NVDA Signs Deal with Six Lenders, Shifting Financing Structure to Diversify Risk

Overview

This is a rating update report from BofA Securities regarding NVIDIA (NVDA), focusing on the event where NVIDIA signed an MOU with six top-tier financial institutions—Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR—to mobilize over $500 billion in third-party capital through independent platforms. BofA's preliminary assessment is positive: this financing structure places the bulk of capital risk on the consortium rather than on NVIDIA itself, thereby freeing up free cash flow for share buybacks. Simultaneously, by leveraging long-term performance guarantees from the CUDA ecosystem, it strengthens NVIDIA's moat in the AI computing market. The report maintains a Buy rating and a $350 price target, viewing NVIDIA as a top pick.

Core views

Shift in Financing Structure: BofA believes that NVIDIA signing an MOU with six top-tier financial institutions to mobilize over $500 billion in third-party capital marks a significant shift—from the previously questioned 'vendor financing' model (such as the initial $100 billion commitment to OpenAI and approx. $250 billion in guarantees) to a third-party financing model where risks are borne by independent platforms and consortia. The key logic lies in risk attribution: BofA's immediate judgment is that 'the burden falls on the consortium side, not on NVIDIA's balance sheet.' The report estimates that NVIDIA's direct ecological equity investments were approximately $70 billion, including $30 billion invested in OpenAI, up to $10 billion in Anthropic, $5 billion in Safe Superintelligence under Ilya Sutskever, $1.5 billion related to Lambda's GPU sale-and-leaseback, plans to lease back idle GPUs from neoclouds at fixed rates, and $860 million in lease obligations to undisclosed partners. These $70 billion in direct commitments represent only about 15% of the approx. $470 billion in free cash flow NVIDIA can generate over the next two years (CY26-27). BofA considers this fully manageable and believes it will not impact the execution of NVIDIA's May 2026 commitment to return 50% of free cash flow to shareholders. Valuation and Buyback Perspective: Current market consensus models assume buyback sizes of approx. $73 billion/$106 billion for CY26/27, accounting for only 36-37% of free cash flow, whereas NVIDIA's committed return ratio exceeds 50%. BofA believes that if the new financing structure allows more free cash flow to be used for buybacks, it is beneficial for a stock trading at a 'significantly suppressed' valuation of less than 20x CY27 expected PE.

Analysis framework

BofA's analytical主线 revolves around 'who bears the risk.' It first identifies that the essence of this MOU is a switch in financing mode—from vendor financing where NVIDIA directly provided guarantees to financing where third-party consortium independent platforms bear the risk. It then quantifies NVIDIA's existing direct commitments (approx. $70 billion) against its free cash flow capacity (approx. $470 billion over two years) to conclude that it is financially fully manageable. It then links this structural change to valuation (less than 20x CY27 PE, approx. 0.4x PEG) and buyback commitments to derive the logic supporting the stock price. The second main line focuses on 'where is the supply-demand bottleneck': The report argues that the bottleneck is capital, not demand. The $500 billion fund pool is equivalent to 2.5 times the scale of Google's network, allowing non-investment-grade buyers to lock in GPUs, power, and data centers at attractive interest rates, thereby supporting BofA's path assessment that the total addressable market (TAM) for AI systems will reach $1.7 trillion by 2030. The third main line is 'how the moat is strengthened by the financing structure': NVIDIA provides residual value guarantees rather than loan guarantees—the transferability and tradability of computing power, combined with CUDA continuously extending service life, keep resale and leasing rates high and depreciation curves healthy. This turns what short sellers feared as depreciation issues into a key element enabling the financing structure.

Methodology notes

• Company Fundamentals & Financial FrameworkFree cash flow analysis

  Measures the financial capacity for ecological investment and shareholder returns using future free cash flow (FCF) size

  The report compares NVIDIA's approx. $70 billion direct ecological investment with its approx. $470 billion free cash flow over two years, concluding it accounts for only 15%, thereby judging that financing commitments will not crowd out cash needed for buybacks.

• Valuation methodsPE/PEG valuation

  Assesses valuation attractiveness using forward PE and PEG (PE divided by earnings growth rate)

  The report notes that NVDA trades at less than 20x CY27 expected PE, corresponding to approx. 0.4x PEG, while EPS compound growth exceeds 46%, considering the valuation to be significantly suppressed.

• Valuation methodsSOTP Segment Valuation

  Sets target price using segment valuation ex-cash (ex-cash PE)

  The $350 price target is based on 26x CY27 expected PE (ex-cash), within NVIDIA's historical range of 25-56x forward PE.

• Industry/Industrial Analysis FrameworkPenetration S-curve

  Provides an anchor for long-term company growth using the path of TAM expansion driven by AI penetration

  The report uses the benchmark of AI system TAM reaching $1.7 trillion by 2030, combined with NVIDIA's approx. 65-70% market share, to derive long-term revenue space.

