Henri Stern (Privy/Stripe): "Use Cases I'd Mapped Out Over a Decade Materialized in Under a Year"

Privy was acquired by Stripe in June 2025. What does this acquisition let you do that you couldn't as an independent startup?

Three things, essentially. The first — and probably the most important — is that we're entering an era where stablecoins and digital assets are becoming a strategic pillar for a lot of companies. Our client profile is shifting. We're moving from deeply crypto-native developers to people who care about outcomes for the end user and don't necessarily worry about the underlying rails. Technically, that means going from raw RPC calls to what we call "wallet actions" — much more abstracted features that give direct access to value. In concrete terms, Bridge brings asset orchestration and issuance, which lets us integrate new stablecoins into Privy and smooth out the fiat-to-crypto transitions through a regulated entity. Stripe, for its part, opens up every fiat rail you can think of. We're evolving from a basic building block — which we still are and still sell — into a far more powerful integrated stack, while remaining an open and modular system. That was actually my biggest fear going into the acquisition: that by verticalizing the stack too aggressively, we'd destroy the value of open systems. The strategy is twofold: a turnkey stack for stablecoin-rail payments, but also a modular approach for those who want to assemble the pieces themselves.

The second thing is access to a global client base. We worked with Stripe and Bridge on Meta's creator payouts. These are very different counterparts from our usual clients. Privy used to talk to fintechs, trading firms, crypto-native players. Stripe talks to virtually every merchant on the planet, to marketplaces. People who will benefit from these rails without treating them as an end in themselves.

And the third, more mundane: access to support functions — legal, corporate, technical infrastructure — that we simply didn't have at that scale.

"The wallet needs to adapt to the audience, not the other way around"

Your bet is that the wallet should be invisible?

My bet is that the wallet should be as invisible as the user expects it to be. The distinction matters. For most people, yes, it should be invisible. But we also serve Uniswap; we serve Hyperliquid — users who know what a wallet is and how to use one. The industry has been a little too paternalistic, saying: "Here's a wallet, this is how it works for everyone, and if you don't like it, don't use this rail." The result? Most people didn't use that rail. If we want this file format for money to go mainstream across the internet, the wallet has to be flexible. Klarna's needs look nothing like Deel's — we just announced a partnership with them — which in turn look nothing like Hyperliquid's. The idea that one solution fits all is completely insane.

A developer walks up and says: "You've rebuilt the bank, except it's Stripe controlling access to my assets." What do you say?

I have a few issues with the premise. But the real question is: how do you design self-custody inside a system like this? The short answer: we give users every means to recreate and manage their keys outside of any system, if they choose to. We allow key export. We were, I believe, the first wallet provider to offer key export from day one. If AWS goes down, if the whole system collapses, the user must be able to recover their assets.

There's a trade-off between convenience and unilateral control. Phantom, MetaMask — these are fantastic tools, I use them a lot, but they're not for everyone. We offer several configurations: self-custody, custody through a third party, or direct custody by the client. In a self-custodial setup, only the end user has access to their keys, but our servers orchestrate the key splitting and recombination. The user can always exit that configuration.

Where it gets interesting is the question of censorship…

Exactly. We've integrated a provider called Blockaid for transaction screening, so users don't unknowingly sign a transaction that would drain their wallet. It's a security feature people really want. When I send a wire to the wrong person, I'm glad I can call my bank and block the transfer. Chargebacks, holds — they're both bugs and features of traditional finance. But looked at from another angle, that same screening capability is censorship. An intermediary telling the user: "You can't make this transaction." One person's censorship is another's security. Our role is to be modular enough that the developer builds the right product for their user.

"The standalone embedded-wallet market isn't dead — it's evolving"

Your main direct competitors are Dynamic (Fireblocks) and Coinbase's CDP. Is the independent embedded-wallet market disappearing?

