SMC Offers "Immersion Cooling" In NVIDIA's Blackwell AI Servers, Scaling Down Power Consumption By 50% 1

Revolutionary Immersion Cooling Technology in SMC’s NVIDIA Blackwell AI Servers Halves Power Consumption

Sustainable Metal Cloud (SMC) has introduced an innovative cooling solution for large-scale AI clusters that involves immersion cooling for systems such as NVIDIA’s Blackwell AI servers.

**Enhanced Data Center Performance with SMC’s Immersion Cooling**

As the computing demands of the industry escalate, data centers are increasingly outfitted with high-performance components. This trend has led to the need for effective and cost-efficient cooling solutions. While NVIDIA has primarily used liquid-cooling in its AI servers, SMC’s new approach features “cubic structures” that allow servers to be submerged in a synthetic oil, significantly improving cooling efficiency.

SMC’s technology employs HyperCubes, which use polyalphaolefin to dissipate heat through immersion. Although immersion cooling has previously faced challenges, SMC claims that HyperCubes can reduce power consumption by up to 50% and lower costs by 28% compared to traditional methods.

“Our HyperCubes enable high-density GPU hosting, which is critical for platforms like NVIDIA’s Grace Blackwell,” said Tim Rosenfield, co-Founder and co-CEO of SMC.

**Universal Deployment and Industry Potential**

SMC asserts that HyperCubes can be universally deployed in data centers worldwide. The company has already partnered with NVIDIA and is seeking to expand its market reach. To achieve this, SMC aims to raise $400 million in equity and $550 million in debt funding for global expansion.

**Increasing Market Demand for Liquid Cooling**

The move towards liquid cooling in data centers has gained significant traction, especially as next-generation AI servers rely heavily on such mechanisms. The rising demand for liquid cooling components has created supply challenges for manufacturers like SMCI and Foxconn, underscoring the industry’s interest in this technology.

While SMC’s HyperCubes hold promise, widespread adoption will depend on overcoming manufacturing complexities and other barriers.

Source: CNBC