Neha Narula started her career building databases — fast, scalable, multi-core systems designed to handle the cloud infrastructure of companies like Google, Microsoft, and Meta. The work was intellectually interesting. It was also, as she eventually concluded, being used mostly to keep people scrolling and clicking on ads. That dissatisfaction sent her looking for harder problems.
She found one in Bitcoin. Not the ideology, and not the investment thesis, but the engineering puzzle: why is a payment system so difficult to scale? The question pulled her into the mechanics of how money actually works — central banks, fractional reserve banking, the basic plumbing of the global economy — and she emerged with a view that has shaped her work ever since. Money is not a natural fact. It is a system built on design choices, and those choices encode political decisions about who gets access, on what terms, and at whose expense.
That view now sits at the centre of some of the most consequential debates in finance. As Director of the Digital Currency Initiative at MIT’s Media Lab, Neha led Project Hamilton, the research collaboration with the Federal Reserve Bank of Boston that laid the technical groundwork for a potential digital dollar. She has written on the hidden risks in stablecoin infrastructure and on what it would actually take for Bitcoin to become quantum-resistant. She sits on the board of Block and advises the Federal Reserve Bank of New York.
In this conversation, Neha explains why CBDCs and stablecoins are less different than most people assume, what Project Hamilton set out to do and what it left unresolved, why zero-knowledge proofs are not a privacy solution on their own, and what a decentralized community like Bitcoin can realistically do to prepare for quantum computing. She also takes the geopolitical temperature: what Iran’s reported willingness to accept Bitcoin payments actually signals, and why, in her view, the dollar remains the game to lose.
Her website is at nehanarula.org. The Digital Currency Initiative is at dci.mit.edu. She is on X at @neha and on LinkedIn.
Marieke: Hi! I’m Marieke Flament.
Nicolas: And I’m Nicolas Colin.
Marieke: And this is the Currency of Power podcast, the companion to the Currency of Power newsletter, where we dig into currencies, power, and the people connecting the two.
Today’s guest is uniquely positioned to bridge deep technical engineering, public policy, and mainstream finance when it comes to crypto and digital assets. Neha Narula started as a senior software engineer at Google and helped relaunch Dig before completing her PhD at MIT, where she built Doppel, a fast, scalable, multi-core database. She currently serves as Director of the Digital Currency Initiative at the MIT Media Lab, a board member at Block — formerly Square — and a member of the Federal Reserve Bank of New York’s Innovation Advisory Council.
Today’s conversation runs in four parts. The first two are for all listeners; the last two for paying members. Thank you to all members for following us, and to those paying, thank you. Part one covers Neha’s career and how she entered the digital asset space. Part two covers stablecoins, Project Hamilton, and why the US does not want a CBDC — or does it — and what that means for other countries. Part three covers privacy and quantum. Part four covers the geopolitical outlook.
Neha, you and I recently met at an ING Orange event in Amsterdam. We shared a ride and couldn’t stop talking about stablecoins and digital assets. I’m delighted to have you here. Thank you for making the time.
Neha: Thank you. It’s wonderful to be with you both, and I’m excited to continue the conversation we started there.
“The success of platforms like Bitcoin has done a great deal to rethink how to build decentralized systems.”
Marieke: Let’s get started. We love understanding people’s journeys — what got them into what they’re doing. Walk us through yours, and in particular the moment when you realized that distributed systems could do something important for how money works.
Neha: I was doing my PhD at MIT in computer science, working on databases and distributed systems — nothing to do with cryptocurrency. I finished around 2015. Bitcoin was launched in 2009. I’d heard of it around 2010 or 2011 but hadn’t paid much attention. A couple of friends were very enthusiastic about it, but I wasn’t convinced. The proof-of-work mechanism seemed odd.
By 2015, I had finished my PhD and reached a moment of real doubt about what I was doing. I had been working on very fast, scalable databases — systems that run the cloud for large tech companies like Microsoft, Meta, Oracle, and Google. But at the time, that work was being used mostly for advertising and for products married to advertising. There was a lot of effort going into keeping people scrolling and clicking, showing them more ads. It was deeply unsatisfying. I wasn’t sure the work I found intellectually interesting was being used to solve real problems.
I took some time off and finally had the chance to read the news. I kept seeing articles about Bitcoin having trouble scaling. 2015 was the height of what we call the “Blocksize wars” — a large debate about how to get Bitcoin to scale, reduce fees, and increase users. I read those articles and thought: what is the problem here? Bitcoin is a payment system. Payment systems are easy to scale. What could possibly be hard? That is how I got — we like to say — nerd-sniped into the cryptocurrency world.
