With AI commoditizing code creation, the sustainable value for software companies shifts. Customers pay for reliability, support, compliance, and security patches—the 'never ending maintenance commitment'—which becomes the key differentiator when anyone can build an initial app quickly.
Snowflake CEO Sridhar Ramaswamy observes that while a few AI labs are far ahead, the pace of innovation means any competitive advantage is fleeting. A year-long lead is now considered an eternity, suggesting constant pressure and rapid shifts in the market.
Snowflake's CEO warns that traditional software firms with walled-garden data models are vulnerable. If they don't develop their own compelling agentic interfaces, they risk being reduced to mere data sources for dominant AI platforms, losing their customer relationship and pricing power.
Sridhar Ramaswamy suggests software valuation multiples are contracting because investors see through the strategy of just adding an 'AI SKU.' The market believes this approach won't win, favoring integrated, consumption-based models where customers only pay for demonstrated value from AI.
The closed nature of leading US AI models has created an information vacuum. Sridhar Ramaswamy notes that academia is now diverging from US industry and instead building upon published work from Chinese companies, which poses a long-term risk to the American innovation ecosystem.
To overcome the sentiment that AI is just hype, Snowflake's CEO advocates for building and using internal AI agents daily. He personally uses a sales agent on his phone in executive meetings, demonstrating its practical value which drives both internal adoption and external credibility.
Former Google SVP Sridhar Ramaswamy reveals that Google has a history of mobilizing intensely against threats, using all-hands-on-deck initiatives. Its recent AI surge isn't surprising to insiders who know its ability to activate a 'war' footing when challenged.
Snowflake's support team transitioned from just using software to actively building solutions with a coding agent. When existing tools fall short, they create new ones and add them to a shared suite. This self-improving system dramatically reduces resolution times for complex cases.
The core conflict in AI is over who owns the user interface. Model makers like OpenAI aim for a universal 'big brain' agent that consumes data, while data platforms like Snowflake are building specialized agents on top of their proprietary data to avoid becoming commoditized data pipes.