Dialogue with Lightspark CEO David Marcus: The Era of Stablecoin Accounts Has Arrived

In the 81st episode of Tokenized, Simon Taylor, Head of Market Development at Tempo, and Cuy Sheffield, Head of Visa's Crypto Business, invited David Marcus, CEO and Co-founder of Lightspark, to discuss: how Lightspark has evolved from the Bitcoin Lightning Network to stablecoins, how to delegate fund flows to AI Agents, and more related topics.

Takeaways:

1. Lightspark's positioning has evolved from "Bitcoin Lightning payment company" to "global accounts and stablecoin infrastructure company"
   David Marcus's core statement is: Bitcoin is more like an underlying neutral network, and users don't necessarily need to perceive "I am using Bitcoin." For Lightspark, Bitcoin can become a "money data packet" similar to TCP/IP, enabling underlying interoperability between dollars, local currencies, stablecoins, card networks, and domestic payment systems.

2. The core of Grid Global Accounts is not that Lightspark creates accounts itself, but that platforms create their own accounts
   David emphasized that many companies are building their own account products and then selling them to platforms; Lightspark is providing the infrastructure that allows platforms to open accounts for creators, sellers, drivers, landlords, and other endpoint users under their own brand. This difference determines who owns customer relationships, data, and account economic benefits.

3. Platforms are realizing: payouts are not a cost center, but a financial entry point
   Creator platforms, marketplaces, and gig platforms used to just send money out, and once it reached a local bank or third-party wallet, the subsequent deposit value, consumption, credit, data, and financial service opportunities were lost. The logic of Grid is to turn the payout endpoint into their own financial account entry.

4. Stablecoin accounts need to have real usability, not just "having USDC/USDT in the wallet"
   David believes that if creators receive stablecoins but cannot spend them, transfer them to local banks, or connect to card networks, then the product's value is weak. The real unlock is: dollar balances can connect to Visa cards, can be converted to local currencies 24/7, can use domestic rails in 65 countries, and can connect to networks like Solana, Base, Tempo, and Pix.

5. The true user language of stablecoins is not "some coin," but "dollar balance"
   When discussing Western Union's issuance of USDPT, David's viewpoint was clear: users do not care about USDP, USDPT, or any branded stablecoin name; what users want is a dollar balance. Platforms can issue their own stablecoins on the backend, but it is best to express it as "dollar accounts" on the frontend, with branding reflected in the account experience rather than forcing users to remember a coin name.

6. The economic significance of having proprietary stablecoins is to reclaim reserve yield and account economic benefits
   If a platform issues its own dollar stablecoin and allows it to interoperate with USDC, Solana, Visa, Pix, and other networks, then the platform can earn reserve yields while user balances remain within its account system, rather than giving this portion of revenue to third-party stablecoin issuers or external wallets.

7. AI Agents will reshape account and payment interfaces, and CLI may become the new UI
   The program repeatedly mentioned that in the past, mobile apps were consumer UIs, APIs were developer UIs, and CLIs are becoming the UIs for Agents. In the future, Agents will need to represent users to move funds, invoke cards, execute payments, and manage accounts, but the premise is that there must be clear authorizations, limits, scopes, and controls.

8. The correct financial model for Agents is not "Agents opening accounts themselves," but "human/business accounts authorizing Agents to act"
   Cuy explicitly opposed a world of Agent wallets without responsible entities. A more reasonable model is that the ultimate customer still completes KYC, and the Agent is an authorized executor. Lightspark's bonded scope delegation sets what Agents can and cannot do, allowing multiple Agent scopes to be configured for one account.

9. Meta's stablecoin creator payouts indicate that the industry environment has changed
   A few years ago, "Meta + stablecoins + payouts" would have triggered huge regulatory and public reactions; now it has hardly stirred any waves. David believes the reasons include clearer regulatory rules, such as the US GENIUS Act and Europe's MiCA; wallet technology has matured; and stablecoin cards and consumption scenarios have become more feasible.

10. Creator platforms have the opportunity to become "creator neobanks"
    Creator platforms hold income data, content data, growth data, and payout data, making them inherently more knowledgeable about creators than traditional banks. In the future, they can not only disburse money but also provide accounts, cards, advances, credit, financial management, and even serve "one-person companies + multiple AI Agents" as new types of creator businesses.