I fell down the rabbit hole. I learned how Bitcoin worked, but more importantly, I learned how money worked, and I had no idea. I learned about central banks, commercial banking, fractional reserve banking, interest rates, the basic infrastructure of the economy. And I thought: this is so different from anything I had imagined. There are a lot of people left behind — people who pay very high fees to access financial infrastructure, people who carry debt their entire lives while others have theirs forgiven. This happens not just at the individual level but at the level of nation states.
These parameters are built into our financial system, but they don’t have to be this way. Cryptocurrencies seemed like the best tool for change, because changing things from within the financial system seemed far too hard. Cryptocurrencies offered a whole new way of thinking about structure — very much parallel to the architecture of the internet, with packets of information and protocols layered on top of one another. And the underlying technology — distributed consensus, peer-to-peer networks — was directly related to the problems I had been working on. It felt genuinely exciting. That is how I got into it.
Nicolas: Would you say that software architecture has already been changed by the lessons learned in crypto?
Neha: Yes, I think that is fair to say. The idea of decentralisation existed before crypto, but the success of platforms like Bitcoin, Ethereum, and stablecoins has done a great deal to rethink how to build decentralized systems. Before Bitcoin and Ethereum, the most well-known decentralized system was BitTorrent — a way of sharing files over the internet without a trusted third party. In 2006, BitTorrent was the majority of traffic on the internet, used primarily for music and films, but also for distributing software and other files.
BitTorrent was a precursor to cryptocurrencies, but crypto has taken that technology well beyond it. And some ideas from cryptocurrency have been filtering back into traditional finance — the idea of digital wallets, more open access, real-time 24/7 payments and settlement.
Marieke: You gave a TED Talk almost exactly ten years ago that now has 2.6 million views. You outlined the programmability of money and what it was going to change. It feels like it took ten years for everyone to arrive at the point you described. What is your view for the next ten years? And when it comes to programmable money, what changes in power are you starting to see in the financial system?
Neha: I stand by that talk, which is remarkable considering it was ten years ago. Back then, everyone was confused about cryptocurrencies. People would start with SHA-256 and miners and electricity and proof of work, and just confuse their audience entirely. The framework wasn’t there. I was trying to separate what would be true in five to ten years from the implementation details that, while important, were not the critical thing to understand.
Going forward, we are still capitalising on programmable money. That is still in progress, but we are seeing a great deal happen. And you have to make AI a central part of any ten-year outlook. People are very excited about AI agents — autonomous software that needs to transact economically, make payments sometimes on behalf of users, but sometimes at a scale and speed at which the user will not be in the loop. Are stablecoins or cryptocurrencies the best tool for that? There are interesting arguments on both sides. One of the actors doing the most in this area is Visa, a very traditional company, which released a command-line interface for AI agent payments. So there is movement there.
But it is not just how AI agents will make payments. It is how AI as a tool will change the way we think about money and how money flows. One example: monetary policy today involves a great deal of data and modelling, but ultimately it is humans deciding. I think it is inevitable that this will become more and more algorithmic — more automated, with perhaps a human in the loop but not at the centre. Another example: most money today sits on the balance sheets of commercial banks. What is the next step? Will stablecoins or new types of fintechs create money? What will the creation of money look like?
Marieke: It sounds as though, if anyone at TED is listening, they should invite you to give the next ten-year talk.
Neha: I would love that.
“Looking back, I think I underestimated how political money really is”
Nicolas: Let’s move to Project Hamilton. I imagine you worked on it before Trump returned to power, because CBDCs are now out of favor with the current administration. And the name — is it Hamilton because Alexander Hamilton was the founding father of the US financial system, and the project was about reinventing or experimenting with that?
Neha: Yes, and it is actually a lovely story. The name did come in part from Alexander Hamilton, but there was another Hamilton the project was named for as well — Margaret Hamilton, a software engineer during the Apollo missions who was critical in writing much of the software for those missions. There is a famous photograph of her standing next to a stack of books representing all the code her team wrote. Jim Cunha, who was a leader at the Boston Fed at the time and pushed for the collaboration, chose the name for both reasons — technology and the person who helped architect the financial system. It is a wonderful coincidence that it is the same name.
At no point had a decision been made to actually build a digital dollar and launch it. This was about doing the research necessary to form a foundation and understand what a digital dollar could be, how it would work, before even deciding whether to explore it as a serious option. Technology sets the rules of any system — what it can do, what is even possible — so we wanted to build that understanding.
Our initial goal was around scalability. Something at the scale of a digital dollar, operating in an economy the size of the United States and potentially globally, would need to handle hundreds of millions or billions of users with very high throughput and very low latency. That was a good match for my expertise. The Boston Fed knew about our work in cryptocurrencies and cybersecurity, and so we collaborated. It was a wonderful partnership — we continued to work with some of those people even after the project ended.
