Lambda, a GPU cloud provider, has successfully deployed its Nvidia GB200 NVL72 rack system at EdgeCloudLink’s (ECL) hydrogen-powered data center in Mountain View, California. This milestone marks another step in advancing sustainable and high-performance AI compute capabilities.
Key Highlights of the Deployment
Hydrogen-Powered Data Center Deployment
In a LinkedIn post, Yuval Bachar, ECL’s founder and CEO, shared:
“Lambda, Supermicro & ECL just powered the first Supermicro NVL72 GB200 based system at the ECL-MV1 site running at 110kw+ based 100 percent on Hydrogen power with Zero emission. The site is running now 3.6 million GPU cores!!! in two rack spaces!!!! Unbelievable.”
The deployment showcases the potential of hydrogen energy in supporting high-density AI workloads.
Specifications of the Nvidia GB200 NVL72 Rack
The GB200 NVL72 liquid-cooled rack features 72 GB200 GPUs, 36 Grace CPUs, and nine NVLink switch trays with dual NVLink switches in each tray. This configuration functions as a unified GPU, significantly enhancing AI performance. Notably, the rack design was revised in late 2024 to address overheating issues, ensuring improved reliability and efficiency.
ECL’s Modular Data Center Design
Rapid Installation and High Flexibility
According to Bachar, the installation process for the GB200 NVL72 rack took only one hour. This speed is attributed to ECL’s modular design, which simplifies modifications. For example, the ECL-MV1 data center—launched in May 2024 with 1MW of IT capacity—can easily transition from 75kW per rack to 150kW as needed.
Accelerating Innovation Cycles
Bachar highlighted the importance of ECL’s adaptable design in meeting the demands of rapidly evolving technologies. He stated, “Today’s cycle of innovation with Nvidia is that every nine to 12 months we see a new technology that comes with different requirements and densities.”
Hydrogen-Powered Operations at ECL-MV1
The ECL-MV1 data center operates entirely on hydrogen power, combining grey, blue, and green hydrogen sources to achieve zero emissions. This sustainable energy solution aligns with the broader industry shift toward greener data center operations.
Lambda’s Strategic Plans and Future Deployments
Anchor Tenant at ECL’s TerraSite-TX1 Campus
Lambda is set to become the anchor tenant at ECL’s upcoming TerraSite-TX1 data center campus in Houston, Texas. This 1GW facility will also utilize hydrogen power, demonstrating a continued commitment to sustainability.
Additional Deployments
Lambda recently delivered another GB200 NVL72 rack to Pegatron SVR, which will similarly bolster its AI compute platform. These deployments reinforce Lambda’s strategy to expand its GPU-based cloud compute offerings.
About Lambda
Founded in 2012, Lambda provides GPU-based cloud compute services at competitive rates. The company also offers colocation space and sells GPU desktops. In mid-2024, Lambda sought $800 million in funding to expand its infrastructure with additional Nvidia GPUs and supporting systems.
FAQ Section
What is the Nvidia GB200 NVL72 rack?
The Nvidia GB200 NVL72 rack is a liquid-cooled server rack featuring 72 GPUs, 36 Grace CPUs, and advanced NVLink connectivity. It enables high-performance AI workloads by functioning as a unified GPU system.
Why is hydrogen power significant for data centers?
Hydrogen power is a sustainable energy source that produces zero emissions, making it an ideal solution for reducing the carbon footprint of high-density data centers.
What makes ECL’s data center design unique?
ECL’s modular design allows for rapid deployment and flexible scaling, enabling faster adaptation to technological advancements and changing workload demands.
What is Lambda’s role at ECL’s data centers?
Lambda deploys high-performance GPU racks at ECL’s facilities and is set to be the anchor tenant at the upcoming TerraSite-TX1 campus.
How does the GB200 NVL72 rack improve AI compute?
With its unified GPU architecture and liquid cooling, the GB200 NVL72 rack significantly boosts performance and reliability for AI workloads.
