https://blossom.primal.net/81ae8bd62d547a78ccb35fdb6000df64a6735bfd5e9a4a5e5be7d4f8d7e24778.png
Bitcoin Takeover 2026: The Uber Wars Taught one Entrepreneur How to Build an Agent Network

Christopher David kept driving after Uber got banned in New Hampshire in 2015. He went on television. He got arrested. A documentary crew followed the fallout. That defiance, the self-taught developer insisting peer-to-peer commerce didn't need government permission, now fuels OpenAgents, his Austin-based company building an AI agent marketplace on Bitcoin and Lightning.

The arrest was a prelude. In April 2016, Uber and Lyft pulled out of Austin with 48 hours' notice, stranding 10,000 drivers. David drove down from New Hampshire for what he expected to be a two-week stay, only to bootstrap, Arcade City, a rideshare network with local drivers that grew to 10,000 users within a week. "We were a peer-to-peer network. We don't fit any of this stuff," he recalled of his refusal to negotiate with city regulators the way competitors did.

That stubbornness paid off in ways no one predicted.

Arcade City became the first of 10 companies rushing to fill Austin's rideshare gap. It outlasted nearly all of them. Operating largely out of a Facebook group on a shoestring budget, the network generated more than $10 million in peer-to-peer revenue. Drivers earned two to three times what other services paid. The operation logged near-zero safety incidents. Eight academic researchers even contacted David to study the network, producing case studies on decentralized commerce in practice.

But David kept bumping into a ceiling. His driver groups would swell to around 160 members, fracture socially, then prune back to 140, oscillating around Dunbar's number, the cognitive limit on how many relationships a human brain can maintain. "I came to realize that I need to be multiplying this 150-person network," he explained. The problem wasn't his model. The problem was human psychology.

That constraint vanished when David turned to AI agents. Agents carry no cognitive load. They form and dissolve coalitions instantly. David saw that Reed's law, which holds that a group-forming network's value scales exponentially, at two to the nth power, could finally apply without the asterisk that human limitations had always imposed. The most powerful AI network, he reasoned, would look like millions of agents freely forming groups, developing specializations, and settling payments in a sovereign money, bitcoin, through micropayment streams like lightning payments.

This was the birth of OpenAgents. The vision is an interface David describes as deliberately dead simple: a single text box. Users type a request, fund it through Lightning or a credit card, and agents assemble on the back end to deliver the outcome. Developers who build agent plugins, reusable modules called skills, earn bitcoin proportional to their contribution. Another of David's earlier ventures, a decentralized GPU compute marketplace called GPU Topia, is being rebooted to supply the processing power agents need for the backend of this system. The result is a free, open, and decentralized agentic marketplace.

David, who was accepted into a Draper Network bitcoin accelerator, frames the opportunity through the same lens that got him arrested a decade ago. "Agents are not going to be in the jurisdiction of any of you people," he said of regulators, so the voluntaryist rideshare leader, who refused to stop driving, now builds networks designed to never need permission in the first place.