In this episode, Host Taylor Baker sits down with Srikanth Satya, co-founder and CEO of QBeast, to discuss why compute not storage has become the biggest cost problem in modern data stacks, and how smarter data layout and indexing can dramatically reduce the price of analytics and AI workloads.
What You’ll Learn
- Why object storage and open table formats have become the default foundation for modern “lakehouse” data architectures and what that shift makes harder.
- How compute costs quietly become the primary operating expense once data is centralized and teams start running queries and ML workflows at scale.
- The core idea behind reducing compute spend: read and scan less data to answer the same questions.
- How multidimensional indexing and clustering can outperform coarse partitioning when teams query across multiple attributes (not just dates).
- Why lowering cost and friction matters: infrastructure solutions win when they fit into existing platforms and workflows without requiring rewrites.
- How AI can be used to automatically recommend the best columns to index based on a company’s real tables and query patterns.
- Where approximate analytics can be more valuable than exact answers, especially as data volumes grow and speed matters.
- Founder lessons from building deep tech: don’t over-rotate on perfect engineering, ship earlier, get customer feedback faster, and iterate quickly.
Across the conversation, Srikanth makes a practical case that as data volumes and AI usage accelerate, the winners will be the teams that make computation more efficient without forcing organizations to “boil the ocean” and replace their existing systems. The episode reinforces a consistent message for builders: focus on measurable customer value like cost savings deliver it in a way that’s easy to adopt, and let real-world usage guide the product forward.
To learn more about Srikanth Satya and their work, visit QBeast.io