Founded by CEO and Co-Founder Ben Lerner, who brings valuable experience from Google’s world-class infrastructure team, Espresso AI has developed an ML-powered solution that automatically tunes and optimizes data warehouses to deliver significant cost savings. The company recently launched “Kubernetes for Snowflake,” a smart scheduling system that routes workloads between data warehouses to maximize efficiency, promising cost reductions of up to 70% while making enterprise-grade optimization accessible to organizations regardless of their internal engineering resources.
In this exclusive VMblog Q&A, Lerner explains how Espresso AI is democratizing the kind of intelligent infrastructure management that was previously only available to the world’s largest technology companies.
++
VMblog: For many folks reading this, it will be their first introduction to Espresso AI. Can you explain the offering at a high level and why you thought it was a necessary solution in the market?
Ben Lerner: Espresso uses ML to automatically tune and optimize your data warehouses. Our team saw a gap between how 99% of Snowflake users run their data warehouses and how large compute systems are managed at places like Google, and realized we could use ML to cross that gap and help Snowflake users save money.
VMblog: How did your experience at Google influence how you approach this problem and building out Espresso AI?
Lerner: Google’s infrastructure has one of the largest compute footprints in the world – there’s a lot to learn from how they’ve managed to scale efficiently. We’re applying learnings and best practices from those systems to data warehousing.
VMblog: Is this solution mostly for teams that are overwhelmed and maybe don’t have the necessary engineering resources to effectively re-architect and optimize their data warehouses? Or can anyone benefit from using Espresso AI?
Lerner: Anyone can benefit from Espresso AI – we expect almost anyone will save at least 30%. Our system makes optimizations that fundamentally rely on ML-based, real-time decisions, and there’s no way to replicate that without our infrastructure and our models.
VMblog: You just launched Kubernetes for Snowflake. Can you explain how it works to save Snowflake users money?
Lerner: Just like Kubernetes uses smart scheduling and routing to increase efficiency and drive down cost, our Snowflake scheduler optimally routes workloads between data warehouses to significantly increase efficiency – reducing costs by as much as 70%.
##
