Nvidia is hunting for a materials engineer to wrangle the thermal challenges of its next-generation networking hardware.

A job listing sees the chip giant advertising for a senior materials engineer to join its Networking Thermal Engineering team, with the remit to look into developing cooling solutions for high-speed communication systems.

The role shows just how far increasing rack densities are pushing the heat problem beyond compute. Nvidia's upcoming Vera Rubin rack packs 1.3 million individual components laced together by ConnectX-9 SuperNICs and BlueField-4 DPUs.

But where liquid cooling used to be a problem reserved for its flagship graphic processing units (GPUs), it also looks to be a networking silicon challenge. The proliferation of 800 Gb/s (800G) and even some 1.6 Tb/s (1.6T) deployments at the hyperscale level have seen switching application-specific integrated circuits (ASICs) and network interface cards (NICs) dissipating heat at levels that mirrored that of compute hardware just a few years prior.

Achieving ultra-dense 3.2T interconnectivity is the pinnacle being pursued by hardware vendors, including Nvidia, which brings the material question to the fore.

Nvidia’s listing reflects that urgency, with successful candidates expected to define thermal material selection strategy, develop characterization and qualification methodologies, and lead root cause analysis on material failures, with particular emphasis on thermal interface materials (TIM) and hands-on experience with liquid cooling environments.

Crucially, the listing isn’t framed as a research role, with Nvidia wanting someone who can take solutions from concept to mass production, working closely with design, analysis, and thermal architecture teams.

The listing was made public a few weeks after industry reports that Nvidia was struggling with power consumption issues with its next-generation Rubin series.

Trendforce reported in early April that, along with memory supply constraints, Nvidia customers may face some shipment delays as it's reportedly facing power issues relating to its new ConnectX line.

The Vera Rubin generation sees the chip giant transition from ConnectX-8 (CX8) to ConnectX-9 (CX9) to provide 800G GPU-to-GPU connectivity. But Trendforce suggested it was addressing issues shifting from CX8 to CX9 owing to “significantly higher power consumption and optimizing performance under more advanced liquid cooling solutions.”

Networking has steadily become a cornerstone of Nvidia’s business, with the segment accounting for $11 billion for its fourth fiscal quarter ended January 25, and CFO Colette Kress labeling the unit a “standout” during that quarter.

The rise hasn’t gone unnoticed at the top level, with both CEO Jensen Huang and networking chief Gilad Shainer proclaiming Nvidia is now the largest networking company in the world.

It’s all part of the firm’s wider strategy to break its perceived image of being a “chip company,” with it instead positioning itself as a full-stack platform player.

“We'll work and integrate Nvidia's technology into whatever platform you would like us to integrate into,” Huang said during a GTC press briefing attended by SDxCentral. “We offer you the software, we offer you libraries, we integrate with your technology so that we can bring accelerated computing to everybody in the world.”