• Event Gaming & Behavioral FinanceAsset Residual Value & Depreciation (Asset Value Preservation Perspective)

  Views GPUs as an investable asset class, core to whether residual value can be maintained

  The report believes NVIDIA's core contribution lies in long-term performance guarantees: the CUDA ecosystem continuously extends GPU service life, making computing assets transferable and tradable. High resale/leasing rates and healthy depreciation curves turn the short sellers' concern about rapid GPU depreciation into a support for the establishment of the financing structure.

Asset mapping & comparison

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

• NVIDIA Corp (NVDA.US)

  Direct Beneficiary: Financing structure diversifies balance sheet risk, frees up free cash flow for buybacks, and strengthens the CUDA ecosystem moat

  Strengths

  Approx. 65-70% AI computing market share; strong lock-in effect from CUDA software ecosystem; approx. $470 billion free cash flow over the past two years; 46%+ EPS compound growth

  Weaknesses

  Market expectations are extremely high; stock price has repeatedly faced pressure after the last four earnings reports; the new financing structure may increase complexity and opacity in the AI construction chain

  Comparison

  BofA views this as a top pick within coverage, trading at less than 20x CY27E PE, approx. 0.4x PEG, significantly below its historical range of 25-56x

  Risks

  Weakness in consumer-driven gaming market; competition from large listed companies, cloud vendors' self-developed projects, and private companies; greater impact of chip export restrictions to China; uncertainty in new market sales; deceleration in capital returns; government antitrust review of its dominant position in AI chips

• Apollo / BlackRock / Blackstone / Goldman Sachs / KKR

  Participating in financing as third-party capital platforms and consortium members, benefiting from the expansion of computing power as an investable asset class

  Strengths

  Opportunity to participate in massive AI infrastructure financing, benefiting from asset management and financing arrangements

  Weaknesses

  Bear the primary capital risk in financing, needing to independently assess residual values of computing assets and final customer returns

  Risks

  Third-party capital still requires real end-customer payments; MOUs do not equal actual deployed funds

Key data

• Scale of Third-Party Capital MobilizationOver $500 BillionMobilized through independent platforms of six financial institutions, approx. 2.5 times the scale of Google's network
• NVIDIA Direct Ecological Equity InvestmentApprox. $70 BillionIncludes $30 billion investment in OpenAI, up to $10 billion in Anthropic, $5 billion in Safe Superintelligence, etc.
• Free Cash Flow for Next Two Years (CY26-27)Approx. $470 BillionThe $70 billion direct investment accounts for approx. 15% of this; BofA considers it financially fully manageable
• Consensus Expectation for CY26/27 BuybacksApprox. $73 Billion / $106 BillionAccounts for only 36-37% of free cash flow, lower than NVIDIA's commitment to return over 50%
• 2030 AI System TAMOver $1.7 TrillionNVIDIA aims to capture approx. 65-70% market share
• Valuation LevelLess than 20x CY27E PE, Approx. 0.4x PEGCompared to 46%+ EPS compound growth, BofA considers the valuation significantly low
• Target Price$350Based on 26x CY27E ex-cash PE

Impact & implications

BofA believes this financing structure is a structural benefit to NVIDIA: it transfers capital risk from NVIDIA's balance sheet to a consortium of six financial institutions, while allowing NVIDIA's free cash flow to be more centrally used to support share buybacks. From an industry perspective, the $500 billion fund pool enables non-investment-grade buyers (AI labs, new cloud service providers, sovereign wealth funds, etc.) to acquire GPUs, power, and data center resources at attractive interest rates, directly alleviating the bottleneck of 'capital, not demand' and reducing hedging demand risks across the entire AI construction chain. From a competitive landscape perspective, by providing long-term performance guarantees through the CUDA ecosystem, NVIDIA essentially backs the residual value of computing assets. This reinforces its core position in AI infrastructure—it not only sells chips but also provides a value anchor for the entire AI asset class, further deepening its moat.

Risks

• Third-party capital still requires real end-customers to pay hard cash; MOUs do not equal actual deployed funds
• Diversifying risk does not answer all ROI questions; uncertainty in computing power returns remains
• Power bottlenecks and regulatory backlash (e.g., cancellation of UK NScale project) along with persistent input cost inflation
• New financing structures may increase complexity and opacity in the AI construction chain, potentially suppressing valuation multiples
• If NVIDIA still needs to bear larger cash flow demands, it may crowd out free cash flow needed for buybacks
• Downside risks such as weak gaming market, intensified competition, expanded export restrictions to China, and government antitrust reviews

What to watch

• NVIDIA's earnings call in late August: Its future role in vendor financing and how free cash flow will be allocated
• Follow-up implementation of MOUs: Actual deployment progress of third-party capital and attracted end-customers
• The role of NVIDIA's next-generation Vera Rubin product cycle in consolidating market share (approx. 70%)
• Impact of changes in power and regulatory environments on the advancement of AI projects