11. AI could expand the creator economy by 100 times, thereby amplifying the demand for financial services
    Cuy's viewpoint is that AI will transform creators from "individual content producers" into "one-person companies," with multiple Agents helping to produce videos, music, images, data, and content. These Agents will also need to pay to access models, tools, and services, so creator accounts will upgrade from simple payment tools to small business financial operating systems.

12. On-chain credit and DeFi private credit may become important funding sources for platform financial services
    Simon mentioned that it is difficult for platforms to make loans in 160 markets and impossible to obtain bank licenses everywhere. However, on-chain private credit, DeFi modular funding sources, and tokenized credit may allow platforms to provide credit services to creators, drivers, and sellers without fully relying on their own balance sheets.

13. Emerging markets facing dollarization with stablecoins should respond by developing local currency stablecoin ecosystems, not by banning them
    David believes there are two regulatory responses: the good response is to acknowledge that users want to hold dollars while also allowing local currencies to have digital stablecoin versions; the bad response is for central banks to try to control everything with CBDCs. He believes that in most countries, goods and services are still priced in local currencies, and there will be real demand for local currency stablecoins.

14. Western Union's stablecoin path represents a typical transformation route for traditional remittance companies
    Western Union first uses stablecoins for internal settlements and correspondent banking between agents, then gradually does off-ramps, stablecoin accounts, and stable cards. This indicates that the landing sequence of stablecoins in traditional institutions often starts with optimizing backend settlement and liquidity before entering user-side account products.

15. Visa's stablecoin card annual on-chain settlement reached $7 billion, but it is still in the very early stages
    Cuy stated that this number is still very small relative to Visa's overall scale, but the growth direction is clear. Stablecoin-linked card projects are naturally suitable for settling with stablecoins because, compared to converting on-chain funds back to fiat for settlement, stablecoins enable faster, more continuous, and less bank holiday-affected fund flows.

16. Tokenized RWA will further compress the necessity of "idle cash"
    The program concluded with a discussion on Securitize and Computershare's collaboration to create tokenized stocks. Cuy extended this to treasury management: if tokenized money market funds and tokenized treasuries can be converted 24/7, in the future, companies and individuals will not want to hold large amounts of non-interest-bearing cash but will let AI treasury agents automatically switch between cash, stablecoins, and yield-generating assets.

17. The stablecoin yield ban may create offshore dollar account arbitrage
    Both David and Simon mentioned that if the US restricts stablecoins from paying yields to holders but allows it outside the US, then a second-order effect similar to eurodollars may occur: users may place funds in offshore dollar stablecoin accounts to earn dollar yields. This may weaken the attractiveness of the US domestic banking system for dollar deposits.

18. The main theme of the entire episode is: stablecoins are not just payment tools, but are reconstructing platform accounts, Agent payments, creator finance, remittance settlements, and global dollar accounts
    This episode is not simply discussing "using stablecoins for faster and cheaper transfers," but is talking about a larger structural change: platforms are beginning to regain control over account endpoints, AI Agents are becoming the interfaces for fund operations, and stablecoins are becoming the global balance and settlement layer, while card networks and local payment systems provide last-mile usability.

Simon Taylor:
Welcome to Tokenized, a show focused on stablecoins and the adoption of tokenized real-world assets by institutions. I am Simon Taylor, your host, author of Fintech Brainfood, and Head of Market Development at Tempo. As always, with me is my friend, colleague, and good brother, Cuy Sheffield, Head of Visa's Crypto Business. How are you doing, brother?

Cuy Sheffield:
It's an important week. A lot has happened. Today we have an absolute legend with us to help us break down one of the biggest announcements so far. So let's get started.

Simon Taylor:
Yes, I don't want to hide the focus, but yes, we have the one and only David Marcus, CEO and Co-founder of Lightspark, and of course, former CEO of PayPal and former head of payments at Meta. David, how are you?

David Marcus:
I'm good. Thank you for having me.

Simon Taylor:
Thank you very much. We'll get to your news shortly, but first, I need to make a quick disclaimer. I want to remind everyone that the views and opinions expressed by today's guests and participants may be their own and do not necessarily represent the positions of their companies. Also, please do not take anything we say as tax, legal, or financial advice. Stay safe and make sure to do your own research.

