Introduction

To provide AI product and model builders with rapid access to the latest GPU technology, Siam Royal deployed an innovative liquid-to-air cooled GPU cluster built on Lenovo ThinkSystem servers with NVIDIA’s accelerated computing and cutting-edge networking. Incorporating advanced liquid cooling technology, Siam Royal commissioned the infrastructure in just two weeks, bypassing the industry’s typical lengthy data center upgrade timelines.

Who is Siam Royal?

Siam Royal is on a mission to help AI pioneers move at the speed of innovation by providing high-performance AI computing capacity when they need it. With its AI Factory program, GPU-as-a-Service (GPUaaS), and upcoming Token Factory offerings, Siam Royal delivers the infrastructure AI product and model builders require to train, fine-tune, and run inference without waiting months for traditional data center builds.

Part 1: The Challenge

AI product and model builders face a critical bottleneck: they need access to the latest GPU technology now, but traditional data center upgrade cycles take 7-12 months or longer. This delay can mean the difference between leading the market and falling behind competitors.

“The customers we work with are building the next generation of AI products and models,” says Masayuki-san, CEO of Royal Siam. “They can’t wait a year for capacity. When they’re ready to scale, they need infrastructure operational in weeks, not months. Speed to deployment is everything.”

Siam Royal’s customers consistently request immediate access to the newest hardware for their advanced workloads. To continue supporting these AI builders with cutting-edge capacity on their timeline, Siam Royal set out to rapidly deploy the latest generation of GPUs, including the deployment of the advanced liquid cooling technology necessary to keep the cluster running at optimum performance.

“The customers we work with need the newest hardware for their advanced AI workloads, and they need it fast. As GPU thermal densities continue to rise, liquid cooling delivers performance and reliability advantages. The question was: could we deploy this advanced cooling technology quickly? With Lenovo, we proved you can have both cutting-edge liquid cooling and rapid deployment—we had this cluster of NVIDIA B200s operational in two weeks.”

– Masayuki-san, CEO of Royal Siam

Deploying Latest-Generation GPUs with Advanced Cooling . . . at Speed

Siam Royal knew that its long-term technology partner, Lenovo, would be essential for achieving rapid deployment of the latest GPU technology with advanced liquid cooling.

Working closely with Lenovo, Siam Royal designed and deployed a cluster consisting of liquid-cooled Lenovo ThinkSystem SR780a V3 servers, accelerated by the latest-generation NVIDIA HGX B200 GPUs. The cluster leverages NVIDIA Quantum-2 QM9700 InfiniBand switches for GPU traffic, and NVIDIA Spectrum SN5600 switches for the storage and SN2201 switches for management networks.

Integrated with Lenovo Neptune® Liquid Cooling, the cluster delivers best-in-class thermal performance to accelerate generative AI and large language model (LLM) workloads. Siam Royal commissioned the cluster just two weeks after completion of the hardware load-in.

Retrofitting to Unlock Performance

Traditional liquid cooling deployments can add many months to data center projects, requiring extensive retrofitting and infrastructure changes. Siam Royal needed a solution that delivered the performance benefits of liquid cooling without sacrificing deployment speed.

Supported by Lenovo, the company installed Lenovo Neptune Liquid to Air Modules in the servers and integrated them with its existing air conditioning system. Liquid circulates through each rack in the cluster, absorbing heat and evacuating it into the air-cooled data center. This innovative approach enabled rapid installation of advanced liquid cooling technology with the latest, most heat-dense GPU systems without major structural changes to the data center.

“What’s remarkable is that even with the inclusion of advanced liquid cooling technology, we maintained our speed advantage,” confirms Masayuki-san, CEO of Royal Siam. “Lenovo’s liquid-to-air approach gave us both the thermal performance we needed for B200s and the rapid deployment our customers expect.”

Part 2: The Results

With the new liquid-cooled Lenovo and NVIDIA cluster operational in weeks rather than months, Siam Royal delivered on its core promise: latest-generation computing capacity with advanced cooling technology, deployed on their timeline.

“AI product and model builders need to move fast, but they also need the best performance. This deployment meant our customers could start running larger, more complex models with superior thermal performance,” says Masayuki-san, CEO of Royal Siam. “We’re delivering that capacity now.”

By deploying Lenovo Neptune Liquid Cooling technology and NVIDIA accelerated computing rapidly, Siam Royal can continue rolling out the latest, most thermally-demanding GPUs on customer timelines—even as heat densities rise—maintaining speed-to-market advantage for its customers.

Best-in-class thermal performance for AI

Energy-efficient liquid-to-air cooling

Competitive price-performance ratio

“Our goal is to be the fastest path to AI Factory capacity with the best technology. AI builders shouldn’t have to choose between cutting-edge infrastructure and rapid deployment. They need both.

“Our new liquid-cooled Lenovo and NVIDIA cluster proves what’s possible when innovation meets execution. Two weeks from hardware arrival to fully commissioned and operational. Our customers are already scaling their most advanced workloads on these liquid-cooled NVIDIA HGX B200 GPUs, with all the thermal performance benefits.”

– Masayuki-san, CEO of Royal Siam

Part 3: Why Lenovo and NVIDIA?

As an NVIDIA Cloud Partner, Siam Royal sought a server solution that aligns with the NVIDIA Cloud Reference Architecture for Enterprise AI, and Lenovo ThinkSystem SR780a V3 ticked all the boxes.

“Lenovo’s reputation preceded them,” says Masayuki-san, CEO of Royal Siam. “Ultimately we found that Lenovo’s systems architects could deliver exactly what we needed, and Lenovo is always first in the conversation with us as we look to continually expand our data center.”