Dataloop
Definition
Dataloop is an Israeli company founded in 2017 that develops a platform for managing, labeling and versioning training data. It was acquired by Dell in December 2025 for $120 million.
The problem: data preparation as grunt work that decides everything
Dataloop was founded in 2017 in Herzliya by Eran Shlomo, Avi Yashar and Nir Buschi. A model's quality is largely determined by the quality of the data it trains on, not only by its architecture. In practice, training a computer vision model requires collecting tens of thousands of images, consistent labeling and version tracking, a process that is often done in scattered spreadsheets and folders.
Human in the loop: a compromise between automation and human judgment
The platform brought this chain together in one place: managing datasets, labeling interfaces, partial automation and version management. At the center of the approach is a "human in the loop" model: the software suggests an initial label, and a person approves, corrects or rejects it. This compromise cuts manual work without giving up human judgment in the hard cases.
Funding up to the acquisition
Up to the acquisition, the company raised about $50 million from investors including NGP Capital and Alpha Wave Global. In December 2025 Calcalist reported that Dell was buying it for $120 million in cash.
March 2026: Dell's official announcement details the integration
According to an official announcement Dell Technologies published in March 2026, together with NVIDIA, Dataloop's technology became a central engine in Dell's new data platform: "The Dell Data Orchestration Engine, powered by technology from Dell’s recent Dataloop acquisition, redefines how enterprises operationalize data for AI," the announcement said. The no-code, low-code engine automatically discovers, labels, enriches and transforms data into AI-ready datasets. This is the first official confirmation from the buyer itself of the technology's integration, beyond the press reports on the deal.
Israeli data-stage companies bought by giant corporations
The company belongs to a group of Israeli companies that worked on the data stage and not on the model itself, a stage that was seen for years as grunt work and turned out to be one of the places where the quality of the system is actually decided. That also explains the interest shown by large hardware makers and cloud providers such as Dell.
FAQ ❓
What exactly does Dataloop do?
It offers a platform for labeling, version management and partial automation of training data, with a "human in the loop" approach that combines automatic label suggestions with human approval.
When, and to whom, was it sold?
To Dell, in an all-cash deal of $120 million reported in December 2025.
Is there official confirmation of the acquisition from Dell?
Yes. An official Dell announcement from March 2026 states explicitly that its new data engine, the Dell Data Orchestration Engine, is powered by technology "from Dell’s recent Dataloop acquisition".
Why is data labeling so important?
Because a model's quality is largely determined by the quality of the data it was trained on, not only by its architecture. It is a stage seen as grunt work, but in practice it decides the result.
Found a mistake, or have a source we should add? Write to us