Alright, the first piece of news, let me read it. David Marcus's Lightspark is launching a banking product as an intermediary between businesses and their AI Agents called Grid Global Accounts. It has a partner network that includes FDIC-insured banks like Evolve, Cross River Bank, and Lead Bank. I believe there's also a small company called Visa involved. So I think they will do quite well one day.

David, thank you for bringing this to the forefront. But David, would you like to introduce us to Grid Global Accounts, what they do, and how they specifically help AI Agents?

David Marcus:
Yes, let me elaborate a bit because this has been in the works for a long time. You know, we have put a lot of thought and work into this. I think the key problem we really want to solve is to return power to individuals, businesses, and platforms.

I feel that all these platforms have been tenants renting someone else's payment tracks and networks. You know, especially at the account level, that's the case. For example, if you are a large platform that is a financial hub for millions or even tens of millions of people, when you send money out, once you send that money to a bank in some random country, you lose all the economic value of those deposits and everything that could happen afterward.

What Grid Global Accounts really solves is allowing these platforms to provide global accounts to all their stakeholders. So if you are a marketplace with a large number of sellers and buyers; if you are a creator platform with many creators earning income on your platform; if you are a gig economy platform like Uber or Airbnb, with landlords and drivers making money on your platform.

So, what if you could actually provide them with accounts under your own brand? These accounts would have dollar balances, a Visa debit card usable at 175 million merchants on the Visa network, and could connect to Lightspark Grid, allowing you to transfer that dollar balance in real-time, 24/7, to your own local currency bank account. Not just local payment tracks, but also global payment tracks, because Grid covers 65 countries where we have already integrated domestic payment systems.

So, that's Grid. Grid Global Accounts is a platform and product designed to allow these platforms to control their own endpoint accounts while providing their stakeholders with a much better product than what people have today. I see it as the most powerful global dollar account. I say dollar, but it will be multi-currency in the future.

And the last point is, you know, in a world where all interfaces are changing, basically all websites and apps were originally designed for clumsy human-computer interaction, right? But that's not what Agents need. So when we turn the interface towards Agents, when we communicate with Agents, and they do things on our behalf, what is the ideal account that Agents need?

It clearly needs to be an account with a dollar balance, maybe also a Bitcoin balance; it needs to be able to use Visa tokens to spend that money; it needs to be able to access global payment tracks. And how do you delegate that control while not losing control, meaning not losing human control over how Agents use their funds? That is also part of what we launched earlier this week with Grid Global Accounts.

Cuy Sheffield:
Yes, there are a lot of very interesting points to unpack here. I want to start with the evolution within Lightspark. I think you have a very interesting perspective on Bitcoin and stablecoins. You know, if I remember correctly, Lightspark started with the Lightning Network, where you were asking how to make Bitcoin more usable as a currency. Meanwhile, everyone else was saying Bitcoin is just digital gold.

Can you talk about how your views on stablecoins have evolved? And where do you see Bitcoin and stablecoins as complementary? How do you see customers wanting to interact with both now?

David Marcus:
Yes, I mean, I think the key insight is that Cuy and I have been through the Libra journey together. So we have a lot of shared scars.

I think the key lesson from that entire epic adventure is that if you really want to build an open internet currency network, and it is a closed, controlled network, then no matter how hard you try to decentralize power, it will be burned down. So the key is actually how you build things on a truly neutral, truly open network that does not belong to us, does not belong to anyone, and operates like the internet.

This is the insight derived from after Libra, and it is also the reason we decided to build on Bitcoin. There is no doubt that Bitcoin is the most neutral currency network in history. So the first step of that journey was actually trying to move Bitcoin quickly and cheaply. And Lightning was the first step.

I think, Cuy, going back to your question, I see Bitcoin both as an asset and as a technology. For us, we are really more interested in the latter, which is how to turn Bitcoin into money's TCP/IP data packets, enabling interoperability between domestic payment systems and other edge networks that can connect to this open network.

This has already taken shape in the way we provide cross-border payments for large banks like SoFi. So if you are a SoFi customer sending dollars to someone in Mexico or sending to India and having them receive local currency, what actually happens is: dollars turn into Bitcoin and then turn into rupees on the other end. And the customer doesn't even know or hear the word "Bitcoin." It is just money's TCP/IP data packet. So that is where we started.

