“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,” Kumo co-founder Jure Leskovec previously said while explaining the company’s technology.
Nvidia has added another artificial intelligence startup to its growing list of acquisitions, purchasing predictive AI company Kumo AI in a move that strengthens its push beyond chips and deeper into enterprise software.
The deal, first reported by Fortune, brings Kumo’s technology and leadership team into Nvidia at a time when competition in the AI industry is expanding beyond hardware and into specialized business applications. While Nvidia has not disclosed the financial terms of the acquisition, reports indicate the company paid more than $400 million for the startup.
Kumo AI may not be a household name, but the startup has built a reputation for helping businesses make highly accurate predictions from their data.
Founded in 2022 by former technology executives and researchers including Vanja Josifovski, Hema Raghavan, and Stanford University professor Jure Leskovec, Kumo developed foundation models designed to analyze structured business data and predict outcomes such as customer churn, fraud risks, payment defaults, and purchasing behavior.
The company’s three co-founders joined Nvidia last month, with their LinkedIn profiles now listing Nvidia as their employer. The Kumo website has not publicly announced the acquisition, and Nvidia declined to comment on the transaction.
The acquisition highlights Nvidia’s broader strategy of building a complete AI ecosystem. Over the past few years, the company has dominated the market for AI chips, becoming one of the world’s most valuable companies as businesses rushed to buy its hardware to train and run AI models. However, Nvidia’s ambitions now extend far beyond semiconductors.
The company has increasingly invested in software, AI models, cloud platforms, and enterprise tools that help businesses deploy artificial intelligence. Acquiring Kumo gives Nvidia access to technology that can help companies extract insights from large amounts of business data without requiring lengthy data science projects.
One of Kumo’s biggest selling points was simplicity. Traditional predictive analytics often requires companies to spend weeks or months building models, cleaning datasets, and engineering features before obtaining useful results. Kumo aimed to reduce that process dramatically.
“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 told Fortune in an earlier interview.
The technology attracted several notable customers, including food delivery company DoorDash, social media platform Reddit, and British supermarket chain Sainsbury’s. These companies used Kumo’s models to improve forecasting and decision-making using their internal data.
Kumo had also attracted strong investor interest before the acquisition. The startup raised approximately $37 million in venture funding from investors including Sequoia Capital. The backing reflected growing confidence in AI tools designed specifically for business applications rather than consumer chatbots.
Industry observers say the acquisition fits neatly into Nvidia’s long-term plans. The company has acquired more than 100 startups in recent years as it works to build a broader AI platform. Previous deals included AI software company Run:ai, data management startup Illumex, and other firms focused on strengthening Nvidia’s software offerings.
Reports from The Information suggest Nvidia paid at least $400 million for Kumo, underscoring the strategic importance of enterprise AI software to the company’s future growth plans.
The move also reflects a growing shift within the AI industry. Much of the public attention surrounding artificial intelligence has focused on chatbots and image-generation tools. Yet many businesses are more interested in systems that can help predict customer behavior, identify risks, improve operations, and generate measurable financial results.
That is precisely the area where Kumo built its reputation. Its models are designed to work with structured business information, including customer records, transaction histories, and operational data. Such information is often more valuable to companies than publicly available internet data because it directly affects revenue and decision-making.
For Nvidia, combining Kumo’s technology with its existing AI infrastructure could create new opportunities to serve enterprise customers looking for practical business applications of artificial intelligence.
The acquisition could also strengthen Nvidia’s AI Foundry platform, which helps businesses build and customize AI systems using Nvidia hardware and software. Analysts believe Kumo’s predictive models may eventually become part of Nvidia’s broader enterprise AI offerings.
As competition intensifies among major AI companies, acquisitions are becoming an increasingly important way to secure talent, technology, and market share.
For Nvidia, buying Kumo is about more than adding another startup to its portfolio. It is another step toward becoming a one-stop provider of AI infrastructure, software, and business tools.
The company’s dominance in AI chips helped launch the current AI boom. Now Nvidia appears determined to play an equally important role in how businesses use artificial intelligence once those chips are installed.
With Kumo now part of its growing AI empire, Nvidia is strengthening its position not just as the company powering AI, but as one helping businesses turn AI into results.
