Cursor Simon Eskildsen on scaling Shopify, building turbopuffer, and the future of databases
4 snips
May 14, 2026 
Simon Eskildsen, CEO of turbopuffer and former long-time Shopify production engineer, shares war stories from scaling Shopify and building resilient infra. He talks about surviving flash sales, MySQL and caching tricks, and the origins of his Napkin Math thinking. Simon also explores why new database companies appear, turbopuffer’s path to extreme scale, and how AI tools change engineering and hiring.
AI Snips
Chapters
Transcript
Episode notes
Use Napkin Math To Validate Benchmarks
- Benchmarks can mislead; use napkin math to compute theoretical floors (e.g., data moved vs DRAM bandwidth) and reconcile gaps with observed benchmarks.
- Simon created Napkin Math to force first-principles checks before accepting benchmarks as evidence for design choices.
TCP Slow Start Caused Multi Second Page Loads
- TCP slow-start and conservative defaults can add seconds for distant clients; tune initial congestion window to reduce round trips for long-RTT connections.
- Simon diagnosed 3s page loads from Australia to US East as repeated TCP ramp-ups and improved latency by increasing initial packet window.
Databases Reappear Every 15 Years For New Workloads
- Big database waves recur ~every 15 years when new workloads appear and enable new storage architectures.
- Examples: 90s OLTP (Oracle/MySQL), mid-2010s OLAP (Snowflake/Databricks), and now AI-connected unstructured data driving new DBs like TurboPuffer.