Morgan Stanley: NVIDIA’s 2U Air-Cooled MGX GB200 NVL2 Still Suffers From “Thermal Issues”

Oct 22, 2024 at 11:31am EDT
NVIDIA
This is not investment advice. The author has no position in any of the stocks mentioned. Wccftech.com has a disclosure and ethics policy.

Morgan Stanley has been quite loquacious on NVIDIA's Blackwell-related production hiccups, having recently penned an entire note on the minor irritants that collectively pushed back the GPU manufacturer's planned production ramp-up for its latest Blackwell GPUs. Now, the Wall Street titan is out with another investment note, this time leveraging its channel checks to expound on NVIDIA's ongoing "technical challenges" vis-a-vis the GB200 NVL-based servers.

As a refresher, while announcing its earnings for the June-ending quarter, NVIDIA confirmed the rumors of a minor Blackwell design flaw, but noted that the shortcomings were rectified by making a few minor changes to the photomask - a specific template that is used to create bespoke patterns on semiconductor wafers. While adding some much-needed color to this brief statement, Morgan Stanley recently noted that NVIDIA only discovered the low-yields issue in the post-packaging stage. The ensuing scrapping of the subpar units then resulted in a loss of "CoWoS and HBM3e," which were already in short supply and served to further aggravate NVIDIA's supply problems.

Related Story MSI Chairman Warns Memory & GPU Shortages Will Drag Into 2026, But Says CPU Supply Will Get Better By Q3

Now, Morgan Stanley has leveraged its supply chain checks to suggest that the 2U air-cooled form factor of NVIDIA's MGX GB200 NVL2, which houses "2x Grace and 2x B200 Blackwell GPUs on the same PCB board, with the GPU module connecting to the main PCB board using an SXM7 module," continues to suffer from "thermal issues."

For the benefit of those who might not be aware, a U is a standard unit - equivalent to 1.75 inches -for measuring server rack heights. Accordingly, a 2U chassis is 3.5 inches high, while a 4U chassis is 7 inches high.

Coming back, Morgan Stanley notes:

"All of the servers showcased at OCP were based on a 2U air-cooled form factor."

The Wall Street behemoth then suggests:

"Our conversations with supply chain partners indicated to us that there are still some thermal issues with the 2U form factor, so this may potentially end up being in a 4U form factor instead."

Elsewhere, Morgan Stanley notes that in light of the ongoing uncertainty around the demand outlook for its MI325 chip, AMD has "trimmed some CoWoS wafer booking at TSMC for 2025." Critically, this freed up capacity was immediately taken up by NVIDIA.

Also, Morgan Stanley believes that Marvell's Trainium 2 chip can see 400,000 units produced in 2025:

"Marvell’s CoWoS booking at TSMC is 3x higher in 2025 vs. 2024. If the output for Trainium 2 is 200k units in 2024, based on CoWoS capacity booking, we believe there could be 400k units of Trainium 2 in 2025 and 500k-600k units of Inferentia 2.5 in 2025."

About the author: Writing is my one incontrovertible passion. Over the past six years, he has authored over 2,200 distinct articles on financial and tech-related topics, spanning nearly 1 million words. And he has been a member of Wcctech mobile team since 2025. As an alumnus of the University of Toronto, Rotman Commerce Program, I bring nuance, in-depth knowledge, and a unique perspective to every topic that I cover. When I'm not writing, I'm traveling the world, exploring hidden confectionaries and restaurants as an aspiring food connoisseur.

Follow Wccftech on Google to get more of our news coverage in your feeds.

Deal of the Day