The audio transcription is done by GPT and may contain errors. Please listen to the complete podcast:
YouTube: https://youtu.be/_7yNaO-jxJU
Spotify: https://open.spotify.com/episode/4kD76ba6JoxVkOvLyqauDD?si=_SdjshxHRhWslB96eFCavA
Introduction to Zama and CEO Rand Hindi’s Background
Ehan: Welcome to WuBlockchain Podcast. For the latest blockchain and crypto insights, subscribe to WuBlockchain on YouTube and follow us on Twitter. Stay informed and join the conversation.
Today we’re excited to be joined by Dr. Rand Hindi, CEO of Zama. Welcome. Could you briefly introduce yourself and the journey that led you to build Zama?
Rand: Hi, everyone. My name is Rand. I’m the CEO at Zama. I’ve been a developer since I was 10 years old, built my first company as a teenager in the ’90s, then did a PhD in artificial intelligence when I was 21, over 20 years ago. I then built one of the first AI companies in Europe, which was acquired by Sonos in 2019. Since 2020, I’ve been building Zama with my co-founder Pascal Paillier, who’s one of the inventors of FHE (Fully Homomorphic Encryption). I’ve also been in crypto since 2013, so I’ve seen four cycles so far. Additionally, I’m an investor in over 100 companies, spanning biotech, AI, and crypto.
Ehan: Was there a fundamental moment when you realized privacy would become the defining constraint for AI and blockchains?
Rand: Well, you know, if I think about how I got interested in privacy, my first company, which I built as a teenager, was a social media network. At the time, I remember seeing all the data coming into the servers and thinking that this wasn’t right. Just because I was the founder of this company didn’t mean I should be able to see all this data. When I did my PhD, I specifically worked on AI applied to genetics. Again, I was manipulating medical data, like DNA, and I thought we clearly needed privacy for AI if it was going to become something big. And finally, once I started working on blockchain, I always found it incredibly weird that in the real world, you would never show your bank accounts to your neighbor. But somehow, in blockchain, we’ve been told that this was normal. It shouldn’t be normal. So for me, privacy isn’t something specific to AI or blockchain. I actually think it’s something we need in every digital product that we end up using.
Explanation of FHE vs. Other Privacy Technologies
Ehan: For those new to the topic, Fully Homomorphic Encryption (FHE) is fundamentally different from ZK and MPC solutions.
Rand: In blockchain privacy today, there are four main technologies: FHE, MPC, ZK, and TEEs. Among them, FHE is the only one that is simultaneously secure, composable, and publicly verifiable.
FHE is secure even against future quantum computers, meaning data encrypted on-chain today will remain safe. By contrast, TEEs have recently been shown to be breakable due to flawed implementations — Intel’s own research confirmed this and even stated that decentralized protocols fall outside their security scope.
MPC is excellent for key management, and we use it in the Zama protocol for decrypting balances. But as a computation layer, it struggles with scalability and verifiability.
ZK is powerful but lacks composability. You can transfer a shielded ZK token, but you can’t easily stake or swap it while keeping it confidential. That limits what you can build.
FHE, however, is fully composable — you can stake, swap, lend, and perform any DeFi action privately, on the same chain where your assets already live. And scaling is straightforward: add more hardware, and it gets faster.
How Zama Uses FHE, MPC, and ZK Together to Enable Confidentiality on Existing Blockchains
Ehan: How do ZK, MPC, and FHE coexist in modern cryptography stacks?
Rand: In Zama, we use all three. Our protocol isn’t an L1 or L2 — it’s an encryption layer that sits on top of existing chains like Ethereum and Solana, adding confidential tokens and confidential DeFi without requiring bridges. Think of it as “HTTPS for blockchains.”
FHE powers all encrypted-state computation on-chain — balances, amounts, and private state updates. MPC is used for secure decryption, splitting the decryption key among multiple operators so no single party controls it. ZK provides scalability, and we’re working on ZK-FHE, enabling ZK rollups with FHE-encrypted states.
Ehan: How does this integrate with EVM ecosystems in practice?
Rand: It doesn’t change how developers build. You still write Solidity, specify what’s encrypted and who can decrypt it, then deploy as usual. Users also continue using the same wallets. If we do our job right, users won’t even notice Zama — the experience stays exactly the same.
Ehan: Are you building Zama to be the default privacy layer for multiple Layer 2s?
Rand: Absolutely. Zama is designed to be multi-chain from day one. We don’t compete with L1s or L2s; we complement them by adding confidentiality to Ethereum, Solana, Base, Tron, BNB, and others. Ultimately, we want users to have privacy no matter which chain they’re on.
Making Encrypted Smart Contracts Easy, and Scaling FHE for Real-World Use
Ehan: What changes for developers when building encrypted smart contracts using Zama’s FHE?
Rand: Almost nothing. You still write Solidity and deploy to Ethereum. Our library lets you mark which fields are encrypted and define who can decrypt them. If you’re building a confidential stablecoin, you just use encrypted integers for balances and specify decryption rules — users can see their own balance, auditors can access specific data if you want compliance. Zama doesn’t force any fixed approach; we simply give developers the tools to build whatever level of privacy or compliance they need.
Ehan: FHE is often seen as slow and expensive. What breakthroughs has Zama made?