In a word, no. Competition is good for developers. We know the Dynamic team well, the people working on the Coinbase system. We work alongside each other and compete head-to-head, and it makes every product better. We weren't even first to this market, by the way. There were solutions like Web3Auth and Magic. We tried them as both consumers and developers and found things we could improve. That's partly why we built Privy. But the landscape is shifting fast. The fact that companies like ours choose to integrate into larger institutions simply reflects a reality: developers want a complete package. And Stripe's commitment to building a modular stack where you can pick your own pieces — that's the best of both worlds. It was a central focus of my due diligence before joining Stripe.

What were the inflection points in Privy's history?

There were three chapters. The first was Friend.tech — a SocialFi product that let you create a self-custodial wallet inside the app and trade with your friends. At the time, there was almost no crypto on mobile. It sounds archaic now, and the experience was frankly rough, but it was revolutionary: a Progressive Web App with push notifications that behaved like a native app. They pulled off some really clever things and changed how people thought about consumer crypto products. We were lucky to work with them. We went from 30 clients to 300 within a few quarters. Around the same time, we were working very closely with Base, with the rail and the account system co-evolving. That's a period where our traffic scaled 40x week over week. It forged both the company and the technology.

The second moment was a string of major clients we signed — OpenSea and others. We shifted from a very developer-centric positioning to a more enterprise dimension.

And the third, roughly a year and a half ago, when we saw everything happening around stablecoins. We invested heavily in our systems to become the best platform for building with stablecoins and digital assets at scale, beyond crypto. We signed some very large institutional fintechs — some I still can't talk about.

"Crypto infrastructure lags stablecoins in valuations. That's the market"

Stripe acquired Bridge for $1.1 billion; Mastercard acquired BVNK for $1.8 billion. Stablecoin payment use cases are valued at over a billion. Infrastructure seems to trail behind. Why?

Good question. Is it fair? The market is what it is. I don't think in those terms. I think in adoption. Stablecoins are a kind of plumbing for monetary rails on the internet — a file format for moving money more easily. The question is: what are the most useful first use cases? Clearly, it's cross-border payments, it's the global fintech play. The orchestration work Bridge does is extremely hard and extremely valuable. Bridge is a regulated financial institution. We are, in many ways, a cryptography software company. Most users start with cross-border. But once you've started, the question becomes: what can I do next? And that's where accounts, wallets become the central node from which you connect rewards, DeFi, card issuance, and so on. These are deeply complementary stacks.

Today, Stripe's stack is significantly more complete with Bridge, Privy, and Tempo…

Absolutely. The thing that's been really cool to watch is the consistency of Stripe's commitment. There are a lot of fair-weather fans in crypto — people who enter and exit the market based on asset prices. We spent a lot of time before the acquisition understanding Stripe's real motivations. And over the past year, despite market fluctuations, the energy and commitment we see from Stripe — externally with our clients and internally with the teams — has only accelerated. That steadiness is extremely refreshing.

Do current market conditions affect your roadmap?

Not really. If anything, it makes things easier. There are fewer distractions. Every time there's a lot of hype in an industry that isn't crypto, I'm grateful for it, because crypto has historically been too fixated on speculation at the expense of building. I'm very excited about the current conditions for building.

"35% of new developers on Privy come for agentic wallets"

AI agents managing money autonomously: what does that actually look like in production, and what's the hardest problem you haven't solved yet?

This is where the bets compound. We entered through the lens of asset management and stablecoins, but Stripe is extraordinarily bullish on agentic commerce. I think it's reasonable to say that agents will account for the majority of monetary flows on the internet within a few years. There's a lot of work underway: SPTs (Shared Payment Tokens) — payment tokens you get through your LLM to pay third-party merchants via Stripe. There's also MPP, the Machine Payments Protocol, and more broadly the support we have for machine-to-machine protocols, so every Stripe merchant can receive micropayments in stablecoins or fiat.

