Read the full transcript
My name is Kevin Matcham. I’m the COO and co-founder of Bentaus. We’re an energy-first NeoCloud.
Some of the biggest challenges we were facing is access to power and access to DC space. Bentaus chose an open reference architecture because our thesis has always been a very agnostic approach to AI infrastructure.
We chose AMD Instinct series GPUs because of the performance and the technical engineering teams behind those GPUs. We then got introduced to DriveNets, who we realized was really setting the benchmark for performance from a networking perspective on AMD reference architecture. We were able to reduce our watts per token and provide some of the more highly performant clusters to our customers.
Working with DriveNets allows us to have some of the most highly performant networking solutions when it comes to AMD Instinct series, allowing us to scale up and scale out across our data center portfolio.
We tie this all together with our Ziani energy orchestration software, which allows us to curtail GPUs based on real-time grid signals. This allows us to run at 30% less power with only a 3% degradation in token throughput.
We have a lot of confidence in continuing to work with DriveNets because of their technical capabilities, their ability to analyze the trends on where the industry is going.
We like working with groups that are open-minded and flexible and highly competent. And we believe that DriveNets checks all those boxes, and we look forward to a continued partnership, ultimately solidifying Bentaus as one of the top NeoClouds in the industry.
Want to learn how DriveNets is reshaping AI infrastructure?
Explore DriveNets AI Fabric full stack networking solutions for AI Infrastructure
