CoinFund Founder: Decentralised AI Is Counterweight to Government Power

Jake Brukhman, founder of CoinFund, draws a clear conclusion after the news that Anthropic complies with US export control regulations for AI: AI models are increasingly becoming a centralising instrument and a target for government steering. According to him, decentralised AI networks are the only real counterforce.

Anthropic as a wake-up call for the crypto world

Brukhman says it has been clear to him from the start that AI models are inherently centralising forces. Anthropic’s decision to comply with US export controls now turns that point into a market reality. Through these large, centralised AI companies, governments can increasingly gain control over the technology and its distribution.

The founder indicates that he has not switched to AI in recent years as many other crypto investors did. Nevertheless, since 2020 he has built a deep understanding of the overlap between AI and decentralised networks. That foundation now forms the basis for his view that decentralised AI is not a nice-to-have, but a necessity.

Distributed GPUs as a solution to the compute problem

Brukhman points to a specific bottleneck: it seems nearly impossible to train large-scale AI models without the centralised infrastructure of big tech companies. But according to him, decentralised AI networks offer an answer to this by using distributed GPU resources to train frontier models.

He mentions a number of teams actively working on distributed training methods, including Gensyn, Prime Intellect, Pluralis and Nous Research. Notably, Pluralis is also exploring a business model around tokenised AI models, where the model weights are split and distributed among participants. This makes it not only technically distributed, but also economically accessible to a broader group.