Snowflake's CRO argues that while large enterprise deals are attractive, a business built solely on them is fragile. He championed a parallel high-velocity motion focused on acquiring new logos of all sizes, creating a more predictable and ultimately larger market over the long term.

To navigate the unpredictable AI landscape, Snowflake's CEO dismantled its specialized, multi-layered structure that had slowed down iteration. This shift prioritized accountability and shorter engineer-to-customer feedback loops, recognizing that speed and adaptability now trump carefully laid out strategies.

ClickHouse CEO Aaron Katz reveals that their database is gaining enterprise customers through an unconventional channel: AI recommendations. He notes that Anthropic itself became a customer after its own model, Claude, suggested using ClickHouse for observability, demonstrating how LLMs are now influencing major technology purchasing decisions within large companies.

Snowflake's initial high-velocity sales model hit a wall with large enterprises. New CEO Frank Slootman mandated a change, forcing CRO Chris Degnan to "rip the bandaid off" and restructure the entire GTM organization in the middle of a fiscal year to create a dedicated enterprise sales motion.

Snowflake hired its first salesperson pre-revenue not to sell, but to get the product into customers' hands to break it. This person acted as a de facto product manager, gathering critical feedback that led to a core architectural change, proving the value of a GTM hire before product-market fit.

Snowflake invested seven months of its entire engineering team's effort to solve a specific clustering problem for one customer, Localytics. This seemingly costly detour created a core feature that became the key to winning major enterprise accounts like Nielsen, proving that bending for the right customer can redefine the product.

Databricks and Snowflake took opposite approaches. Snowflake optimized for fast queries on curated, proprietary "downstream" data. Databricks focused on large-scale, messy "upstream" data ingestion using open formats. Databricks found it easier to add speed than it was for Snowflake to move upstream and abandon its proprietary lock-in.

In Snowflake's consumption model, a salesperson's job isn't done at signing. They have separate quotas for bookings (the commitment) and consumption (actual usage). This structure forces them to act as a long-term business partner, ensuring the customer successfully adopts and uses the platform.

ClickHouse's revenue ramp (0, 12, 50, 200, >500M) is faster than any database before it, yet more gradual than today's AI apps. This highlights that infrastructure adoption, while explosive, builds on durability and high switching costs, not just viral growth.