NVIDIA earnings put its AI chip supply chain to the test

Subscribe to our free newsletter today to keep up to date with the latest manufacturing news.

NVIDIA earnings for the third quarter of fiscal 2027 showed revenue of $115.8 billion, up 103% year over year and 20% on the previous quarter. Data Center revenue reached $107.6 billion, up 110%, and the company expects fourth-quarter revenue of $130 billion, plus or minus 3%.

The results, published Nov. 17 for the quarter ended Oct. 25, are the first since Jensen Huang resigned as president, chief executive, and board member on Nov. 12. Chief Financial Officer Colette Kress is serving as interim CEO while the board searches for a permanent successor, and former Synopsys chief executive Harvey C. Jones has become the board’s first chair. The filing also refers to an ongoing investigation by an independent committee of the board.

For manufacturers, the leadership change is only part of the picture.

The bigger question is whether the physical supply chain behind NVIDIA’s systems can keep pace with numbers of this size.

Vera Rubin is ramping faster than any NVIDIA product before it

NVIDIA said its Vera Rubin NVL72 system has shipped, and it described the launch as the fastest product ramp in the company’s history. Racks are running at CoreWeave, Google Cloud, Microsoft Azure, Oracle Cloud Infrastructure, and Nebius.

The ramp already shows up on the balance sheet. Inventory rose to $39.7 billion from $31.6 billion in the previous quarter, which NVIDIA attributed to supporting Rubin production. GAAP gross margin was 74.8%, slightly down on the prior quarter, a dip the company linked to the initial Rubin ramp. It expects gross margin of 71% to 72% in the fourth quarter.

That pattern will be familiar to anyone who has brought a new product line to volume. Early output brings new tooling, yield learning, and higher unit costs before a process settles.

Demand is also spreading beyond the largest cloud providers. Hyperscale customers accounted for $60.1 billion of Data Center revenue, up 71%. The category NVIDIA calls AI Clouds, Industrial, and Enterprise reached $47.5 billion, up 195%.

China remains a separate story. Data Center Hopper shipments to China were $2.4 billion in the quarter, but the fourth-quarter outlook assumes no Data Center compute revenue from the country.

Packaging and substrates have become the pressure points

The results release names 6 suppliers NVIDIA plans to work with to scale critical manufacturing capabilities for its infrastructure: Amkor Technology, IBIDEN, King Yuan Electronics, Samsung Electro-Mechanics, Siliconware Precision Industries, and Unimicron Technology Corporation.

None of these companies makes the processors themselves. Their work sits around the chip: assembling and testing packages, and supplying the substrates that connect large AI processors and their memory to the rest of a system. As accelerators grow larger and combine more silicon in a single package, those steps carry more of the manufacturing load.

NVIDIA did not give locations, timelines, or investment figures for these plans. It is too early to say how much new capacity they represent or where that capacity will be built.

Its procurement figures are clearer. Supply commitments rose to $390 billion from $279 billion, mainly because of multi-year memory purchasing. Of that total, $203 billion falls in fiscal 2028. The company said it has partnered with suppliers to secure the components it needs to meet demand for the next several years.

For component makers, that is a long and valuable order book. It also concentrates risk, since commitments on that scale assume demand holds.

The supply chain now runs without its founder

Leadership changes at this level raise practical questions for suppliers. Who signs off on new capacity, how quickly decisions move, and whether long-term priorities shift all matter to companies building plants around NVIDIA’s roadmap.

The board has said it expects operations, customer commitments, and the product roadmap to continue without interruption. In the results release, Kress said “our roadmap is on track,” and pointed to an explosive start for Vera Rubin.

Those figures support that message for now. Revenue grew 20% in a single quarter, and the outlook calls for further growth in a fourth quarter that includes an additional week.

What the filing cannot yet show is how the partnership plans with packaging and substrate suppliers will turn into installed capacity, or how quickly. Margins will also need watching as Rubin moves past its initial ramp.

Then there is the permanent CEO search, which will consider internal and external candidates. Until that appointment is made, suppliers committing capital to NVIDIA’s next generation of systems will be planning around a roadmap whose long-term owner is still to be named.

Source:
NVIDIA (SEC filing)

Ross Prudames

Ross is a Digital Marketing Executive specializing in B2B content, email marketing, and brand strategy. Alongside producing newsletters and digital campaigns, he writes news analysis and thought leadership for a portfolio of industry publications, creating content that helps professional audiences understand the trends and issues shaping their industries.