Snowflake Summit 2025 - highlights for geospatial

Jun 12, 2025

Updated Sep 7, 2026

3 mins read

TL;DR

Snowflake Summit 2025, held in San Francisco in June 2025, moved geospatial in two directions: interoperability and AI. On interoperability, Snowflake announced native GEOGRAPHY and GEOMETRY types for Apache Iceberg, developed with input from CARTO, and GeoParquet consolidated as the format for exchanging spatial data in columnar form, the combination letting teams write spatial data once and read it from any engine without conversions or lock-in. On AI, Cortex AISQL and Document AI made unstructured sources such as permits and scanned maps queryable with SQL and natural language, Cortex Agents answered natural-language questions about spatial datasets, and Semantic Views entered private preview so reusable spatial metrics like footfall density or drive-time coverage could be shared across teams. Several of these were announced as previews, so check Snowflake’s current documentation before assuming any of them shipped as described here in June 2025.

The Snowflake Summit has to be one of our favorite events of the year! Held in San Francisco, this year’s event showcased the next generation of AI, apps, and data platforms - and CARTO was proud to be at the heart of it all. From major announcements around agentic AI and interoperability to a landmark shift for geospatial data in the cloud-native ecosystem, the momentum around Location Intelligence has never been stronger.

In this post, we’ll share the top highlights from Summit - and what they mean for the future of geospatial analytics.

A photograph of the CARTO team — *Team CARTO at the Snowflake Summit*

What we’re most excited about: spatial taking center stage

At Snowflake Summit 2025, the announcements confirmed what many in the spatial community have long anticipated: geospatial data is finally stepping into the cloud-native mainstream. Two themes stood out most - interoperability and AI - and both are transforming how spatial data is used, shared, and understood.‍

‍Interoperability: open formats are changing the game.

Spatial data has historically been siloed - stuck in proprietary formats or isolated systems. Snowflake Summit cemented why that’s no longer the case.

Native geospatial support in Apache Iceberg. Snowflake’s announcement of GEOGRAPHY and GEOMETRY types on Iceberg is a major leap forward. Developed with input from CARTO, this brings native spatial support to one of the fastest-growing open table formats - enabling true multi-engine geospatial analytics. You can learn more about this development here.
GeoParquet becomes the standard. With strong community backing, GeoParquet is now the default for storing and exchanging spatial data in a columnar, compressed format. It unlocks better performance, lower storage costs, and full compatibility across various engines.
Access once, use anywhere. Together, Iceberg and GeoParquet allow users to write spatial data once and read it from anywhere - without conversions, rewrites, or vendor lock-in. This marks a foundational shift in how organizations manage and scale location data.

CARTO has played a key role in both initiatives, from defining specifications to helping cloud platforms like Snowflake support spatial interoperability by design.

AI: smarter spatial insights with Snowflake Cortex

The second big theme? AI-native workflows - and spatial is front and center here, too.

Cortex AISQL & Document AI. Anyone who works with spatial data knows the pain of being told “please find the data attached”... only to open a PDF. Snowflake is taking that pain away with their new AI SQL capabilities which let you query unstructured data (like PDFs or scanned maps) using SQL and natural language! This is huge for geospatial teams dealing with things like permits or planning documents.
Cortex Agents for Data Exploration. With agentic workflows, users can now ask natural-language questions about their spatial datasets and get contextual answers, visualizations, or next steps - instantly.
Semantic Models + Spatial Metrics. Snowflake’s private preview of Semantic Views and the sharing of Semantic Models allows teams to define and reuse spatial dimensions and KPIs - such as footfall density or drive-time coverage - across different teams and apps.

So… how can you start taking advantage of these developments?

A photograph of a presentation at the Snowflake Summit — *Christian Kleinerman, EVP of Product @ Snowflake*

CARTO + Snowflake: Seamless Location Intelligence

CARTO is deeply integrated with Snowflake at every layer - from storage to analysis to app development. The entire platform is run natively inside Snowflake - meaning spatial analysis without time-consuming ETL or painful data governance. It’s spatial analytics that scales with your cloud - fast, governed, and cloud-native by default. With openness at its core, CARTO extends easily to BI tools like Power BI and QGIS, letting teams connect spatial insights across their wider analytics ecosystem.

To make things even more seamless, CARTO is available as a Snowflake Native App and can be easily deployed via the Snowflake Marketplace.

An architecture diagram of CARTO + Snowflake — *Seamless integration between CARTO & Snowflake*

Conclusion: the future of spatial is cloud-native

The announcements at Snowflake Summit 2025 mark a turning point: geospatial data is no longer confined to niche systems or workflows. With developments across interoperability and AI, spatial analytics is finally becoming open, scalable, and accessible.

Want to see how you can take advantage of these updates? Request a demo to connect with our experts and explore the possibilities.

Frequently Asked Questions

What were the main geospatial announcements at Snowflake Summit 2025?

Two themes dominated the June 2025 event in San Francisco. Interoperability: native GEOGRAPHY and GEOMETRY data types for Apache Iceberg, and GeoParquet settling in as the standard columnar format for storing and exchanging spatial data. AI: Cortex AISQL and Document AI for querying unstructured documents, Cortex Agents for natural-language exploration of datasets, and a private preview of Semantic Views for defining and sharing reusable metrics. Taken together, CARTO read them as the point at which geospatial data entered the cloud-native mainstream rather than remaining a specialist workload.

What does native GEOGRAPHY and GEOMETRY support in Apache Iceberg mean?

Iceberg is an open table format that lets multiple query engines read and write the same tables in object storage. Adding GEOGRAPHY and GEOMETRY as first-class types means spatial data can live in Iceberg tables and be analyzed by any engine that supports the format, instead of being locked into one vendor’s proprietary spatial columns. Snowflake announced this at Summit in June 2025 and developed it with input from CARTO.

Why does GeoParquet matter for spatial data?

GeoParquet stores spatial data in a columnar, compressed layout, which cuts storage cost and speeds up the column-selective scans that analytical queries perform, while remaining readable across engines rather than tied to one tool. Paired with spatial types in Iceberg, it supports what the post calls access once, use anywhere: write the data a single time and read it from wherever, with no format conversions, rewrites or vendor lock-in. CARTO contributed to defining the specifications behind both.

How is Snowflake Cortex used with spatial data?

In three ways, as announced in June 2025. Cortex AISQL and Document AI let you query unstructured sources with SQL and natural language, which matters for geospatial teams whose inputs arrive as PDFs, planning documents or scanned maps rather than tables. Cortex Agents let users ask natural-language questions about spatial datasets and get answers, visualizations or suggested next steps. Semantic Views, in private preview at the time, let teams define spatial dimensions and KPIs such as footfall density or drive-time coverage once and reuse them across teams and applications.

How does CARTO run inside Snowflake?

As described in June 2025, CARTO is available as a Snowflake Native App and can be deployed from the Snowflake Marketplace, so the platform runs natively inside Snowflake at every layer from storage through analysis to app development. Because analysis executes where the data sits, there is no ETL step and no second copy of the data to govern. CARTO also connects outward to tools like Power BI and QGIS so spatial results reach the rest of an analytics stack.