Enquire nowLog in

17 March 2024


Director of Product & AI: Dr. Peter Blain on the Race for Compute

With his PhD in Artificial Intelligence & Computer Systems Engineering and diverse experiences across commercial, research, and financial domains, Peter is our technical lead and foremost authority in all things AI, DevOps, Machine Learning & Supercomputing.

He’s a busy man to get a hold of, but we caught up in between deployments to talk about SMC, the sustainable AI opportunity, and how current AI infrastructure is holding up.

1. What’s your role at SMC?

I joined our parent company Firmus three and a half years ago to help define and build out the AI cloud services that leverage our unique approach to immersion cooling. My role is to make SMC's energy-efficient GPU compute usable by data scientists, machine learning engineers and others in the AI field. My team builds the cloud orchestration layer and MLOps tools that our customers use to interact with the underlying compute.

When I joined it was before AI was everywhere, but we could tell GPU cloud computing was the next big technology advancement. We saw workloads running in traditional air-cooled high-performance compute and cloud computing data centres consuming excessive amounts of energy, and knew we could eliminate this problem with our advanced immersion cooling solution.

2. As GPU requirements scale, how can data centres and businesses prepare themselves?

All businesses will need to adopt AI or expand their existing AI capabilities, and they will need to ensure that they have access to the necessary compute resources to support this. Large language models, in particular, require large amounts of GPUs coupled with high-speed interconnects and fast storage arrays.

The release of chips that begin to challenge existing rack densities has become an important consideration. If you’re a business looking to scale, strategic infrastructure planning is essential. Is it future-proof?

“ Put simply, the world is at a pivotal moment and businesses that fail to secure cost-effective access to suitable AI infrastructure, may not be in business for much longer. ”

3. What do you anticipate will have the most significant impact on cloud computing in the next 5 years?

Looking ahead to the next five years, the landscape of cloud computing will be driven by an escalating arms race for accelerated compute. As businesses increasingly view AI capabilities as critical for survival, the demand for more powerful GPUs intensifies. With each new generation of GPUs pushing the boundaries of performance, they consequently run hotter.

At SMC, we're not just ‘anticipating’ these changes. We’re hoping to be a major influence on the industry — pushing for global adoption of accelerated, sustainable compute. SMC is scaling out its immersion cooling solution across the globe to ensure that energy-efficient GPU compute is available to our customers. We're working with NVIDIA to bring the latest software ecosystem into our product, which includes NVIDIA AI Enterprise and integration with NGC. The software layer allows SMC customers to get up and running quickly and easily with the latest AI models and MLOps tools.

This software ecosystem will run counter to the ‘arms race’ and instead will democratize access to accelerated compute.

4. Can you discuss the importance of the industry's move towards sustainable cloud solutions?

The shift towards energy efficiency in HPC and GPU-centric cloud computing is a key component of the broader transition towards sustainable development, demanding collaborative efforts from governments, industries, and communities worldwide.

It’s why SMC's drive towards sustainable GPU cloud services is not just an environmental imperative but also a strategic business decision.

“ It offers a path towards sustainable growth, innovation, and leadership in the digital age, aligning with global efforts to combat climate change and transition to a low-carbon economy. ”