Matt Aslett's Analyst Perspectives

As I explained in our recent Buyers Guide for Data Platforms, the popularization of generative artificial intelligence (GenAI) has had a significant impact on the requirements for data platforms in the last 18 months. While there is an ongoing need for data platforms to support data warehousing workloads involving analytic reports and dashboards, there is increasing demand for analytic data platform providers to add dedicated functionality for data engineering, including the development, training and tuning of machine learning (ML) and GenAI models. Snowflake is a prime example. The company remains best known as a cloud data platform provider for data warehousing workloads but, like many other data platform providers, has improved its support for AI workloads in the last year. In particular, the company made several announcements at its Snowflake Summit 2024 customer event in June that highlighted considerable advancements to its capabilities in relation to AI and GenAI.

Snowflake was founded in 2012 to build a business around its cloud-based data warehouse with built-in data-sharing capabilities. Snowflake has expanded its reach over the years to address data engineering and data science, and long ago moved beyond being seen as just a cloud data warehouse. The company was rated Exemplary in our recent Buyers Guide for Analytic Data Platforms and a Provider of Assurance in our Buyers Guide for AI Platforms. Snowflake is not alone in adding support for AI workloads to its data platform. I assert that through 2026, analytic data platform software providers will increase their native support for AI and ML algorithms, functions and tools to enable in-database data science. The extent to which the company is positioning itself for AI and GenAI workloads was a focus at Snowflake Summit this year, with Snowflake describing its platform as an AI Data Cloud. The company did more than just add the letters A and I to its former Data Cloud branding, however, as it also made numerous announcements that justified its increased emphasis on AI. These included improvements to its Snowflake Cortex managed service for developing applications based on large language models (LLMs), as well as its Snowflake ML offering for training and operationalizing ML models for predictive analytics. Snowflake also enhanced its capabilities for data management and data governance, which will improve support for both analytics and AI workloads.

Snowflake has been hugely successful in helping to drive adoption of cloud-based analytic databases, establishing itself as one of the major cloud data platforms for SQL-based data warehousing. The launch of the Snowpark developer environment in 2020 was significant in widening the target workloads by enabling data engineers, data scientists and developers to execute custom Python, Java and Scala code against data in Snowflake. This was followed in 2023 by the launch of Snowpark Container Services, which enables customers to run their own choice of third-party software, including programming languages, data science libraries and GenAI models, on the platform. This avoids the need for users to move data out of Snowflake for processing and potentially reduces complexity and infrastructure resource requirements.

I noted in the wake of the Snowflake Summit customer event last year that the company’s plans for native support for GenAI were comparatively nascent relative to some rival providers that had been quicker to add support for GenAI models and development. Snowflake’s GenAI strategy took a leap forward in late 2023 with the launch of the Cortex AI development platform, which provides users with access to LLMs, AI models and vector search functionality. That was followed in April by the delivery of Snowflake’s own Arctic family of LLMs. At Snowflake Summit 2024, the company announced a number of enhancements to Cortex, including the preview of Snowflake Cortex Analyst, which provides an interface for natural language querying of structured data by business users, and Snowflake Cortex Search, which provides developers with a combination of keyword and vector search to support enterprise search and retrieval augmented generation workloads. Natural language search and query are amongst the most popular early use cases for GenAI, with 99% of participants in the ISG Market Lens AI Study having seen positive outcomes from natural language search and 97% having seen positive outcomes from the interpretation of data. Snowflake also added Snowflake Cortex Guard to automatically identify inappropriate language in the output of LLMs.

In addition to GenAI enhancements, Snowflake also expanded the MLOps capabilities of its Snowflake ML offering, including the general availability of Snowflake Model Registry and the preview of Snowflake Feature Store and ML Lineage. Snowflake also announced several data management and governance enhancements that will support both analytic and AI workloads, including the general availability of support for Apache Iceberg Tables and the launch of the Polaris Catalog open-source project built on the Apache Iceberg REST protocol. Polaris is designed to provide a vendor-neutral technical catalog for Iceberg tables but complements the data governance and discovery capabilities of Snowflake Horizon, which was also recently enhanced with the preview availability of Internal Marketplace, which enables the creation, sharing and consumption of data products. Many of the new capabilities announced at Snowflake Summit 2024 remain in private or public preview. As such, they were not included for assessment in our recent Buyers Guide reports. They nevertheless represent a significant enhancement of the company’s native support for AI and GenAI that will improve our evaluation of Snowflake in future reports. As a result, I recommend that enterprises include Snowflake in their evaluations for AI platforms, as well as analytic data platforms.

Regards,

Matt Aslett