NVIDIA Acquires Predictive AI Startup Kumo for Over $400 Million

According to Analytica Beating monitoring, NVIDIA has acquired the five-year-old predictive AI startup Kumo AI for a reported amount exceeding $400 million. The acquisition aims to leverage Kumo's predictive large model to provide predictive analytics services based on structured data to enterprise clients. Kumo's three co-founders, including former Pinterest CTO Vanja Josifovski, Stanford professor Jure Leskovec, and former LinkedIn AI head Hema Raghavan, have joined NVIDIA in May 2026.

Enterprise structured data consists of interrelated multiple data tables, which are generally challenging for large language models to parse directly. Kumo combines graph machine learning technology with synthetic data generated in a simulated enterprise environment to train a proprietary large model capable of predicting customer churn rate or payment defaults, thus achieving intelligent forecasting of enterprise data warehouses.

NVIDIA may integrate Kumo's technology into the AI Foundry software service to assist enterprises in combining private data with domain knowledge to customize proprietary models that fit their business scenarios. In April 2026, Kumo unveiled its latest prediction model, KumoRFM-2. Currently, Kumo's partners and clients include DoorDash, Reddit, Databricks, and Snowflake, among other enterprises.

In contrast to the $20 billion acquisition of Groq's inference chip technology at the end of 2025, NVIDIA tends to favor small and medium-sized deals in M&A. According to regulatory filings, NVIDIA's cumulative acquisition spending over the past 5 years amounts to approximately $3 billion.