Enterprise Data Platforms Are Becoming AI Platforms — What That Means for Your Data Strategy

Snowflake's fiscal Q2 revenue jumped 35% year-over-year, and the company specifically credited its CoCo AI coding agent — now at 9,100 accounts, up over 2,000 in the quarter — as a real contributor, not a forward-looking pilot metric. That's a genuinely useful data point for anyone evaluating a data platform partly on its AI capabilities: a vendor reporting specific adoption numbers alongside earnings is giving you something far more concrete than a roadmap slide, and it's worth using that same bar when you evaluate any platform's AI claims, including your current one.

The Shift From Add-On AI to Native AI

There's a meaningful architectural difference between a data platform that lets you connect an external AI tool and one that integrates AI capability natively into the platform itself — the way Snowflake has built Cortex AI directly into its data cloud, including integrating OpenAI's models within it rather than requiring a separate tooling layer. Native integration generally means lower latency, less data movement (a real security and cost consideration), and a simpler operational model — but it also means your AI capability is now more tightly coupled to your data platform vendor's own roadmap and pricing decisions, a real trade-off worth naming explicitly rather than treating native integration as an unambiguous win.

Questions to Ask Any Data Platform About Its AI Roadmap

• What's the actual adoption number for your AI features among existing customers — not the number of customers with access, but the number actively using it, the same distinction Snowflake's reported 9,100 CoCo accounts makes explicit
• Is the AI capability built natively into the platform, or does it require exporting data to a separate service — and what does that mean for your data governance and residency requirements specifically
• Can the vendor show quarter-over-quarter adoption growth for a specific AI feature, not just an aggregate "AI revenue" number that could be counting anything
• What's the pricing model for AI features specifically, and how does it scale — a platform's AI capability being genuinely valuable doesn't guarantee its pricing model won't become a cost problem at your actual usage volume

Reading Vendor Earnings Calls as Due Diligence

Public data-platform vendors' quarterly earnings calls are an underused due-diligence resource for enterprises evaluating them — a vendor's own reported adoption metrics, guidance changes, and specific product callouts (like Snowflake naming CoCo specifically rather than speaking only in aggregate AI-revenue terms) are more reliable signals than sales-conversation claims, precisely because they're made under securities-disclosure obligations, not sales incentives. Treating a prospective vendor's last two quarters of earnings commentary as required reading before a renewal or RFP decision costs an hour and surfaces real signal a sales call won't volunteer unprompted.

What This Means for Your Own Data Strategy, Not Just Vendor Selection

Beyond vendor evaluation, the broader signal — data platforms converting native AI integration into real, measurable revenue growth — is a leading indicator that AI capability is becoming a baseline expectation for enterprise data infrastructure, not a differentiator reserved for cutting-edge adopters. Organizations still treating their data platform's AI features as optional or experimental are increasingly out of step with where the rest of the market has already moved, and the cost of catching up later only grows as the gap widens.