Summary

Coinbase sued its regulator, and Armstrong says it won—no fines and no changes to the company—after roughly 30 SEC meetings in which Coinbase asked for rules and was told to talk to its lawyer. Armstrong estimates legal and related costs at “$50 million to $100 million” and stock damage at “probably… $10 billion to $20 billion, maybe more.” He says the proposed market-structure fix is the CLARITY Act pursued in the House and a Senate version in draft, clarifying CFTC commodities versus SEC securities so “a future Gary Gensler couldn’t come in and try to kill the industry.”
Armstrong’s account of the lawfare mechanism is the episode’s sharpest institutional insight: Elizabeth Warren exercised extrajudicial pressure through bank regulators rather than Congress. Regulators “can choose to lose your paperwork” for 90 days or five years, so when they express “deep concerns” about crypto clients, banks comply without a law being passed—the same approach, he says, used against oil, gas, and firearms lending. The SEC crackdown helped push crypto companies offshore, benefiting places like the UAE and the Bahamas.
Coinbase is repositioning from crypto exchange to “Everything Exchange”—stocks, commodities, prediction markets, loans, and a spend card—targeting a “multitrillion-dollar market,” and Armstrong is explicit this was not a master plan. “I knew it was massive, I just didn’t know exactly how it’d play out”: he didn’t foresee stablecoins or prediction markets, only that Bitcoin was a wedge into updating financial services.
The internal capital-allocation system is a live signal for where product upside can come from: twice a year, any employee can pitch “Next Bets,” and one yes from a budget-holder can greenlight it— inverting the five-yes committee veto. Armstrong voted no on USDC; someone else funded it, and he thinks Coinbase generated “about $800 million” in 2025 revenue from it, “or something.”
AI is already structural at Coinbase—more than 50% of code is written by agents and about 60% of customer-support inquiries are answered by them—and the crypto-specific angle is giving AI agents stablecoin wallets for machine-to-machine payments, since traditional corporate cards can’t be issued to nonhuman entities. This is new in “the last few months,” but Armstrong says it is getting traction.
The 2020 “Mission First” apolitical stance—triggered by a 300-employee virtual walkout over BLM—led to 5% of the company taking severance, and Armstrong connects it to the same willingness to act despite being disliked that led Coinbase to sue the SEC. His model was Lee Kuan Yew’s “iron in my veins” speech: “I could go back to being on my laptop again if I had to.”
Outside Coinbase he committed $100 million of his own money to NewLimit, an epigenetic-reprogramming longevity company, whose first drug candidate enters clinical trials “probably next year”—faster than the five-to-six-year pure-research phase he expected, with three to five candidates hoped for over five years.
The formative macro thesis came from lived experience, not theory: in Argentina, Armstrong recalls reading that the country had been a top-10 economy around 1908 and saw hyperinflation and decay; Senra supplied the comparison to the “100th-richest” economy. At Airbnb, payouts through Latin American cash-pickup oligopolies carried 7%–12% fees; for Ecuador, or a similar country, “we basically just decided to send $100” to see what arrived. The conclusion was a financial system that is “fast, cheap, permissionless, decentralized.”

Deep dive

1. Market structure legislation is the endgame of a two-agency turf war

Armstrong goes to Washington “once or twice a quarter” because the industry is “right at the crux” of market-structure legislation: the CLARITY Act was being pursued in the House, the Senate is drafting its own version, and the substance is classifying crypto assets as CFTC commodities versus SEC securities.
His framing of why it matters: the ambiguity is “a totally parochial issue in the United States”—the UK and Singapore have one financial regulator and don’t care—but here the gap “was really weaponized by Gary Gensler… and Elizabeth Warren, and some people like that who tried, in my view, to unlawfully kill the industry.”
The stakes as he defines them: Bitcoin is clearly a commodity “like oil, gold, or copper,” but token issuers raising money at “various stages of decentralization” sat in a gray zone Congress had not clarified—and legislation is his proposed durable protection against a repeat.

