Databricks CEO 回击 Snowflake：你们没在做 AI，不如来买我们的 AI token
‘You’re Not Doing This AI Stuff,’ Databricks CEO Tells Snowflake
Databricks CEO Ali Ghodsi 在 X 上回应 Snowflake CEO Sridhar Ramaswamy 的质疑，称 Snowflake“没在做 AI 这件事”，建议其通过 Databricks 获取 OpenAI、Anthropic、Google 和 xAI 等模型 token，并推介提供模型路由、预算管理与访问控制的 Unity Gateway。Ghodsi 称多数客户近几个月 token 用量上升的同时 token 支出减少约 30%。Ramaswamy 此前要求 Databricks 披露剔除 pass-through 推理收入后的增长、自由现金流与利润率等数据，并回应称“我问的是收入，不是来贡献收入的”。
主题
正文
Databricks CEO Ali Ghodsi took a jab at rival Snowflake over its AI strategy, telling the company to consider buying AI tokens through Databricks after Snowflake CEO Sridhar Ramaswamy questioned the data software company’s revenue growth.
“I don’t think we should fight. The two companies are complementary,” Ghodsi wrote on X. “Since you’re not doing this AI stuff, I really think Snowflake should consider getting its AI tokens through Databricks.”
Ghodsi pitched Databricks’ platform as a way for Snowflake to access open-source and frontier AI models, including those from OpenAI, Anthropic, Google and xAI. He also promoted Unity Gateway, which offers model routing, budget management and access controls.
“Most of our customers used this to reduce token spend ~30% last few months while token count went up,” Ghodsi wrote, referring to customers using the platform. He ended the post by asking, “Quick POC?”
“Generous. I was asking about the revenue, not contributing to it,” Ramaswamy responded.
The exchange began after Ramaswamy called on Databricks to disclose more details about its financial performance. Quoting a post by venture capitalist Ben Horowitz titled Databricks & the Future of the Firm, Ramaswamy wrote, “Release all the numbers.”
He questioned whether Databricks’ reported 80% year-on-year revenue growth would fall to 65% after excluding pass-through inference revenue. He also asked how much growth came from AI-native customers and how much reflected the transition from one-time software licences to service-based revenue.
Ramaswamy sought further details on free cash flow, remaining performance obligations (RPO), current remaining performance obligations (cRPO), GAAP and non-GAAP operating margins, and share dilution.
Databricks announced in August that its annualised revenue run-rate had crossed $7 billion, with revenue growing more than 80% year-on-year in the second quarter. It also raised $5 billion in a funding round at a $190 billion valuation and reported positive adjusted cash flow over the preceding 12 months.
The company said its Lakehouse data warehousing business had crossed a $1.5 billion revenue run-rate, while its Lakebase database product had exceeded $100 million.
Databricks has been expanding its AI offerings as enterprises look to deploy AI models and agents using their corporate data. Its Unity AI Gateway provides tools to manage access to models and control AI spending.
Snowflake, meanwhile, offers AI capabilities through its Cortex platform, which allows customers to work with AI models and build applications using their data.