On the Privy side, roughly 35% of new developers signing up today are coming to build agentic wallets and agentic commerce features. What's interesting is that all the work we did for financial institutions — wallet controls, key quorums, policy engines — has become extremely relevant for agents. Because you don't want your agent to hallucinate and blow up your portfolio.

We work with companies like Bankr and Merit that are building agentic wallets on our platform. The models vary: some run a fleet of 100 agents sharing the same capital pool, each agent with a daily allocation; others create one wallet per agent. The constants are: simple funding to distribute capital to agents, strict controls to cap their spending and define where they can spend, and real-time visibility into what the agent is doing.

The biggest challenge remains twofold. First, the interface between the human and their agent. How do you move from a fully autonomous world to one where the human manages their money, and back? How can the agent interact with the human, request more funds? It's a really interesting problem space, and Stripe Link is doing a lot of work on that layer. Then there's agent authentication. And a third topic: service discovery. That's exactly what Stripe Projects does, announced at Sessions — a system that lets you automatically discover and provision third-party services within an LLM context. The most annoying problem in vibe coding is that you're almost done, the agent tells you it works, and then you're missing the API key. Projects solves that. Privy is already on Projects, alongside PostHog, Vercel, and others.

What's the realistic timeline for fully autonomous finance? When will an agent be able to execute a million-dollar transaction without human confirmation?

I don't think that will ever happen in absolute terms — in the sense that you'd hand your agent open access to your bank account with no instructions and hope for the best. But pattern recognition, full handling of predictable and recurring workflows — that, I think, will happen within the next two to three years. I can very easily picture a world where I have a vendor I pay a million dollars a year, it's a recurring transaction with high visibility, I've already pre-approved it, and I no longer need to intervene. The devil is in the details. You have to educate users, set up guardrails, and all of that remains complex. But that's precisely where simplicity of interfaces, abstraction, and intelligibility for the end user matter so much. And it's also why the integration at Stripe is so valuable: I can think about service discovery and API keys just as much as wallet provisioning and how to put fiat in the agent's hands, because Stripe is looking at the entire stack.

"I sleep like a baby — and asset prices help"

You sold to Stripe. You didn't IPO, you didn't raise another round. In hindsight, was it the right call?

I think about it across three dimensions. The first is financial value, and I'll just say that I'm a very satisfied Stripe shareholder. The second is impact. The speed at which we can do the things we want to do is phenomenal. Stripe is working with Meta on creator payouts. These are use cases I would have mapped out on a ten-year horizon, and they materialized in under a year since the acquisition. I'm not taking credit — I'm simply saying that the volume of things we can do and the pace at which we do them in this setup are extraordinary. The third is our ability to move the needle in the industry. And this is where I have a contrarian take: I think this is a much better time to be working in crypto than in AI. There's so much opportunity for what we're building, and relatively so little competition. Our ability to shape these rails in meaningful ways is very high. So yes, I sleep like a baby. And asset prices help, too. But above all, we're very clear-eyed about the fact that we're still at the very beginning. That was my biggest worry: would Privy's intensity drop after the acquisition? It's been the opposite. The ambition and appetite we found on the other side energized us.

Coinbase and Kraken come from an exchange background, but they're expanding into payments, infrastructure, yield. How do you position them?

My position — whether from Privy's standpoint or Stripe's — is that one minute spent focusing on the customer is worth a hundred minutes spent watching the competition. Otherwise, you end up in echo loops where you're mimicking people who are mimicking you, with no real customer signal. You have to put blinders on and focus on user signal. That's the hardest part of a startup: getting real signals. And at Stripe, we have so much to do for our users.

And the reality is, these systems are very dynamic. We work very closely with Coinbase. Privy works with Base. Stripe collaborates with Coinbase in multiple ways, as does Bridge. And we serve Kraken — Kraken Earn is built on Privy. In all these cases, it's not a binary question of competition or partnership. There are places where we compete and many more where we're partners. As long as you stay focused on the user, it works.

>> John Egan (Stripe): "We want to build a global network for payments and treasury"