2. How Warren’s leverage actually worked: regulate by examination, not legislation

Armstrong’s mechanism, spelled out: only Congress makes laws, but bank regulators have discretion—“they can choose to lose your paperwork and not approve something for 90 days, or two years, or five years.” So when an examiner says serving crypto companies “is not illegal per se, but we’re going to have a lot of questions… in the next exam,” everyone inside the bank gets the message. “If they say, ‘Jump,’ you sometimes want to say, ‘How high?’”
His characterization of Warren—hedged as his view but categorical in substance—is that “she’s a socialist” who believes the government should run financial services; he says she used appointed regulators to pressure banks into stopping loans to oil, gas, and firearms industries before crypto arrived as “a new system operating outside of that.”
The SEC’s conduct as Armstrong tells it: roughly 30 meetings after Coinbase became public, where Coinbase said “you tell us the rules, and we follow the rules,” were met with “We’re not going to give you any advice. Go talk to your lawyer”—followed, he says, by enforcement actions without a citation of what law was broken.

3. Suing the SEC: a three-to-four-month decision with years of short-term pain

Both sides sued: the SEC brought an enforcement action and lawsuit; Coinbase proactively sued under the Administrative Procedure Act because the agency had failed its legal duty to engage with the industry to promulgate rules. People Armstrong talked to said, “Do not sue the SEC,” but he had spoken with financial-services CEOs who had sued and won, and cites SpaceX’s NASA suit and Palantir as precedents: “there are moments where you have to stand up and sue the regulator… to actually get the right outcome.”
The decision took “3 or 4 months” as the temperature rose. Armstrong estimates legal and related costs at “maybe in the $50 million to $100 million range” and stock damage at “probably… $10 billion to $20 billion, maybe more” over a couple of years.
The outcome as he describes it: “we didn’t pay a single dollar in fines. We didn’t have to change a single thing”; the SEC withdrew the case under the new administration, and several judges published opinions saying the SEC had behaved in an “arbitrary and capricious manner.” He keeps a commemoration in his office.
The counterfactual is worth flagging: had this happened before Coinbase had accumulated more resources, Armstrong says it “probably” could not have afforded the fight, and “a lot of startups did die as a result.” Resources made the litigation possible.

4. The long-term orientation came from short-term failures—and maybe the spectrum

Origin of the time horizon: everything is difficult, “even if you’re running a sandwich shop,” so “you might as well pick something that you care about, that’s the really big thing”—along with his frustration at entrepreneurs who defer their real ambition: “you should just go for that now.”
On autism, Senra pushes back (“You’re not autistic to me”), and Armstrong holds his ground: “I mask it well.” He cites difficulty reading faces, overstimulation, and a compensating ability to focus for nearly 12 hours on interesting work.
The functional connection is tentative: Armstrong says he may be less concerned with “social cohesion or what other people think,” but he does not know whether that is an autism-spectrum trait. “It’s not that I like being disliked, it actually causes me a fair amount of stress too, but I don’t let that stop me.”

5. The Mission First post: a walkout, a severance offer, and Lee Kuan Yew

The trigger: at a COVID-era remote town hall, an employee asked whether Coinbase would endorse Black Lives Matter; Armstrong said “I don’t know. I haven’t looked into it,” Slack erupted, and 300 employees staged a virtual walkout—effectively closing their laptops. After a 48-hour statement supporting equality, “I felt like I had compromised something about myself.”
His diagnosis came from Jonathan Haidt’s The Coddling of the American Mind: campuses were training activists who joined companies not to advance the mission but “to hold truth to power… and actually reform the company as an activist.” The response was apolitical at work, “unless it has to do with our mission, crypto, and economic freedom.”
Internal readers “begged me not to post it.” He offered severance to anyone unaligned; internal guesses ran as high as 50% resigning, but 5% took the package. Armstrong later judged that the apparent division had been driven by a very vocal 1% minority, with others sympathetic to the cause.
The steel behind it, from Lee Kuan Yew’s “iron in the veins” speech about a strike involving an airline or air-traffic workers, was: “I will rebuild it all from scratch.” Armstrong’s contrast is between a founder and a president of a company: “I started it when it was just me on a laptop… I could go back to being on my laptop again if I had to. That’s leadership.”

