Bezos Backs Toloka: A Strategic Bet on Human-AI Collaboration in the AI Infrastructure Boom

Amazon founder Jeff Bezos has placed a major wager on the future of artificial intelligence, leading a $72 million investment in Toloka, a Dutch-based AI data solutions company under the Nebius Group. The move underscores growing investor confidence in human-AI collaboration as a critical pillar for ethical, scalable AI systems.

The Investment’s Strategic Weight
Bezos Expeditions, alongside Mikhail Parakhin (CTO of Shopify and a leading AI figure), is repositioning Toloka as an independent entity. The deal reshapes governance: Parakhin becomes Executive Chairman, while founder Olga Megorskaya retains her CEO role and joins the board. Nebius, Toloka’s parent company, steps back from direct control, ceding voting rights while maintaining an economic stake. This restructuring aims to free Toloka to scale globally, particularly in the U.S., where it had been constrained by sanctions linked to its Yandex heritage.

Why Toloka Stands Out
Toloka’s platform manages over 200,000 experts across 50 domains, 100+ countries, and 40 languages—a human network critical for training AI models. Its services include red-teaming (testing AI against malicious inputs), reward modeling (via RLHF and DPO methods), and code generation for programming assistants. Clients like Amazon, Microsoft, and Anthropic rely on Toloka to ensure AI systems align with human values.

Nebius, now reporting Toloka as an equity investment, has long prioritized long-term AI infrastructure projects. The firm’s cloud platform for intensive AI workloads and ventures like database tool ClickHouse signal a broader vision. By unshackling Toloka, Nebius can focus on its core strengths while enabling its star subsidiary to thrive.

The Human-AI Collaboration Trend
The $72 million infusion arrives as investors increasingly prioritize “safety-first” AI development. A 2024 McKinsey report found that 78% of enterprises view ethical AI alignment as critical to long-term success. Toloka’s focus on hybrid human-AI solutions directly addresses this demand, offering safeguards against bias, misuse, and technical failure.

Market Potential and Risks
The global AI data market is projected to grow at a 34% CAGR through 2030, driven by rising demand for training datasets. Toloka’s expertise in niche areas like code generation and red-teaming positions it to capture a significant share. However, competition is intensifying: startups like Scale AI and established giants like Google Cloud offer rival platforms. Toloka’s edge lies in its global human network and Nebius’s infrastructure support.

Conclusion: A Pivotal Moment for Ethical AI
Bezos’s investment in Toloka marks a bet on two critical trends: the need for human oversight in AI development and the structural shift toward independent AI data firms. With $72 million in fresh capital, Toloka can scale its 200,000-expert network and expand into markets like the U.S., where its services are in high demand.

The move also highlights Nebius’s strategic foresight. By divesting voting control, it avoids the pitfalls of overconcentration while still benefiting financially—a model that could reshape corporate venturing in the AI sector. As AI infrastructure stocks like Nebius and competitors surge (up 47% year-to-date in some indices), investors are signaling that the era of human-AI collaboration is here to stay.

Toloka’s success hinges on executing its vision: leveraging its platform to democratize access to ethical AI tools. With Bezos and Parakhin’s backing, it’s well-positioned to lead that charge—and redefine the standards for safe, reliable AI systems in the decade ahead.