Country Manager Shannon Katschilo at the opening of Snowflake Canada’s new Toronto offices, with mayor Olivia Chow in the foreground, April 2024. Photo by Jen Evans.
Snowflake started life as a cloud data warehouse. In 2025, it’s something much bigger: a neutral, multi-cloud AI Data Cloud that wants to be the place where enterprises keep all their mission-critical data and run all of their AI. Snowflake’s evolution from “fast analytics database” to “AI substrate” is what puts it at the center of the current AI infrastructure race – and in direct competition with Databricks and Microsoft Fabric.
From Cloud Data Warehouse to AI Data Cloud
Snowflake’s growth, especially in Canada, has been fast, even explosive. Its stunning offices on Queens Quay in Toronto with one of the best views in the city speak eloquently to its success. And its value proposition is elegant, profitable and simple: you don’t move sensitive data to AI – you bring AI to governed data.
Photo: Mayor Olivia Chow,
Minister Vic Fedeli and Premier Doug Ford watch a presentation by a Snowflake executive, April 2024. Photo by Jen Evans.
The company now describes its platform as an AI Data Cloud: a unified environment where organizations store, govern, share, and process data, and then run AI and applications directly on top.
Key architectural pieces:
- Decoupled compute and storage, with elastic, consumption-based pricing, still aimed at SQL and BI use cases.
- Cross-cloud reach (AWS, Azure, GCP), allowing multi-cloud and data-sharing scenarios that don’t force customers into a single hyperscaler.
- The Snowflake Marketplace / AI Data Cloud network, where customers can share and monetize data and models with thousands of other organizations.
On top of that, Snowflake has layered a growing AI stack:
- Snowflake Cortex AI – a fully managed layer for running LLMs, vector search, and AI functions (summarization, translation, classification, etc.) directly inside Snowflake.
- Model access – including Snowflake’s own Arctic model plus third-party models from Anthropic, Meta, Mistral, OpenAI, and others, all abstracted behind governed APIs within the platform.
- Snowflake Intelligence & agentic AI – a relatively new layer that lets users query organizational data in natural language and orchestrate “agents” that can perform multi-step analytic and operational tasks.
The recent $200M multi-year deal with Anthropic underscores how serious Snowflake is about being an AI runtime, not just a data warehouse. Anthropic’s Claude models (Sonnet, Opus, and successors) are being integrated deeply into Snowflake Cortex and Snowflake Intelligence to power “agentic AI” workloads directly on customer data, particularly in regulated industries like finance and healthcare.
The scale of this matters. Snowflake now reports:
- Over 12,600 customers globally.
- AI features used by 7,300+ businesses weekly, with Snowflake Intelligence landing ~1,200 customers in its first month.
In short, Snowflake’s role in the AI ecosystem is to be the governed, neutral execution layer for enterprise AI: a place where internal, partner, and public data can live, be shared, and be acted on by AI models and agents without data leaving a secure environment.
Competitor #1: Databricks – The AI-First Lakehouse
If Snowflake is the neutral AI Data Cloud, Databricks is the AI engineer’s workshop.
Databricks built its reputation on Apache Spark and the lakehouse concept: an architecture that merges the flexibility of data lakes with the governance of warehouses. Its Data Intelligence Platform targets the full spectrum of data workloads – ETL, streaming, analytics, and machine learning – on a unified lakehouse foundation.
Where Snowflake historically leaned into SQL analytics and governance, Databricks leaned into:
- Advanced data engineering and big data processing (Spark, Delta Lake).
- ML and MLOps via MLflow and now Mosaic AI.
- Data-centric generative AI, especially with Lakehouse AI, which pushes the idea that the best generative systems are built where your training and feature data already lives.
Databricks’ AI posture has been reinforced by major moves:
- Launch of Lakehouse AI and Data Intelligence features to make the platform more “AI-native” rather than just a big data engine.
- A strategic partnership with OpenAI to integrate GPT-5 and other models into the platform and its Agent Bricks product, aimed at enabling 20,000+ customers to build and scale AI applications.
- A funding round valuing Databricks at around $100B, driven largely by investor conviction that Databricks is a core AI infrastructure provider.
