Nvidia’s pitch now goes beyond selling chips. As Advanced Micro Devices and Google parent Alphabet chip away at its technology lead, Jensen Huang is leaning on an advantage that’s much harder to match: capital.
With free cash flow up 18-fold over three years to $48.5B last quarter, Nvidia is using its balance sheet to deepen its reach across the AI economy.
Nvidia finances its own moat
Huang’s recent moves make that strategy hard to miss. Nvidia agreed to back up to $105B in guarantees for OpenAI’s massive Ohio data center and partnered with Goldman Sachs, Apollo, and BlackRock among others to mobilize more than $500B in third-party capital for AI infrastructure.
Bank of America estimates Nvidia has committed roughly $300B to ecosystem partners, including about $70B in equity investments and $230B in residual value guarantees.
The logic is simple. Frontier AI labs like OpenAI are growing faster than their balance sheets can support, and Nvidia is helping close the funding gap while keeping demand flowing toward its infrastructure.
Cantor analysts see this as evidence of an AI investment cycle that will be “elongated and durable.” Nvidia's equity holdings in AI companies reached $30.2B as of the most recent quarter, up from $12.9B a year earlier.
The company pledged to return roughly 50% of free cash flow to shareholders this year through buybacks and dividends, including an $80B repurchase plan and a raised quarterly dividend of 25 cents per share.
Where the thesis cracks
None of this is without risk. Bank of America’s Vivek Arya says Nvidia is moving to secure every critical input behind the AI buildout, from chips and land to power and infrastructure. That's a dangerous combination for a company carrying $300B in ecosystem commitments.
Then there’s the risk of using chips as collateral. One critic compared GPU-backed debt to assets with “the shelf life of lettuce,” while the $500B Wall Street financing structure remains untested, with no capital raised under the new partnerships yet.
Michael Burry has also flagged the structure as a late-cycle warning, drawing parallels to credit practices that preceded past market dislocations.
Export controls add a slower, quieter threat. Chinese semiconductor firms Hua Hong Grace and SMIC posted profit surges of 385% and 267% respectively last quarter, fueled by domestic AI demand that US chipmakers can't touch.
The deeper damage, beyond lost revenue, is the erosion of customer relationships and the evolving market knowledge that comes from them.
The valuation gap
Here's where the investment case gets interesting. BofA's sum-of-parts analysis concludes Nvidia trades at a 34% to 50% discount to intrinsic value, even after pricing in financing risks. Arya carries a $350 price target against a stock sitting near $219.
Goldman Sachs pegs fair value at 14.4x its 2027 EPS estimate and calls the discount wide, though it wants to see disciplined capital allocation before re-rating.
Both banks flag the same lever: buybacks. Nvidia's 50% free cash flow return to shareholders sits well below peers, which return 75% to 100%, accelerating buybacks would be the fastest and most legible signal that the balance sheet bets are under control.
Nvidia's capital strategy has made it the financial backbone of the AI buildout, not just a hardware vendor. If that buildout stalls, the exposure is bilateral.
Nvidia will have bought itself a lock on the entire ecosystem that no competitor, however technically capable, can price-match or replicate.
