Bank of America Reaffirms $350 NVDA Target on AI Financing Thesis

NVDA News

Bank of America has kept its $350 price target for Nvidia (NVDA), arguing that the shares trade 34% to 50% below fair value after the company moved to cover up to $105 billion in lease obligations at a new data center campus in Ohio. Analyst Vivek Arya laid out the case in a note published Monday, the same day Nvidia’s official announcement confirmed the guarantee for a site developed by SB Energy, a company backed by SoftBank and OpenAI. Located on a Cold War-era uranium enrichment parcel in Pike County, the campus will be leased to OpenAI for 20 years. Wall Street’s worry, Arya said, is that Nvidia is quietly becoming a bank for the AI boom: it sells GPUs to OpenAI, has pledged as much as $100 billion under a 2025 partnership, and now stands behind OpenAI’s rent. Critics describe the money loop as circular. Arya argues the market is misreading the structure. Nvidia does not guarantee OpenAI’s full rent; it covers only the gap if OpenAI defaults and the site is re-leased or sold, with the total bill capped at $105 billion rather than the $250 billion floated in earlier reports. In return, Nvidia becomes the exclusive AI compute provider on the campus, keeping rivals away from scarce land and power. CEO Jensen Huang framed the tradeoff in the release: “land, power and shell have become vital in the age of AI.” From the $219.74 level where NVDA traded at the time of the note, a move to $350 would be roughly 59% upside and would add about $3 trillion to a $5.45 trillion market capitalization. In Arya’s view, ecosystem investments of this kind, especially into frontier labs and neoclouds — the smaller cloud firms built to rent out GPUs — are critical to accelerating the AI cycle even if they pressure earnings quality today.

The bullish call sits well above the Street average and highlights how far Nvidia’s valuation has fallen in the AI-bank debate. Analyst consensus data shows 36 of 37 analysts rate NVDA a Buy; the average target stands at $309.94, and even the lowest target on the Street, $250, is above the $219.74 level where NVDA traded at the time of the note. Arya arrives at his $350 target with a sum-of-the-parts model built on free cash flow, and he says the shares still trade 34% to 50% below fair value after every financing risk is loaded in. His proposed remedy is also part of the thesis: Nvidia returns only about half of its free cash flow through buybacks, while peers return 75% to 100%. A larger buyback program would hand cash back to shareholders, ease doubts about earnings quality and could lift the multiple. That proposal is aimed at a specific criticism: that Nvidia’s rising commitments to OpenAI and other partners are inflating its reported earnings while true cash returns to shareholders lag behind peers. The nearest catalyst is August 26, when Nvidia reports earnings and is expected to separate its off-balance-sheet commitments for the first time. Arya says a clean disclosure could start closing the discount he identifies in the near term. The risks are equally concrete: if AI demand cools, re-leasing a huge campus in Ohio becomes much harder, and the stock has fallen after earnings six times since August 2024. That history explains why traders are treating the upcoming report as a binary catalyst for NVDA, with AI trading bots and options desks likely to react instantly to the numbers.

Both stories point to one arc: investors are pricing NVDA as an AI financier, not just a chipmaker. Nvidia’s official announcement caps the guarantee at $105 billion and gives it exclusive AI compute rights on the Ohio campus, turning an open-ended credit line into a contained backstop. Until August 26, the uncertainty premium will be difficult to remove. If AI demand weakens, re-leasing that campus becomes harder, and the bear market narrative could return. The trust test resembles AI crypto wallets: protection only matters at the moment of default. NVDA now trades with narrative sensitivity familiar from altcoin markets, and the upcoming disclosure will show whether the backstop is a real shield or a financial shell.