Jeff Bezos, through his personal firm Bezos Expeditions, is leading a $72 million funding round in Toloka, an AI data solutions company aiming to scale its global presence, particularly in the United States, Toloka told Reuters on Wednesday.
Toloka specializes in training and evaluating AI models using a global network of human experts and testers, providing high-quality data labeling and validation. The company is part of Nebius Group (NBIS.O), an AI infrastructure firm listed on Nasdaq and formerly affiliated with Russian tech giant Yandex.
The investment marks a significant milestone for CEO and founder Olga Megorskaya, who said the funding would accelerate product development by fostering collaboration between AI agents and human experts.
“There will always be the need for control, verification, and help from human experts to ensure that the result is actually of high quality,” she said.
Strategic Backing and Global Shift
The deal comes after Nebius successfully split from Yandex in a $5.4 billion exit from Russia, the largest corporate withdrawal since the 2022 Ukraine invasion. The restructuring allowed Nebius and Toloka to pursue Western capital without violating sanctions.
Other notable participants in the round include Mikhail Parakhin, CTO of Shopify, who will also serve as Toloka’s executive chairman. Parakhin emphasized the urgent global demand for trusted AI data solutions.
In late 2023, Nvidia invested in a $700 million private placement in Nebius, highlighting growing institutional interest in AI infrastructure and tools.
With this latest funding round:
-
Bezos Expeditions and other new investors gain equity
-
Nebius retains a majority economic stake, but gives up majority voting control, enabling Toloka to operate independently
-
A future funding round is anticipated, Megorskaya said
The investment underscores a broader trend of scaling AI companies focused on high-quality data pipelines, as tech giants like Amazon, Microsoft, and Anthropic increasingly rely on curated training datasets for safe and effective AI model development.