AI data security could draw on Bitcoin proof of work as compute concentration raises concerns

Concentrated AI data centers can increase pressure on natural resources and create infrastructure risks. Distributing computing capacity and linking data security to electricity or other scarce resources through proof of work could offer a way to address those concerns.

In an interview posted on Anthony Pompliano's X account on November 10, Tillman Holloway, CEO of ArchPublic, said the AI industry could learn from Bitcoin's network design by distributing the infrastructure on which users depend.

Holloway said clustering large AI data centers in a single area puts pressure on natural resources and can provoke political opposition in small rural towns. Drawing a parallel with the concentration of Bitcoin mining among publicly traded companies, he argued that the burdens of centralized AI infrastructure could push computing capacity toward a more distributed model.

Individuals could contribute computing capacity much as owners of solar installations supply electricity to the grid and receive payment, he said. Even household appliances could provide processing power. Holloway argued that securing data against expanding AI threats would require tying security to electricity consumption or another scarce natural resource through a proof-of-work network.

Device security and protection for enterprise databases would both be necessary, he said. Phone-based liveness checks can help verify identity, while large repositories of sensitive information also need protection. In his view, combining proof of work with oracle technology, which brings external data onto blockchains, could strengthen resistance to brute-force attacks while allowing real-world information to enter securely in real time.

Holloway also argued that AI is making data collection and analysis more accessible, opening up new uses for information that was previously burdensome to process. He emphasized proficiency with the tools, saying people who know how to use AI would displace workers who do not.

Looking over the next 10 years, Holloway identified natural resources and electricity demand management as areas where data could have a substantial impact. Power systems must accommodate peak demand even though it occurs infrequently, he said. Improving battery capacity and rapid adjustments enabled by AI could make more grid capacity available, linking advances in data collection and analysis to the management and storage of electricity.