depreciation eating away profits" does not hold here.

$500 billion financing extends AI construction window

Bank of America points out that the bottleneck is funding, not demand. Every major provider is complaining about limited supply. The $500 billion funding pool (equivalent to 2.5 times the scale of Google's network) allows non-investment-grade buyers to obtain GPUs, power, and data centers at favorable rates, reducing purchasing risks.

This effectively transforms AI computing power from a "capital-intensive purchase" to a "financing lease acquisition." The $500 billion is not the endpoint; it turns the AI landscape from "who can spend money" to "who needs computing power." Bank of America expects this structure will support a $17 trillion AI system TAM by 2030.

Five major risks still need attention

Bank of America also pointed out five risk points. Third-party capital still needs actual end customers to pay real funds; a memorandum of understanding does not equal deployed capital. Transferring and diversifying risks does not mean every ROI issue has been answered. Power supply and regulatory resistance still exist, as evidenced by the cancellation of the UK's NScale project, and cost inflation is ongoing. The new financing structure may increase the opacity and complexity of AI construction, putting pressure on transaction multiples. Although Bank of America believes this arrangement limits NVIDIA's further exposure on its balance sheet, the company's exact role in this structure still needs more details, and any cash demand beyond expectations could squeeze the free cash flow available for buybacks.

Bank of America maintains a buy rating and a target price of $350, based on a 26 times expected earnings per share in 2027. Trading at less than 20 times the expected price-to-earnings ratio for 2027, 0.4 times PEG vs. 46%+ compound earnings per share growth rate, and capturing 65% to 70% of the market share in a $17 trillion TAM by 2030, NVIDIA remains attractive. This financing platform consolidates the CUDA moat while shifting capital risk away from NVIDIA, constituting a structural benefit. The upcoming earnings call will be the next significant catalyst. Bank of America believes that the Vera Rubin product cycle will further solidify the company's approximately 70% market share.

Disclaimer

This article is a整理与解读 of a third-party brokerage research report (Bank of America Securities, August 10, 2026) by Chao Xiang Research, combined with整理 of publicly available market information. The ratings, target prices, earnings forecasts, and related judgments quoted in the text are solely the opinions of the brokerage's analysts, representing the stance of their respective organizations, and do not represent the views of Chao Xiang Research, nor do they constitute any investment advice.

Markets carry risks; decisions should be made independently. This article should not be used as the basis for buying or selling any securities.