The chipmaker's long-term supply and capacity commitments more than doubled in a quarter, testing the limits of its asset-light fabless model
Nvidia built an empire by designing chips and letting other companies worry about the factories. That arrangement now comes with a $279 billion footnote.
Ahead of its fiscal third-quarter 2027 results, the company’s latest 10-Q shows long-term supply and capacity agreements totaling $279 billion as of July 26, 2026. One quarter earlier, the figure stood at $119 billion.
What Nvidia is guiding for
The earnings release is expected around November 17 to 18, 2026. Nvidia has guided for Q3 FY27 revenue of $108 billion, plus or minus 2%.
Gross margin guidance sits at 74%, plus or minus 50 basis points.
Some projections circulating ahead of the print suggest Nvidia could clear the top of its own range. Those estimates land between $113 billion and $114 billion.
The expected drivers are familiar ones. Data Center demand remains the engine, with the anticipated Vera Rubin platform also expected to contribute.
The last quarter sets a high bar. In Q2 FY27, Nvidia reported revenue of $96.22 billion, up 106% year over year.
Adjusted earnings per share came in at $2.22, beating expectations by $0.13. Data Center revenue hit $89 billion, a 117% year-over-year jump.
Hitting the $108 billion midpoint would mean adding more than $11 billion in revenue in a single quarter.
Why the fabless model matters here
Nvidia’s fabless model has historically let the company pour resources into design and software while external partners shouldered the capital-heavy work of manufacturing.
A Seeking Alpha contributor who published a panel regression on Nvidia in March flagged just how lucrative this structure has been. The standout finding was a structural operating margin premium of about 39 percentage points. The author discloses a long position in NVDA.
The $279 billion in agreements is primarily focused on memory components. These are long-term obligations to secure supply and capacity, not one-off purchases.
Nvidia still does not own the factories. But it is increasingly locking in what those factories will produce for it, years in advance, with real dollars attached.
What this means for investors and the chip sector
The bull reading is straightforward. A company does not commit $279 billion to memory and capacity unless it expects to need it, so the filing doubles as a signal of how much demand Nvidia sees coming.
The bear reading is about flexibility. Supply commitments are obligations, and obligations do not shrink just because demand does.
If AI spending cools, Nvidia would still be on the hook for agreements signed during the boom. That creates questions around cash flow and balance sheet health that simply did not exist when the number was $119 billion.
That tension is what makes the gross margin guide so important. Holding margins near 74% while carrying these commitments would suggest Nvidia can absorb the extra obligations without denting profitability.
Key items to watch on the call include whether the supply commitment figure keeps climbing, how management characterizes the memory agreements, and how much Vera Rubin contributes to the quarter.
Revenue relative to the $108 billion guide will grab the headlines. The more telling detail may be whether a company that never needed factories is quietly becoming responsible for a meaningful share of their output.
