Teradata joins Snowflake, Databricks in expanding MCP ecosystem

Teradata has released its first Model Context Protocol (MCP) server, designed to enable enterprises to connect their agents to data stored within the data analytics provider’s databases, amid rising demand for agentic applications.

The Teradata MCP server, hosted on GitHub and supported by the Teradata community, focuses on helping enterprises take the first step in testing their agentic workflows for accessing and analyzing data stored within Teradata.

‘Full support’ version is up next

The data analytics software provider has confirmed that it is planning to launch another version of its MCP server that will come with “full support,” which is missing with the Community Edition.

The new version, which is expected to be released by the first half of 2026, will include capabilities critical for enterprises such as security, observability, scalability, workload management, and compliance, making it suitable for production deployment, Meeta Vouk, VP of product management for AI and analytics at Teradata, told InfoWorld.

“The future releases will target advanced security and context handling capabilities, and resource optimization. Customers will be able to migrate from the community server to the new product,” Vouk added.

In contrast to the Community Edition that includes tools for running common development tasks, the upcoming version will come with additional tools that target more complex tasks, such as SQL generation and optimization, multi-modal data retrieval, performing data engineering tasks, running in-database analytics and machine learning pipelines, and executing custom Python code.

Releasing a Community Edition before a fully supported version, according to Moor Insights and Strategy principal analyst Robert Kramer, is a strategy to drive early adoption, while also laying the groundwork for a future commercial version.

Capabilities of the Community Edition

The Community Edition comes with sets of tools and prompts, grouped as modules, to enable agents to query, analyze, and manage data inside Teradata’s databases.

These sets include Base Tools for general platform engagement, DBA Tools for database administration tasks, and Data Quality Tools that accelerate exploratory data analysis.

In addition, the MCP Server comes with specialized tool sets such as Security Tools for managing permissions, Feature Store Tools for overseeing enterprise feature stores, and RAG Tools to facilitate vector store creation and usage.

A Custom Semantic Layer is also available, designed to enable enterprises to build domain-specific tools and prompts aligned with their unique business data needs, Teradata said.

The Community Edition, according to Kramer, is designed to help customers connect AI agents directly to operational data with built-in governance, metadata management, and tools like vector stores and RAG — key for generative AI use cases that rely on combining structured data with large language models (LLMs).

This access to data inside Teradata’s databases will allow agents to gain more contextual understanding and help deliver more accurate responses, Kramer said.

Teradata, however, isn’t the only data analytics provider that is offering an MCP server. While rival Databricks offers Managed MCP servers to enable governed access to structured data and unstructured data via Genie Space MCP Server, Vector Search MCP Server, and UC Function MCP Server, Snowflake, this month, released open source resources that simplify the creation of MCP servers connected to Snowflake services.

Separately, a contributor-driven Snowflake MCP server has been available on GitHub, even before Snowflake released its resources.