6. Before Coinbase: passive-income dreams, The Dip, and one clarifying piece of paper

The pre-history: a college tutoring-matching startup, University Tutor, born from turning roughly $7–$8-an-hour library work into $60-an-hour tutoring, then Houston rental houses—The 4-Hour Workweek-era attempts at “$100,000 a year passively” with no plan beyond.
Seth Godin’s The Dip broke the pattern: most people quit between beginner and the top 1%. Armstrong wrote down what he would do for 20 years “even if I saw little or no success”; the only answer was tech entrepreneurship. He sold the rentals, moved to Silicon Valley, and “within 7 years of that decision, Coinbase had a billion-dollar valuation.”

7. Argentina plus Airbnb equals the thesis

Argentina, where he lived for a year seeking adventure, supplied the macro backdrop. Armstrong recalls reading that around 1900—he thinks in 1908—it was one of the world’s top 10 economies and saw hyperinflation, pessimism, and grand buildings in decay. Senra described the longer fall to the “100th-richest” economy; Armstrong also said that mortgages were largely unavailable, leaving only wealthy people able to buy real estate.
At Airbnb, after reading the Bitcoin white paper in December 2010, Latin American payouts ran through Western Union-style cash-pickup oligopolies at “7% to 12%” fees. For Ecuador, he thinks, or one of those countries, the documentation was so opaque “we basically just decided to send $100” and found someone local to report what arrived.
“It gave me such a visceral sense of how broken the global financial system is”: each country had its own costly, opaque rails. The analogy was a foreign webpage that required an exchange fee, arrived in another language, and took seven days.

8. Nights and weekends, cofounder dates, and the Fred Ehrsam pairing

The grind: 8:30 p.m. to midnight five days a week after Airbnb shifts, on his own laptop, reimplementing “a whole Bitcoin node in Ruby.” He sought a cofounder largely because Paul Graham’s essays said complementary skill sets improved the odds of getting into YC and eventually succeeding, then failed to find one for a year and a half.
The YC application with Ben Reeves of Blockchain.info—someone he had only recently met over coffee—was, Armstrong says, “a bad idea. You should really cofound with people you’ve known for a long time.” It collapsed within three months, and Armstrong went through the program solo.
Fred Ehrsam, a former Goldman Sachs FX trader, arrived after the seed round. Armstrong says Coinbase probably would not have succeeded without him: he was “an absolute killer” through years of near-death experiences, including spotting in their first three weeks that Coinbase was losing money on every Bitcoin trade.

9. Product-market fit as near-death: golf clubs, a $30K legal opinion, and a one-graph raise

The first Coinbase had no buy button—Armstrong thought he was building a payments wallet until user interviews surfaced the obvious: “the app was pretty cool, but I just don’t have any Bitcoin.” Adding ACH buying, through a YC introduction to Silicon Valley Bank and a $30,000 five-page legal opinion arguing Coinbase might not be a money transmitter, produced instant product-market fit. The license would have cost $5–$10 million and taken three or four years against $600,000 raised: “instead of pushing a boulder uphill every day, the boulder was rolling down the hill.”
Support backlogs hit 10,000-plus tickets; users triangulated the office from a photo’s background buildings and showed up at odd hours—“once in a while, Fred would go answer the door holding a golf club,” and the company sometimes wrote walk-ins a physical check. Senra confirms he was one of the angry early customers.
Then the cash-flow crisis: roughly $550,000 was moving through the account daily against a $600,000 balance while Coinbase pre-bought Bitcoin. The banker warned, “If you just have one error, you’re insolvent.” With no deck, they showed investors a few pieces of data—“an up-and-to-the-right graph of demand”—and raised $25 million in a week from Union Square Ventures and Ribbit. According to Senra’s recap, a16z came in at the Series B.