But the work was just the tip of the iceberg. We started with scalability and security. We did some work on programmability. We did not get to touch privacy fully, though we did design the initial system so that it did not store personally identifying information, did not store balances in clear text, and did not store user addresses or pseudonyms in clear text. That was rooted in the idea that central banks do not want a honeypot of sensitive information. They want to focus on building a currency that fulfils their mandates — in the United States, employment and price stability. They did not want to know all the payments users were making.
We kept the work very technical — distributed systems, software architecture, published papers. We tried not to get into policy because that was not our domain. That was the job of the Federal Reserve Board of Governors, Congress, and the executive branch. But looking back, I think I underestimated how political money really is. Even clearly technical work that is, I believe, quite neutral is very hard to divorce from people’s feelings about money — their worries, their fears, their incentives, political and otherwise.
There are real fears around surveillance. People are worried about the government, their employer, friends, family, and coworkers potentially seeing their financial transactions, which reveal a great deal about who we are. But people are not just worried about data leaking. They are worried about control.
There is a conversation happening around the world about what I want to call restricted money — as distinct from programmable money. The idea that government-issued money could restrict how it is used. SNAP benefits in the United States are a good example: payments for food that can only be used on certain things. But we do not call SNAP money. We think of it as a coupon or a voucher. When you start to mix the idea of restricting money with the concept of money itself, that becomes very worrisome. There were fears that a CBDC would prevent you from buying certain things, donating to certain political parties, or that spending data would be used to deny visas. I never want to build a system that can be used in that manner. The system we built cannot, I believe, be used that way. But I understand the concerns.
The 2024 election was highly charged. The idea of a CBDC immediately fell along partisan lines. We now have an executive order from Trump rescinding Biden’s executive order on digital assets, and an anti-CBDC Surveillance Act that has passed the House. There is currently a moratorium on this type of research. For what it is worth, I do not know whether it is a good idea for the United States to launch a digital currency. I was clear about that throughout — I was not advocating for it. But I think it is a serious mistake to suppress research. That starts to encroach on questions of free speech and academic freedom. It is important to be able to study things.
Marieke: Is your reading of the Anti-CBDC Surveillance Act that it blocks both wholesale and retail CBDC? And what is the actual US position, given that other countries are still testing and embracing CBDCs?
Neha: I have not read the act recently, so I may be slightly wrong, but I believe it is primarily focused on retail. Wholesale CBDC concerns them far less. The Federal Reserve Bank of New York is still working with the BIS on some wholesale CBDC concepts. A wholesale CBDC would only be used by large financial institutions — and the question then becomes why it is better than FedNow or Fedwire. You could add programmability, atomic settlement, and so on. But then why can you not do that with APIs on top of FedNow or Fedwire? These are genuinely interesting questions.
To answer your direct question: it was primarily about retail, and some wholesale CBDC research is still ongoing. However, the act was written quite broadly. It essentially said the Federal Reserve could not issue a digital version of the dollar, and I think even existing central bank reserves could fall under that definition. It needs to be sorted out before anything is implemented.
Nicolas: Because money is largely digital already. The concern seems to be about preventing a permanent connection between the dollar you use in daily life and the central bank. How are you looking at the geopolitical landscape, with different countries going in different directions?
Neha: We work across all those categories — CBDC, stablecoins, tokenized deposits — and I think they are more connected than people realize. A lot of people see CBDC versus stablecoins as a binary. I do not think that is true at all. One way of viewing a stablecoin is as a wrapped CBDC. Stablecoins currently wrap Treasuries in the United States, and to some extent commercial bank accounts and money market funds, which ultimately wrap Treasuries.
A lot of stablecoin providers would love access to central bank balance sheets. In the UK, they are actually thinking about that — stablecoins backed by the central bank balance sheet. That is a wholesale CBDC. You are broadening access to the central bank balance sheet, and once you wrap that in a stablecoin, it becomes a retail instrument issued by the private sector instead of the central bank. That is essentially a two-tier CBDC, which is exactly the CBDC architecture where banks issue the tokens.
The whole point is that we are tokenising money — turning it into this packetized thing. Whether it is issued by a central bank or the private sector, what matters are the design choices: what is it backed by, what are the legal agreements behind that backing, what happens in bankruptcy, can users access something of value when institutions fail? From a technology standpoint, a CBDC and a stablecoin are the same idea with different choices for those parameters.
I do not want to minimise the differences between something issued by a central bank and something issued by a private actor — the legal frameworks are quite different. But at a high level, they are the same thing. People do not quite understand how connected they are, or how design choices can blend them together.
Nicolas: Can you give an accessible explanation of the difference between wrapping Treasuries, as stablecoins do under the GENIUS Act, and having access to the central bank balance sheet? Is it fundamentally about power — who governs the issuance and use of money?