Then, regarding how it interacts with stablecoins, I have also gone through a kind of split journey on stablecoins myself. This is also related to what happened with Libra. We were trying very hard to build on the most open, neutral network. So if all transactions actually happen on a large, centralized stablecoin, because stablecoins are completely centralized, that kind of defeats the purpose. So I was in a bit of a state: this is not right.

Then I thought, wait a minute, we have connected the fiat payment tracks of 65 countries, so why not connect all networks? Why not connect Solana, Tempo, Base, and everything, and support all stablecoins?

Then ensure that the Grid Global Accounts we actually launch are compatible with all these networks. Because if you make a balance more interchangeable, allowing it to be spent not only on the Visa network but also on Solana and in Brazil's Pix, then that account will have more utility, and more individuals and businesses will want to use that account. And this will actually strengthen the entire system.

However, that account exists on Spark, which is our Bitcoin layer two, built on top of Bitcoin. So this network is not ours, nor anyone else's. Therefore, the premise of building on something truly open and neutral still holds.

Simon Taylor:
I think this is a fascinating set of ideas, layered like Russian nesting dolls. The more you look outward, the more you can see the structure within. The Grid thing, honestly, I hadn't really realized before this conversation how powerful it is to be an off-ramp in 65 different markets; just that alone makes it a very useful one-stop product.

Then you add in financial accounts, and you see Ramp has launched a similar product, and Stripe has just launched a similar product. How many businesses want a frictionless, globally available financial account that is available 24/7? The cherry on top is that it's not just businesses that need these; they now want their Agents to do this.

So, can you talk about that last piece: how you make these accounts applicable to Agents? What have you seen in this area? This is something Cuy and I have been discussing repeatedly, but I would love to hear your perspective.

David Marcus:
Yes, so let me first make a very, very important key distinction.

Now everyone who is building is trying to create their own accounts and then sell those accounts to all these platforms, right? The companies you just mentioned are all doing this, and they are indeed very good companies that we respect.

But what we are building is essentially the infrastructure that allows these platforms to create their own accounts, accounts that carry their own brand. And this account itself is a self-custodied wallet, while also having the best technology embedded in the wallet, meaning you will never lose your funds. You can log in with a passkey, Google, or any login method you want, and you will never lose your keys.

I think this is a very important distinction because it fundamentally changes a dynamic: who owns customer relationships, and what data you will leak to other platforms, which those platforms will then use to build products based on your backend data, right? I think this is a very important distinction.

As for the Agents, what we really want to do is, you know, there is a world now where everyone is trying to build competing Agent-to-Agent payment protocols, like agentic commerce and things like that. That’s great; we will definitely need those, and something may win out at some point. But right now, that is not the most pressing issue because there isn't that much Agent-to-Agent transaction volume yet.

So the core challenge we posed to ourselves is: how do we actually create a mass-market product that is human-facing but can safely delegate the powerful fund flow capabilities of Grid Global Accounts to Agents?

The way we are doing this is by building a mechanism we call bonded scope delegation. That is, you basically bind an Agent to a Grid Global Account and then limit that delegation with explicit restrictions on what it can and cannot do. You can also create multiple Agent scopes for one account.

My own experience has been to launch an OpenClaw and connect it to a Grid Global Account. I have been playing with this account for about a month now. It is really crazy to see what an OpenClaw can do with global fund flow capabilities, a Visa card, and all these things.

Cuy, I know you have also been playing with this, and I really love the Visa CLI experience. This is indeed the direction of future development, right? All these Agents will have a CLI that connects everything, and then they can represent individuals and businesses to stitch together a fantastic experience. I think what we are trying to do here is just create an account, a very powerful account, while allowing Agents to use it safely.

Cuy Sheffield:
Yes, this ecosystem is really interesting right now. I love this example because it also responds to something I have been particularly annoyed by recently. Whenever someone says, "Oh, Agents can't have bank accounts. Agents can't have accounts." Then they assume the whole world has to run on stablecoins.

