Snowflake's ongoing pitch: bring AI to data rather than data to AI
Customers are 'excited' says one solution provider
Snowflake is putting cash and kinetic energy behind the idea that AI works best in its platform.
Whether it's the $200 million deal signed with OpenAI, its impdending acquisition of Observe, or its new Project SnowWork, Snowflake partner Gary McConnell said that the platform is constantly coming up in conversations, because of this effort.
“What's compelling about Snowflake's recent moves isn't just the dollar amounts — it's the consistency,” he told The Register. “Snowflake has been aggressive on the feature roadmap. They're also making investments in observability which should play to enterprise support as complexity scales.”
McConnell, who's the CEO of solution provider VirtuIT, said Snowflake’s recent moves have put a focus on helping customers achieve an actual return on investment for AI, an important topic among his enterprise customers.
“The OpenAI partnership, the Observe acquisition intent, and Project SnowWork all point to the same thesis: your data platform should be the place where AI work actually happens, not just a source you export from,” he told The Register. “For our customers, that's a meaningful shift. Historically, organizations had to stitch together a data warehouse, a feature store, and a separate AI/ML environment. Snowflake is collapsing that stack, and we're seeing real interest in that consolidation story.”
Snowflake has been aggressive with its feature roadmap and making investments in observability a critical component as the complexity of AI data scales, McConnell said.
“Customers are excited about being able to bring AI workloads to the data rather than moving the data to the AI,” he said. “The governance story of knowing where your data is and who touched it also resonates strongly in regulated industries such as pharma, legal, and finance to name a few.”
Snowflake is adding thousands of customers annually, growing from 7,800 in January 2023 to 13,330 this January, a 70 percent increase in its customer base in three years.
In that same timeframe, it has also added more enterprise customers, growing those from 573 of the Forbes Global 2000 in 2023 to 790 as of January 2026. Those enterprises contributed 43 percent of the company’s $4.7 billion in revenue during the most recent fiscal year.
Snowflake kicked off the year by announcing a partnership with Google that brought the Chocolate Factory's Gemini model into Snowflake’s Cortex AI, its application-to-inference service. It also announced plans to buy Observe AI which engineers can use to detect anomalies, identify root causes faster, and improve operational resilience.
In February it announced a $200 million partnership with OpenAI to develop custom AI solutions for enterprise customers. It added Semantic View Autopilot, a service that gives AI agents a shared set of business metrics for more consistent and reliable data outcomes. Then came Snowflake Postgres which is powered by pg_lake, a set of open source PostgreSQL extensions that allow Postgres to work within an organization’s data lakehouse.
Last week the company announced that it was beta testing Project SnowWork, which uses role-based AI personas that understand common business workflows, terminology, and KPIs. The idea is to give business tasks to the business persona that it matches, with Snowflake providing pre-configured capabilities for finance, sales, marketing, operations.
“We’re not assuming every sales or marketing team works the same way, but there are clear patterns in how these functions operate—how pipeline is tracked, how campaigns are measured, how forecasts are built,” Snowflake's Bala Kasiviswanathan, VP of Developer and AI Experiences, told The Register. “Those patterns, observed across thousands of customers, give us a strong starting point.”
- Snowflake gets frosty with Google Gemini
- Snowflake plugs PostgreSQL into its AI Data Cloud
- Snowflake update caused a blizzard of failures worldwide
- Snowflake bets $200M that OpenAI makes databases more chatty
He said that Snowflake is using Project SnowWork internally with its sales teams, which can now generate data-backed QBRs, pitch decks, and customer emails all from one place. Executives get a personalized intelligence feed with the metrics that matter to them, tailored to their role. And Snowflake said it has begun to automate its earnings prep using SnowWork to ease the burden of a weeks-long, cross-team effort.
“The system is grounded in each customer’s own data, definitions, and workflows, and teams can layer in their own logic,” Kasiviswanathan said. “Over time, it also improves through usage and feedback. So it’s less a fixed “persona” and more a starting point that quickly becomes specific to how each company actually runs. This is also a key part of what we are trying to learn and codify during our research preview.”
In terms of security, he said that every action Project SnowWork takes inherits role-based access controls, data policies, and audit logging automatically. That means it can only act on data the user is allowed to see, and every step is fully traceable, he said.
“Enterprises can inspect the steps, validate outputs, and maintain control over how and when actions are executed,” he said. ®