10. Hiring for spikes: the lumberjack over the Google manager

Armstrong endorses the Karim/Daniel Ek “hire for spikes” doctrine: look at past outlier work rather than resumes, and favor interviews where you leave with “more energy than I went in.” Early Coinbase could not compete on heat—“the first 5 or 10 people who joined were crypto zealots.”
The best specimen was first hire Olaf Carlson-Wee, whose prior job was literally lumberjack. He arrived disheveled in “some ill-fitting suit he’d bought on the way to the interview,” having written his college thesis on Bitcoin, and beat a credentialed Google AdSense manager whose interview was low-energy.
“The guy was just superbright, superpassionate… he crushed it.” Armstrong adds that Olaf became a billionaire and went on to create a crypto venture fund. The broader bet was on people who were high-agency, smart, and got things done despite weak credentials on paper.

11. Answering the decentralization purists—and Jobs’s appliance logic

To the Reddit-era charge that a centralized Bitcoin company betrayed the point, Armstrong’s analogy is: “email is a decentralized protocol. But you can use Gmail, Outlook, or whatever.” Coinbase eventually shipped a self-custodial wallet for people who wanted to custody their own crypto.
Armstrong estimates that “something like 80% or 90%” of the world’s money is tied up in financial institutions rather than retail users. Institutions found self-custody “super scary” and wanted enterprise custody; that decision came from customer conversations, while the retail ease-of-use decision was his intuition.
Senra’s parallel—Jobs’s “appliance” insight that easier products could make the market a thousand times bigger—lands: many people may use crypto without knowing it, such as using stablecoins for remittances instead of paying 11% at Western Union, or using DeFi for a cheaper loan approved in 30 seconds.
On the “Everything Exchange” ambition, Armstrong refuses a Bezos-style master-plan retcon: “I think that would be intellectually dishonest for me to say that I knew exactly how that was going to play out… I knew it was massive, I just didn’t know exactly how it’d play out.”

12. Running Coinbase now: founder-operator pairing, bottleneck hunting, internal VC

The organizational thesis is that pairing a technical founder with a great operator—president and COO Emilie Choi—generates enterprise value. In Armstrong’s stereotype, an operator alone runs efficiently but can miss the next wave of innovation, while a founder alone can “blow the place up.” He cites Zuckerberg/Sandberg and Eric Schmidt with Larry Page and Sergey Brin as precedents.
His personal operating principle, borrowed from Elon, is: “what is the limiting factor at any given time? And I go dive deep on that.” He also supplies risk tolerance: a 20% chance at a 20x outcome “you should take that bet all day long… If it fails, it’s on me.”
The “Next Bets” system inverts corporate veto structures. Instead of five yeses up the chain, a product-group leader or another budget-holder in the room—including the CFO, Choi, Armstrong, or potentially a talented young engineer—can fund a pitch from that person’s budget.
The proof: “I am actually embarrassed to admit I voted ‘No’” on USDC; someone else funded it, and Armstrong thinks that in 2025 Coinbase generated about $800 million in revenue from it, “or something.” His standing fear is that brilliant young engineers could leave if they cannot get their ideas funded; Wozniak pitching HP on a personal computer is the cautionary example.

13. Everything is content: earnings calls, meme-format investor videos, and the Buffett gap

Against finance-team pleas to “stay on script. It’s supposed to be boring,” Armstrong reframed earnings as marketing: “We’re supposed to be selling some stock, right? Let’s go out and tell the story of the company.” That led to a filmed pitch-deck walkthrough paired by an Internet-native marketer with vertical-video gameplay whose retention, Senra says, goes “through the roof.”
“We do need to get the word out in the way that people actually consume content today,” since “99% of people aren’t going to read our shareholder letter.” Analysts and major funds may read it, but most retail investors consume podcasts, X, blogs, Substacks, and other direct content.
Senra’s challenge—Buffett’s letters as “the most successful example of content marketing in history,” and Bezos teaching philosophy through 21 years of letters—draws a genuine concession: “I hadn’t thought about it… Those guys went deep. That was very atypical. So, I’ll think about that.” Senra’s Ogilvy maxim seals it: “You can’t save souls in an empty church.”

