Nvidia acquires Kumo AI for over $400 million, expanding enterprise AI capabilities

Nvidia Acquisition Brings Predictive AI Capabilities

Nvidia has acquired Kumo AI, a Mountain View-based startup that develops foundation models for business predictions, in a deal valued at more than $400 million 2

2

3

3

. The four-year-old AI software firm Kumo specializes in predictive enterprise AI software designed to analyze structured business data, including customer information and payment data that traditional AI models struggle to process effectively 2

2

.

Kumo AI's three co-founders—Vanja Josifovski, former chief technology officer at Airbnb; Jure Leskovec, a Stanford University computer science professor; and Hema Raghavan, a former AI lead at LinkedIn—transitioned to Nvidia last month 1

1

2

2

. While their LinkedIn profiles now list them as Nvidia employees, the Kumo website has not yet acknowledged the change 1

1

.

Expanding the Full-Stack AI Ecosystem

The deal marks a strategic shift for Nvidia, which has traditionally invested heavily in open-weight models. Kumo's proprietary technology represents a departure from this approach, offering specialized capabilities for enterprise AI applications 2

2

. The startup had raised $37 million in venture capital funding across two rounds in 2022 from investors including Sequoia Capital 1

1

.

Kumo's technology has already proven valuable to major companies including DoorDash, Reddit, and U.K. grocery chain Sainsbury's 1

1

. The startup's RFM model can instantly tackle predictions around customer churn and credit risk without additional training, a significant improvement over traditional, time-consuming predictive analytics 1

1

.

"With the foundation model, you point it to your data, you define what you mean by churn, and a second later, you get the prediction," Leskovec explained in a previous interview 1

1

.

Integration into Nvidia's AI Foundry Software

While Nvidia declined to comment on the acquisition, industry observers speculate the company could deploy Kumo's models within its AI Foundry software, designed to help companies build custom AI models 2

2

. Nvidia could also leverage Kumo's researchers to develop new, business-centric foundation models optimized for Nvidia hardware 3

3

.

This acquisition continues Nvidia's M&A spree of over 100 startups in recent years as it builds out its full-stack AI ecosystem 1

1

. The $5 trillion company has spent approximately $3 billion on acquisitions over the last five years 2

2

. Notable recent deals include a $20 billion agreement with AI inference company Groq in December 2025, the acquisition of data semantics company Illumex in February 2026, and a $700 million purchase of orchestration software company Run.ai in April 2024 1

1

.

Broader Push into Agentic Computing

The Kumo acquisition arrives alongside Nvidia's partnership with Microsoft to create Windows PCs built for agentic computing, shipping this fall 2

2

. This initiative complements Nvidia's March introduction of NemoClaw, an enterprise-hardened version of OpenClaw that adds security controls and policy enforcement for businesses in regulated industries 2

2

.

The timing also aligns with Nvidia's unveiling of new AI software and hardware products at the COMPUTEX conference in Taiwan, including a line of new CPU chips for PCs 3

3

. For enterprise buyers, the combined capabilities offer the ability to run AI agents locally, keeping data and decisions within their own infrastructure while reducing latency, privacy exposure, and cloud inference costs 2

2

.