From Snowflake’s perspective, Databricks is the most direct technology rival: both want to be the place enterprises centralize their data and run AI. Financial coverage repeatedly frames Databricks as Snowflake’s primary competitor as both race to capture AI workloads.
How they differ in the AI landscape:
- Snowflake:
- Stronger with BI teams, SQL analysts, data-sharing and governance.
- AI is increasingly “embedded” into that experience (Cortex, Intelligence, agentic workflows).
- Plays the neutral, multi-cloud, app-builder role.
- Databricks:
- Stronger with data scientists, ML engineers, and organizations building custom models.
- AI is treated as an extension of its existing ML and big-data pipeline strengths.
- Capitalizes on open formats and open-source tooling.
In the real world, many large enterprises now run both: Databricks as a modeling and engineering environment and Snowflake as the governed analytics and application substrate. The competitive tension shows up when each tries to expand into the other’s home turf: Snowflake pushing deeper into ML and agents, Databricks pushing harder into BI and SQL-friendly workloads.
Competitor #2: Microsoft Fabric – The Integrated Data+AI Universe
If Databricks is Snowflake’s open, engineering-heavy rival, Microsoft Fabric is the hyperscaler-integrated one.
Fabric is Microsoft’s unified analytics and AI platform that bundles Power BI, Azure Synapse, and Azure Data Factory into a single SaaS experience, tightly integrated with Microsoft 365 and the broader Azure ecosystem.
Fabric’s value proposition:
- A single surface where business users and technical teams can go from raw data to dashboards, ML models, and AI copilots in one interface.
- Deep integration with Microsoft’s GenAI ecosystem (Copilot, Azure OpenAI), giving every knowledge worker AI-enhanced workflows “for free” if they’re already in the Microsoft stack.
- Familiar governance and identity model for organizations already standardized on Azure and Microsoft 365.
Analysts and practitioners now routinely compare Databricks, Snowflake, and Microsoft Fabric as the three core contenders for the “AI & Data Cloud crown.”
In parallel, broader coverage of the AI data-platform race highlights Alphabet (BigQuery + Gemini + Vertex AI) and Microsoft (Fabric + Azure) as major structural threats to independent platforms like Snowflake, because they can bundle storage, compute, AI models, and productivity tools into one tightly integrated offer.
From Snowflake’s vantage point, Microsoft Fabric is dangerous for a few reasons:
- Default advantage
If you’re an enterprise already living in Microsoft 365, Teams, Power BI, and Azure, Fabric is the easiest default upgrade path to “AI-powered analytics.” - AI everywhere narrative
Microsoft positions AI as embedded in every layer of the stack – from code copilots to Office to data. That narrative resonates with business leaders who want instant wins rather than re-platforming their data stack around a neutral vendor. - Platform lock-in
Fabric subtly reduces the need for a neutral platform like Snowflake by keeping data, analytics, and AI tightly within the Microsoft universe. As one analysis notes, Fabric competes with Snowflake and Databricks by offering a comprehensive data platform with integrated AI and a unified pricing model.
Where Snowflake Fits – And Where the Battle Lines Are
Put together, the competitive picture looks like this:
- Snowflake wants to be the multi-cloud, neutral AI Data Cloud – the shared substrate where governed data, models, and agents live, and where enterprises can avoid betting everything on one hyperscaler.
- Databricks wants to be the data-intelligent workbench for AI-first organizations, particularly those with strong data science and engineering teams and complex pipelines.
- Microsoft Fabric wants to be the fully integrated analytics and AI fabric for organizations already committed to Microsoft – where every user, from analyst to executive, lives in an AI-enhanced interface tied to their productivity tools.
For enterprises making AI platform bets, the practical questions become:
- Do you want open, engineering-heavy control (Databricks),
- neutral, governed multi-cloud reach (Snowflake),
- or tight integration with your existing productivity and cloud stack (Microsoft Fabric)?
Snowflake’s role in the AI ecosystem will ultimately be decided by how convincingly it can keep AI “close to the data” while still giving customers enough flexibility to mix models, clouds, and applications. Its expanding partnerships (Anthropic, AWS, Google, and others) suggest it understands that neutrality plus AI is its differentiator.
In a world where every vendor claims to be an “AI platform,” Snowflake’s real test is whether it can remain the place where enterprise reality lives – the governed source of truth that everyone else’s models have to come to if they want to matter.
