Nvidia acquires Kumo AI, a predictive AI startup known for its extreme accuracy

Nvidia Corp. has acquired another artificial intelligence startup, acquiring four-year-old model maker Kumo AI Inc.

The company designs AI models focused on making highly accurate business predictions. This is the latest in a series of startups acquired by the chipmaker over the past few years. Nvidia did not disclose the terms of the deal, but reports from Fortune and The Information say the deal has already been completed.

Kumo AI is known for highly accurate machine learning models that are trained and fine-tuned on customers’ own business data. One of its main advantages is convenience. Rather than spending months building complex data pipelines and feature engineering, companies can simply connect Kumo’s models to a data warehouse like Snowflake or Databricks and start predicting results using queries like a simple structured query language.

Kumo’s platform automatically handles all the work involved in preparing, cleaning, and combining historical business data, eliminating as much as 95% of the manual work required to set up predictive models. The company also translates users’ natural language queries into its own predictive query language. This means you can ask simple questions like “Will this customer churn?” or “What kind of demand do we expect to see in the next quarter?” Instantly generate answers based on your own company data.

The secret sauce of Kumo’s predictive model is a graph neural network. While traditional AI models examine each row of data in a database individually, Kumo’s graph learning considers them together, creating a large interconnected network of nodes and edges that maps the relationships between them. By mapping business data in this way, the model can see how each customer is connected not only to other users, but also to the various products and services the company sells. This enables you to generate more accurate and context-aware business insights.

Kumo co-founder and CEO Vanja Josifovski appeared on SiliconANGLE Media’s mobile livestreaming studio theCUBE during coverage of the 2024 Data Cloud Summit. He was joined by Brian Rikuda, Executive Vice President of Strategy, Operations and Programming Scheduling for Black Entertainment Television LLC, one of its largest customers, who spoke about how beneficial the platform has been.

“If you use a service like Kumo… [it allows us to] “By observing customer behavior, we can take stock of the content we have and start making more dynamic recommendations. If we can solve the retention problem of weekly or monthly subscriber loss, we are actually solving a huge problem for the overall growth of our business,” Rikuta said.



It’s not immediately clear how Nvidia plans to use Kumo’s predictive models or what motivated it to acquire the startup. But Josifovski and his co-founders Hema Raghavan and Jure Leskovec are already listed as Nvidia employees, according to their LinkedIn profiles, and have already moved on to the chipmaker. However, Kumo’s website makes no mention of the acquisition.

Mountain View, California-based Kumo has raised $37 million in venture capital funding through two rounds, both in 2022, with Sequoia Capital as its most prominent investor. The company’s customers include Reddit Inc., DoorDash Inc., and British grocery chain J Sainsbury plc.

Nvidia has acquired dozens of startups in recent years, but many of its deals have remained under the radar. One of its biggest recent acquisitions was a reported $20 billion “acquisition” that included buying technology assets and poaching key talent from AI inference company Groq Inc. Most recently, in February, it acquired a data semantics startup called Illumex Ltd. for about $75 million.

Image: Cloud AI