Snowflake

Product revenue growth has been decelerating pretty steadily — from 70%+ down to the mid-20s — and management keeps guiding conservatively and then barely beating. At 15x forward revenue this thing is still priced like a hypergrowth company, not a maturing data warehouse vendor competing against free tiers from AWS and Google.

How do people think about Snowflake's competitive positioning specifically in the financial services vertical versus Bloomberg's data infrastructure and cloud-native alternatives like Databricks' financial services cloud? I work adjacent to a few big banks and the conversations I'm hearing are more fragmented than I'd expect for a supposed category leader.

Frank Slootman's departure was underestimated as a negative signal. He was the credibility anchor with large enterprise buyers and the person who drove the hyper-efficient sales culture. Ramaswamy is a product guy from Google who's never run a sales-heavy enterprise company at this scale, and that transition risk isn't priced in.

Everyone is sleeping on Snowflake's enterprise sales motion — they have 510+ customers spending over $1M annually and that number keeps growing even as the macro is tough. These are multi-year, deeply embedded workloads that don't churn easily. The comparison to a SaaS company with weak retention is wrong; migration costs out of Snowflake are enormous.

Does anyone have a clear view on how Snowflake's Marketplace is actually tracking? They talk about it a lot on earnings calls as a monetization driver but I've never seen a clean breakout of revenue or even active listings growth. Is this a real business or just a talking point?

Snowflake announced expanded support for Apache Iceberg open table format, allowing customers to use Snowflake as a compute engine over data they own in their own S3 buckets without full lock-in. This is a direct response to competitive pressure from Databricks and the broader open lakehouse movement.

Snowflake reported its most recent quarter with product revenue of approximately $829 million, up 29% year over year, and raised full-year product revenue guidance to around $3.43 billion. Remaining performance obligations came in above $5 billion, suggesting the forward pipeline remains healthy despite macro headwinds.

Snowflake's Cortex AI features are genuinely differentiated — the ability to run LLMs directly inside the data cloud without moving data out is a real architectural advantage over having to pipe everything through a separate AI stack. Enterprises hate data egress costs and compliance headaches, so keeping inference inside the warehouse is a legitimate moat. Sridhar Ramaswamy's focus on making this a native experience rather than bolted-on is the right call.

Databricks just raised another massive round at a valuation that implies it's closing the gap fast, and it's winning deals in the data engineering and ML pipeline space that Snowflake used to get by default. The narrative that Snowflake owns analytics and Databricks owns ML is breaking down as both companies move toward a unified lakehouse.

Something that doesn't get discussed enough is Snowflake's global expansion — international markets are still a relatively small percentage of revenue compared to US, and there's real ceiling-raising potential if they can crack European enterprise at scale, especially with new EU sovereign cloud regions addressing data residency concerns.

The Snowflake-NVIDIA partnership to run GPU-accelerated inference through Cortex is the kind of infrastructure deal that takes years to copy. If Snowflake can be the place where enterprises both store data and run AI inference without a second vendor, the revenue per customer ceiling goes way up.

Snowflake's consumption-based pricing model is a double-edged sword — customers love the flexibility and it's a great sales motion, but it creates real revenue volatility quarter to quarter depending on how hard customers are optimizing their queries. The push toward Snowpark and Cortex is partly an attempt to add stickier, less optimizable workloads to the mix.