14. Radicalized into going direct: the New York Times hit team

After the Mission First post, Armstrong says insiders later told him that The New York Times assigned a team to “go dig up dirt on this company.” He says they had the headline before finding anything and published articles implying racism and underpayment of minorities—“false information.”
His verdict is that traditional outlets are “more like political propaganda machines,” not journalism. That experience shifted him toward direct distribution: “every company is a media company now,” publishing through its own blog, social accounts, and sometimes podcast rather than relying on an intermediary’s framing.
The liberation thesis is worth quoting: “everybody at some point in their life should get The New York Times to write a hit piece on them, because you stop fearing it… Why don’t I actually just do the thing that I think is good, regardless of how people perceive it?”

15. Building a company and an industry: Wright brothers, Gandhi’s stages, and the banks flipping

Senra’s frame—Armstrong as an early automobile founder, building the company and the industry simultaneously—prompts the Gandhi progression: “First they ignore you, and they laugh at you, and they fight you… And then you win.” Armstrong says crypto is at stage three; five G-SIB banks are now working with Coinbase on crypto integrations and hiring crypto engineers. “This is like a little blip on the policy radar.”
On the Wright brothers, Senra recalls that they solved a centuries-old problem with roughly $1,500 from a bicycle shop, despite better-funded competitors, and cites a War Department dismissal with uncertainty. Armstrong positions himself carefully as commercializer, not inventor: Satoshi is the equivalent of the Wright brothers, while “I did not ‘discover’ flight.”
The Thiel caveat is that entrepreneurs must be “contrarian but right” and “willing to be misunderstood for a long time.” Armstrong’s own contribution is the instinct—or “nose”—to recognize an opportunity and commercialize it with a successful company.

16. Beyond Coinbase: NewLimit, agent wallets, and special economic zones

NewLimit began after Armstrong hosted dinners with biotech CEOs and PhDs asking what was “underfunded or underinvested in.” The answer was epigenetic reprogramming, including Yamanaka’s Nobel-recognized work reprogramming skin cells into stem cells. Armstrong committed “$100 million of my own money”; the company says it has demonstrated reprogramming human cells to restore function, and its first drug candidate will enter clinical trials “probably next year”—faster than the five-to-six years of pure research he expected.
Day-to-day NewLimit is run by Jacob Kimmel; Coinbase remains Armstrong’s full-time job, though he spent roughly 5%–10% of his time on NewLimit when it was getting started. He is primarily an investor and board member, while helping with operations and fundraising.
Inside Coinbase, AI writes “more than 50%” of code and answers about 60% of support inquiries. Internal data from Docs, Slack, GitHub, and Salesforce feeds tools including LibreChat, Glean, Slackbot, and Gemini, allowing Armstrong to ask, “What should I be more aware of as CEO?” and receive, for example, “Did you know this team is not aligned on the strategy?”
The crypto-native piece is stablecoin wallets for AI agents doing machine-to-machine payments: agents may need to buy cloud resources, pass through paywalls, purchase domains, or launch marketing programs, while traditional corporate cards cannot be issued to nonhuman entities.
Two honest experiments: the Base App’s SocialFi launch, with optional coins for posts and creators, was “kind of polarizing”; Armstrong says, “I don’t think it quite worked.” The tokenomics did not yet seem durable, and the app has since focused on self-custodial trading, though he thinks something in the social-token space will eventually work.
His next possible cause is US special economic zones on federal land—sandboxes for nuclear-reactor design, accelerated biotech trials, or drones outside traditional FAA rules—modeled on Shenzhen and other zones abroad. “I might work on that at some point.”

Cambridge PhD 相子恒 on Bringing DeepOptica’s AI to Mining Situational Awareness in Government, with UK AISI Chief Scientist Geoffrey Irving