Databricks pushes into 'physical AI'

Databricks is moving beyond analytics software and into the systems companies use to run warehouses, stores and power grids. (databricks.com) On April 13, Databricks said app builder Lovable can now connect directly to Databricks, letting teams build internal tools on company data with plain-English prompts instead of waiting for engineers to wire up dashboards or forms. (databricks.com) On April 13, Wiliot said it is building its Physical AI supply-chain platform on Databricks, feeding in live data from battery-free, postage-stamp-sized sensors that track inventory, shipments and temperature conditions. (rfidjournal.com) Physical artificial intelligence, in this case, means software that reacts to what is happening in the physical world rather than only analyzing stored records after the fact. Wiliot said Databricks will be the layer that ingests and analyzes those item-level data streams inside a governed environment. (rfidjournal.com) The same pitch is showing up in utilities. Tata Power said on April 9 that it will adopt Databricks across all business clusters, combining edge, operational and enterprise data and deploying Databricks Genie, a natural-language tool for querying company data. (tatapower.com) (databricks.com) That is a broader role than the one Databricks built its name on. The company said in December 2025 that it was investing in products including Databricks Apps, while its February 2026 financing closed at a $134 billion valuation after annualized revenue passed $5.4 billion in the January quarter. (databricks.com) (cnbc.com) Databricks is also adding deal capacity around that ecosystem. Global Corporate Venturing reported on April 13 that former ServiceNow investor Mike Kirk joined as head of corporate development, alongside Databricks Ventures, which says it backs startups building on data, analytics and artificial intelligence. (globalventuring.com) (databricks.com) Microsoft’s Azure Databricks documentation describes Lovable as a no-code app platform that can query lakehouse data through the Databricks application programming interface and machine-to-machine authentication. That turns Databricks from a back-end data store into part of the app layer employees actually touch. (learn.microsoft.com) The bet is that companies will not just ask Databricks to answer questions about their business. They will ask it to sit underneath the software that senses what is happening, decides what to do next and triggers the next action. (rfidjournal.com) (databricks.com)