But in reality, every week you see new products emerging that are built for humans but treat Agents as delegated actors, entities that humans can introduce and authorize to act on their behalf. I think that is the right model. That is, you still have a customer who completes KYC, but if that customer wants to introduce an authorized entity to act on their behalf and has the appropriate control mechanisms, we should allow that.

I personally think this is the best outcome for consumers. At the same time, this is indeed what is needed for agentic commerce to truly develop because it must have some control mechanisms. I am actually very concerned about another world: a bunch of Agents out there with wallets, but without responsible entities and not under any human control. I think that would lead to a lot of strange and bad outcomes. So finding the right model will be very important.

But all of this starts with the willingness and ability to experiment. So for me, it is always interesting to see the difference between two types of people. One type dives into the OpenClaw rabbit hole, sets up OpenClaw, experiments, and creates products around it. The other type just talks about agentic commerce, and when you ask them, "Have you actually done a transaction?" they say, "No." Well, then it is hard to continue the conversation.

Simon, I would also love to hear your thoughts on how these parts start to fit together.

Simon Taylor:
I have been saying this week that CLI is the new UI, right? For customers, we built mobile apps and dashboards. Then fintech came along and made it a complete experience. Those disruptors created very good experiences. Then Stripe did the same thing with developer-facing APIs. And now, CLI is doing this for Agents. So you need to have a CLI, but you also have to do this with permission controls in place.

I think just two days ago, on the 28th, while we were recording this episode, the FIDO Alliance, the one that invented and promotes passkeys, announced their standards for Agent trusted delegation. So if we start to have these standards, I think we can start to secure it because Cuy, I share your concerns.

You know, I think Mercury announced their CLI a few days ago and also launched accounts for Agents. That’s great. They already have fee accounts with scopes and some permission management infrastructure. But I can see two things.

First, I go back to your point, David: do I want to own this economic benefit, or do I want to give it to the bank? For example, as a platform, do I want to own my Agent economic benefits? So I agree with that.

But I think another point is, maybe I want my Agent to operate across multiple countries and multiple bank accounts rather than being bound to a specific bank account. That’s where I think a more open mechanism, like OpenClaw, becomes very, very interesting.

On this story, I want to leave you with a thought. Yesterday at Stripe Sessions, Stripe did a fantastic launch. They launched their command line interface seven years ago. You know, at first, it just slowly developed, with some tech enthusiasts using it. Then by December 2025, suddenly, it appeared to accelerate vertically. Of course, now it is being used by Agents.

So I think the Collison brothers cheekily proposed that January 1, 2026, is the day the singularity begins, and we are now on day 120. So welcome to the singularity, everyone. And I am glad that this singularity actually includes many different people trying many different things, and I think we are still in a very early stage.

Folks, if you will forgive me, I want to take us to the next story, which is about Meta. Just right, we have you here, David, to hear your thoughts. They have launched stablecoin creator payouts, currently available to some creators in Colombia and the Philippines, using Stripe Link as the receiving wallet, and have already gone live with Tempo, Polygon, and Solana, planning to expand to 160 countries.

David, you just mentioned some scars. I have always found it interesting that this story has hardly received any coverage. Meta plus stablecoins plus payouts, if placed five years ago, would have sounded alarm bells on the street, completely crazy. Is this a sign of changing times? What do you think…

Narrator/Crowd:
Laughter.

David Marcus:
I don't know.

Simon Taylor:
Does this indicate that the position we are in now, perhaps in a post-GENIUS world, is somewhat different? Also, what do you think about the business case for giving creators payouts?

David Marcus:
Yes, I mean, first of all, I am really, really happy to see platforms and companies like Meta trying more with digital assets and stablecoins, experimenting with it, piloting it, and exploring it in various ways. I think that is great.

I have two reactions to this. First, times have indeed changed. I think when we launched Grid Global Accounts, I said the world has changed, and this change has made things like Grid Global Accounts possible, which wouldn't have been possible even a year ago.

So, to answer your earlier question about how conditions have matured, the first change is regulatory. Clearly, there is the GENIUS Act in the US, MiCA in Europe, and similar legislation around the world that actually establishes road rules for how you interact with digital assets and stablecoins. So this makes platforms like Meta and others more willing to try again, which is a good thing.