Neha: Yes, exactly. If it is backed by Treasuries, it is issued by the Treasury, operating in concert with the Federal Reserve. Treasuries are managed through the New York Fed’s Market Desk. But the key question is: who decides how much is issued into the system? And most importantly, what happens when things fail? That is what gives money its value — the trust that you will have something of worth at the end of the day if institutions fail. That is the bankruptcy question. Can consumers and users get at something still valuable when that happens?
Nicolas: So we are prisoners of current categories. If you had to reinvent the wording, how would you describe the space?
Neha: CBDC has become a dirty word, so I would throw that one out. I was annoyed for a while by the conversation about tokenized versus account-based systems, but I am coming around on it — tokenized money versus non-tokenized money is a meaningful distinction. I think the more fundamental question is: how decentralized is the substrate? Does the system have the property of censorship resistance? Cash has that property. Tokenized money is supposed to have it, but if it is issued by Circle or Tether, it does not. Tether just froze 500 million dollars’ worth of USDT. So that is the real question. I am not sure I have the right vocabulary yet. It would take months, if not years, to really sort that out properly.
Marieke: This is very interesting, because Nicolas and I have been working with a concept of a tokenized money stack. I love the word you use — wrap — rather than stack. In a stack it feels layered. But it is actually that you wrap one thing within another. And ultimately, as you say, it comes back to bankruptcy: who is the lender of last resort?
You recently wrote a paper on the hidden plumbing of stablecoins. Walk us through what it found, and what could go wrong — especially in an international system where parties are operating under different rules.
Neha: The interoperability question is central. Within one blockchain ecosystem — say, the Ethereum ecosystem, using the EVM and ERC-20 tokens — you get interoperability almost for free. That is the whole point. You can use any ERC-20 token in Aave or Compound. It is the concept of money Legos.
It becomes much harder across different blockchain systems. Ethereum to Solana, for example, is traditionally done with bridges. Bridges have been a major source of hacks because they rely on multisig or trusted actors managing the movement of money. Anyone can start a bridge, and there have been many. But interoperability between chains is not solved.
In the traditional finance world, large institutions — the BIS, CLS, Swift — come together to manage interoperability through messaging standards and FX holdings. They have projects they are starting to productionize. It will be interesting to see how that develops.
As for the hidden plumbing paper: everyone was focused on reserve backing. The GENIUS Act is primarily focused on backing — are the reserves there, how are they accessed? But we thought there were many other risks that people were not thinking about, starting with the underlying blockchain infrastructure. Circle and Tether issue on dozens of different chains. The largest chains by volume are Tron and Binance Smart Chain, which most people have not heard of — not what you would call blue-chip chains like Ethereum. These are chains with fewer validators, less decentralisation, shorter track records, and weaker security. Blockchains are not a universal, well-understood substrate. They are all quite different.
Then there is the question of how stablecoin issuance actually works. Not everyone can redeem directly with a stablecoin issuer. Most people have to sell through a cryptocurrency exchange, which introduces additional risk for maintaining par value. When Silicon Valley Bank failed and Circle had three billion dollars there, Circle resolved all redemptions at one dollar — for the institutions allowed to redeem directly with Circle. But ordinary users had to sell USDC on a cryptocurrency exchange, and on exchanges it broke par. That is an important structural risk that is not widely understood.
The third element is the structure of the Treasury market itself. Even if you are redeeming with a stablecoin issuer, can they actually sell the underlying Treasuries quickly enough to give you commercial bank account money? There are weaknesses there. I am not sure these risks are understood by the people writing or voting on the laws.
Nicolas: What is the systemic reason why ordinary users cannot redeem directly with issuers?
Neha: From the issuer’s perspective, it is overhead. They have to KYC everyone, maintain business relationships, manage all those accounts. Circle typically only redeems with a set of institutional customers; Tether with an even smaller set. Interestingly, under MiCA in Europe, stablecoin issuers are required to offer redemption to all customers. I am not sure how that is playing out in practice, but it is a significant development.
Marieke: On the Clarity Act and the question of yield on stablecoins — within this stack there are many actors participating in the stablecoin ecosystem, and the yield can be distributed by different companies at different levels. Are banks right to be worried about stablecoins?
Neha: Banks are right to be worried broadly. Coinbase would love to have JP Morgan’s business — that is very clear. But should they specifically be worried about yield? I am not 100% sure the arguments they are presenting are entirely valid. I am not sure stablecoins take money out of the banking system in a material way. They may take customers from the banking system, which is different. This is really a question about distribution and who owns the relationship with the end customer. Smaller and mid-tier banks are most concerned about interchange fees. If a large share of payments move to stablecoins, they lose on interchange. Yield is the first thing people have identified, but there are other aspects of their business that will be eaten away. Yield is essentially a customer acquisition tool — the incentive that draws people in and determines where they go.
