CoinFund’s Jake Brukhman says Anthropic’s export control compliance proves AI is centralizing fast. Here’s what he wants to do about it.
CoinFund founder Jake Brukhman is sounding the alarm on AI centralization. He pointed to Anthropic’s compliance with U.S.
AI export controls as proof that governments are tightening their grip on AI models.
Brukhman, who has tracked the intersection of AI and decentralized networks since 2020, says the risk is real. He argues decentralized AI networks are the only viable counterweight. For him, this is not a hypothetical debate.
Read also:
Bitcoin ETFs Bleed $10.5B as Binance USDT Data Signals Weak Demand
Decentralized AI as a Check on Government Control
Brukhman posted his concerns on X, stating that AI models have always been a centralizing force and a prime target for government control. He described Anthropic’s export control compliance as the moment that turned that view into market fact.
His position is that sovereign, open, and public AI running on decentralized networks is the necessary alternative. Without it, he warns, AI could fall under censorship and unilateral state control.
He also noted that his investment thesis did not shift to AI in recent years the way many crypto investors did.
Instead, he built what he calls deep expertise at the crossroads of AI and decentralized networks. That context shapes how he reads the Anthropic development.
Unlike many investors in crypto, I did not pivot to AI in the last few years. However, since 2020, I built some of the deepest understanding in this industry on the intersection of AI and decentralized networks (crypto, web3).
From the start, it was very clear that AI models are…
— Jake Brukhman (@jbrukh) June 13, 2026
The Compute Problem Blocking Decentralized Frontier Training
Brukhman identifies the compute gap as the core challenge for decentralized AI. He argues that enough commodity GPU resources exist globally to compete with large corporations on frontier model training.
The obstacle is not availability but algorithms. Training on distributed hardware requires new technical approaches that centralized infrastructure does not need.
He cited several teams working on that problem directly. Gensyn, Prime Intellect, Pluralis, Nous Research, Macrocosmos, and Covenant AI are among the projects he named.
Brukhman said these teams pursued distributed training research at a time when the broader industry considered it impossible. He now describes the results as not only viable but potentially cheaper and close in efficiency to traditional centralized approaches.
Tokenized AI Models and the Missing Business Case
Brukhman also raised the question of economic sustainability for open-source AI. He said open-source models, while valuable, currently lack a workable business model.
In his view, that limits their long-term viability. Among decentralized AI projects, he singled out Pluralis as the only one with a credible answer so far.
Pluralis is exploring a model where the weights of an AI model are split among multiple participants. That structure creates a tokenized ownership framework and an underlying business model.
Brukhman presented this as a meaningful distinction in the decentralized AI landscape, though he stopped short of predicting outcomes.
Leave a Reply
You must be logged in to post a comment.