xAI brings Grok models to Databricks Agent Bricks for enterprise data

xAI integrated its Grok models into Databricks Agent Bricks, the developer platform for building AI agents, announced at the 2026 Data + AI Summit. The integration lets enterprise teams run Grok models directly against proprietary structured and unstructured data in Databricks' Lakehouse architecture, enabling agents to reason over company data without routing it to external systems. Databricks confirmed that xAI, as a model partner, does not retain data submitted through these features — a governance assurance aimed at enterprise data-trust concerns.

The official xAI account promoted the launch, branding the models as 'SpaceXAI's latest,' a nod to the reported corporate entanglement with SpaceX that has surfaced repeatedly this week. The Databricks deal is part of a broader multi-model trend — Databricks positions Agent Bricks as model-agnostic, hosting Grok alongside other frontier and open models so customers can route per task.

Competitively, this puts Grok in front of Databricks' large enterprise data customer base, a meaningful distribution win at a moment when model choice and data-locality are top enterprise priorities (echoing AWS Bedrock's multi-model approach and Microsoft's Copilot multi-model GA). For xAI, enterprise distribution diversifies beyond consumer Grok and X integration.

The shadow over the announcement is unavoidable: it lands the same week the Pentagon disclosed Grok's role in firing 2,000 munitions, which may give some enterprise and international customers pause regardless of the data-retention guarantees. Watch whether the military-use disclosure affects enterprise adoption of Grok via Databricks.