The second is better wallet technology. I think if you look at embedded wallets, they now have truly better modern cryptography that allows different parts of your keys to be distributed in different places and completely abstracts all of this from the customer. There is really no longer a need for seed phrases, no longer a need for mnemonics, and no longer a risk of losing private keys. So that is the second prerequisite.

The third prerequisite, and Cuy, you should really get a lot of credit for this, I would even say it when you are not around, is that Visa is really actively investing in stablecoin-supported cards. This is a key point, right? Because if you are a platform actually sending USDC or USDT to a random wallet for a creator in a certain country, but they cannot really spend that balance and it is stuck in the wallet, well, good luck, right? No one wants that. So I think this is a huge unlock.

So I think these three things have changed the conditions to make such products truly possible today.

Then, specifically to the question here, I think there is indeed a real question for these platforms. The question is: do you want to create an ecosystem where your creators operate on your own accounts? Or do you want third parties to create accounts for them and then cross-sell a bunch of things to them, while you never see the economic benefits?

I absolutely believe that most platforms will want the former.

So while I think it is great that Meta and other platforms are trying this, I would strongly encourage them to take control of their destiny and create these accounts for their own endpoints. Coincidentally, starting this week, we happen to have a very good product that can help them do that.

Narrator/Crowd:
Laughter.

Cuy Sheffield:
Looking back at the early days of Libra, it is really interesting. You know, I am also grateful that Libra pushed the entire industry forward at such a fast pace. Without Libra, I think we might have had to wait another five years to get to where we are today.

But the infrastructure did not exist back then. You pushed it forward, and I appreciate that, but the infrastructure really did not exist. So if you take the same use case, I remember discussing creator payouts back in 2018 and 2019. This was a very obvious problem that needed to be solved, and the concept of stablecoins seemed to be the solution.

But back then, the problem was: well, the blockchain wasn't fast enough or cheap enough, so you had to build a new blockchain. There were no stablecoins that could scale, so you had to build a new stablecoin. There were no real embedded wallets, so you had to build the wallet infrastructure for Diem from scratch. You had to build all these core components from the ground up because they simply did not exist at that time.

We did not have any stablecoin cards in 2018. I remember when we first discussed it, we said we would love to work with Diem and Libra to do cards. So to really get to this use case, a lot of work had to be put in to build those core primitives.

And now it is interesting that Meta can say, well, the same use case, but they basically don't need to build anything themselves; they can achieve it through integrating third parties.

I think what is really interesting is how deep this integration will go. When you are in pilot mode, like testing a few markets to see if this is one of the multiple payout options, that is one way to look at creator payouts.

Another way to look at it is that we want to build a huge new bank for creators. This could become an entirely new business. You could say, we have so much money flowing to creators, why don't we monetize it and provide them with every service they need? Maybe they need loans, maybe they need all sorts of things. If you can control that experience, you can embed those things.

So in my view, at some point, the idea of "creator platforms adding stablecoin payouts" ideally should no longer be news but just become a given. Yes, this is what everyone will do. But how deeply they do it will become an interesting factor that distinguishes one creator platform from another.

Simon Taylor:
Yes, this reminds me of the early days of embedded finance. At that time, every platform was trying to embed a card, but it never really took off because these platforms were global. So I think you need the global properties and acceptability of stablecoins, while not needing the bank pipeline infrastructure, industrial loan company licenses, and all the accompanying things. This is a very different model, and I think it is very powerful.

I can really imagine that if you are a very risk-averse person, starting with something like Stripe Link, which is completely owned and operated by someone else, is a good way to test the waters and see if anyone will chop your head off. Then you can always continue from there. I just throw this idea out there.

But I think what you just said resonates perfectly.

David Marcus:
I think that is absolutely correct. I feel that when you think about this, if you are a creator platform or a gig economy worker platform, right? No matter what platform it is. You are the center of the economic life of all these endpoints.

And you have the data, right? So going back to your earlier point about "the new bank box for creators or drivers or landlords," for those creators, drivers, landlords, or whatever roles on your network endpoints, you have the best data, so you should be in a position to offer customers prepayments, credit, and all these services because you have the best data.

Leaking that data to edge-positioned banks, leaking it to various third parties, before AI, that might just be the cost of doing business. But in an AI world, you can actually mine that