Brian Armstrong: Iron in His Veins

Armstrong describes the SEC lawsuit as a calculated gamble driven by long-term mission over short-term optics. When Gary Gensler and Elizabeth Warren allegedly weaponized regulatory ambiguity to kill crypto in the U.S., Armstrong knew compliance was impossible because the SEC refused to publish clear rules. After 30 meetings where the SEC said «go talk to your lawyer» and then issued enforcement actions the next day, Armstrong decided to sue.

Most investors and advisors told him not to do it. Public market investors hesitated to buy stock in a company suing its regulator. But Armstrong had talked to other financial services CEOs who had sued the SEC and won, including references to SpaceX and Palantir suing the government. He believed the fight was necessary to prevent a future regulator from unlawfully killing the industry again. The decision took three to four months of deliberation as the «temperature was rising.»

The lawsuit cost $50–100 million and caused $10–20 billion in stock damage over a few years. But Coinbase won: the SEC withdrew the case under the new administration, and judges published opinions stating the SEC behaved «in an arbitrary and capricious manner.» Armstrong keeps a memento in his office commemorating the victory. He now believes that standing up to regulators, even at great cost, is sometimes the only path to the right outcome.

During the George Floyd protests and COVID isolation, Coinbase town halls shifted from product questions to social issues. When Armstrong was asked if Coinbase would support Black Lives Matter, he said he'd look into it. An employee held the mic and demanded an answer. Armstrong said he didn't know enough yet. 300 employees staged a walkout in protest — which, in a remote environment, meant closing their laptops.

Armstrong felt he had compromised something about himself when he issued a bland statement 48 hours later to get people back to work. He started reading books like Jonathan Haidt's «The Coddling of the American Mind,» which argued that college campuses were training activists who saw their job as reforming companies rather than advancing the mission. Armstrong drafted a blog post saying Coinbase would not jump into every hot social issue but would focus on its mission: increasing economic freedom.

People inside the company begged him not to post it. He did it anyway and offered a severance package to anyone not aligned with the new direction. They had internal bets that 50% might resign. Only 5% took the exit. Armstrong later realized the vocal minority made it feel like half the company was against him. He was prepared to rebuild from zero if necessary, drawing inspiration from Lee Kuan Yew's «iron in my veins» speech about being willing to rebuild Singapore from scratch.

Armstrong inverts the typical corporate approval process. In most companies, you need your boss, your boss's boss, and several layers up to all say yes — a committee structure that makes companies risk-averse. At Coinbase, employees pitch ideas twice a year to a group that includes product leaders, the CFO, Armstrong, and a few talented young engineers. If any one person says yes and agrees to fund it out of their budget, the project is green-lit.

This system is modeled on pitching ten venture capitalists: you only need one yes. Armstrong admits he voted no on USDC, the stablecoin that generated roughly $800 million in revenue in 2025. Luckily, someone else funded it. He references the Steve Wozniak story: Wozniak pitched HP on the personal computer, they said no, and he left to found Apple with Steve Jobs. Armstrong fears missing the next big idea because a founder pitches internally and gets rejected by a single gatekeeper.

Armstrong also pushes decision-making down the org by designating clear DRIs (directly responsible individuals) and giving short time frames. He tries to provide risk tolerance, encouraging people to take bets with 20% success probability if the upside is 20x. Most companies are too risk-averse to approve such bets. Armstrong gives people «air cover» and takes responsibility if it fails.

Armstrong's long-term perspective came from trying short-term projects in college and after graduation. He started a tutoring company, bought rental properties in Houston, and dabbled in passive income ideas inspired by Tim Ferriss's «The 4-Hour Work Week.» He realized that running a sandwich shop is just as difficult as running a rocket company — everything is hard. If you're going to dedicate a decade or more to something, you might as well pick something that will have a major impact if it works.

Armstrong wrote down what he was passionate enough about to work on for 20 years even with little success. The only thing he could think of was tech entrepreneurship. He moved to Silicon Valley, shut down his other projects, sold his rental properties, and within a few years founded Coinbase. Within seven years, Coinbase had a billion-dollar valuation. He attributes this inflection point to committing fully to the big, long-term thing rather than chasing short-term wins.

He contrasts founders who tell him about their current project but reveal in conversation that their «real» big idea is something more ambitious they think is too difficult or needs more capital. Armstrong's advice: go for the big thing now, because you could spend the next two decades on the smaller thing and still find it difficult. You might as well work on something that will have major impact if it succeeds.

Armstrong describes a major shift happening in the last few months: AI agents need to make payments to get work done, but traditional corporate cards can't be issued to non-human entities. Coinbase is giving agents stablecoin wallets so they can autonomously pay for AWS resources, access paywalled research, buy domain names, or spin up marketing programs without bugging a human for approval every time.

This is a unique crypto application. Over 50% of code at Coinbase is now written by agents, and about 60% of customer support inquiries are answered by agents. Coinbase uses both vendor tools (like Glean, Slackbot, Gemini) and custom internal models. Armstrong says the key was ingesting all internal data — Google Docs, Slack messages, GitHub commits, Salesforce — so agents can answer questions like «what should I be more aware of as CEO?» and surface misalignments the executive team didn't know about.

Armstrong uses AI personally for research, drafting, and decision-making. Coinbase now includes a row in its decision-making framework for the AI agent to write its input, which the team compares to human recommendations. Armstrong believes machine-to-machine payments via stablecoins will become a major use case as agents become more autonomous digital employees.