Daniil and David Liberman, the brothers who previously sold a company to Snapchat for $64 million, are now turning their attention to decentralized AI infrastructure. Their latest message is a direct warning about the growing concentration of AI power in the hands of a small group of major technology companies, which they believe could leave users and builders increasingly dependent on centralized platforms.
A warning against concentrated AI control
The Liberman brothers argue that the long-term trajectory of AI points to a world with 10 billion robots, dramatically increasing demand for compute, inference, and network infrastructure. In their view, if that future is built primarily on centralized cloud systems, society risks surrendering computational sovereignty to algorithmic gatekeepers and becoming dependent on a narrow group of dominant operators.
Gonka’s pitch: decentralized compute for AI inference
To counter that trend, they are advocating for Gonka, a decentralized AI computing network designed to convert large-scale distributed computing power, including power associated with networks like Bitcoin, into useful AI inference through a decentralized protocol. According to the source material, this structure can cut GPU computing costs substantially compared with traditional cloud services.
The project’s early growth is being presented as evidence of demand. Within its first 100 days, Gonka’s computing power reportedly increased from 60 to more than 10,000 units. While the article does not provide further detail on how those units are defined or monetized, the jump suggests that decentralized AI infrastructure is attracting attention at a time when access to affordable compute has become a critical issue across the industry.
Decentralized infrastructure as an anti-monopoly thesis
For the Libermans, the case for decentralized AI is about more than price. They argue that open, distributed infrastructure is essential to prevent future AI resources from being controlled by a combination of large tech firms and financial giants. Their broader thesis is that wider participation in compute supply could help preserve openness and reduce the risk of monopoly control over the next generation of digital infrastructure.
More broadly, Gonka taps into one of the most closely watched debates in both crypto and AI: who will own and operate the infrastructure behind large-scale inference. As demand for AI compute continues to expand, the market will be watching whether decentralized networks can deliver sustainable supply, lower costs, and real-world adoption.