Rand: That was true before Zama. When we started, a confidential transfer took around 10 minutes. Now we’re nearly 1,000× faster thanks to deep research — we have 37 PhDs and one of FHE’s inventors on the team. Today, FHE is already faster than Ethereum. With GPUs, we reach 500–1,000 TPS per chain (Solana, Base, Tron, etc.). And we’re building a dedicated FHE ASIC that will push performance to ~100,000 TPS on a single server with lower energy usage.
FHE is no longer a math problem — we solved that. It’s now a pure compute problem: more hardware means more speed and more use cases.
Ehan: What’s the biggest technical bottleneck now?
Rand: Hardware. Building an FHE ASIC is similar in complexity to a Bitcoin mining ASIC — around a $30–50M investment. It’s a big number, but small relative to enabling global encrypted payments. Today the challenge is time and capital, not the underlying science.
Ehan: How close are we to FHE apps that serve millions of users?
Rand: We can already support Ethereum-scale apps — 100 million users. For heavier workloads like perpetuals or confidential AI, we’ll need stronger hardware and ASICs. But most use cases with millions of users are already doable today; the extremely high-throughput ones will be fully feasible within the next four years.
Ehan: What’s the path to real-time encrypted computation?
Rand: Again, better hardware. FHE is fully secure — even against quantum attacks — and can support any computation: DeFi, stablecoins, trading, AI. We just need to make it faster.
Ensuring Compliance and Auditability in Private Computations Using Zama
Ehan: One of the industry’s challenges is balancing privacy with auditability. How does Zama ensure private computation does not undermine security or regulatory transparency?
Rand: I want to make something clear: Zama is for legitimate use cases. Everything we’re doing is to enable banks, financial institutions, companies, and startups to build confidential applications. We’re building a protocol from the ground up to enable compliance at the application level. The Zama protocol itself doesn’t encrypt anything. We provide tools for developers to build confidential tokens and applications. It’s up to each developer and token issuer to decide what’s encrypted and who is allowed to access the encrypted data in their contract.
Let me give you an example. If you’re a stablecoin issuer and you want to enable confidential transfers, the balances will be encrypted, and the amounts will be encrypted. You want the user to be able to see their own balance, of course, but you can also include in your smart contract the ability for you, as the token issuer, to see all the transactions of your own users. By doing this, you can comply off-chain, just like banks do today with anti-money laundering and other regulations.
And if you think about it, that’s how finance works today. You have a bank account, you see your bank account, your bank sees your bank account, but your neighbor doesn’t. We’re trying to enable the same kind of model on public blockchains.
Ehan: Does encrypted computation affect MEV ordering or fairness on Layer 1 or Layer 2 networks?
Rand: It changes nothing about the security of the blockchain itself because we sit on top of L1s and L2s. We don’t replace anything. We don’t replace the sequencing, we don’t replace the consensus, and we don’t change the way the system is architected. All of the rules for what’s encrypted and who can decrypt it are handled on the L1 where the application is running.
Think of the Zama protocol as a co-processor for public blockchains, adding confidentiality capabilities to them, but we don’t change anything in the blockchain itself. In fact, blockchains don’t even have to integrate us directly. We just deploy the Zama contracts on those chains, and it works. It becomes available for any developers and users on those blockchains.
FHE in Real-World Crypto Applications: Confidential Payments, Token Distribution, and Composable Private DeFi
Ehan: Which crypto use cases will adopt FHE first?
Rand: We see many, but payments are the biggest. If you’re using stablecoins like a bank account, you need confidentiality. Projects like Ray Cash are already building fully on-chain, non-custodial “banks” using encrypted stablecoins — your balance stays private, but you can still stake, swap, spend with a card, or make transfers. In places where banks fail or governments seize funds, having encrypted on-chain assets makes a huge difference.
Another clear use case is token distribution. We’re distributing Zama tokens as confidential tokens — everyone receives their allocation, but no one sees how much others get. This matters for fairness and avoiding unnecessary tension.
Trading is another major one. Today, when large investors move tokens, social media panics. Confidential swaps, deposits, and trades eliminate that market noise.
So it’s not “what can FHE be used for,” but rather “what won’t use FHE in the future?” I can’t think of many exceptions.
Ehan: Why are encrypted auctions a good FHE use case?
Rand: Auctions work beautifully with privacy. Google used a sealed-bid Dutch auction for its IPO, ensuring fair distribution and real price discovery. We built the same mechanism on-chain using FHE. In the Zama token auction, you bid with encrypted stablecoins — the price tier is public, but your bid amount stays private. At the end, everyone pays the same clearing price, with refunds handled automatically. And it all runs on Ethereum mainnet through Zama.
Ehan: What applications are only possible with FHE, not ZK?
Rand: Anything requiring privacy and composability. ZK can shield transfers, but it struggles with staking, swapping, lending, or using private identities across multiple contracts. FHE handles all of that.
Ehan: How are you making FHE development easier?
Rand: By integrating with developers’ existing tools. On Ethereum or Base, you still write Solidity — just import the Zama SDK. On Solana, you use our Rust library. No new languages, no new mental model. Developers can keep building and shipping the way they’re used to. That’s the whole point.
FHE Enabling Confidential AI and Private AI Agent Interactions
Ehan: You were involved in AI before Zama. How do you see FHE enabling private AI, both on-chain and off-chain?
Rand: Well, it’s very obvious that we’re never going to have AI on-chain if everybody can see your prompts. It’s just not going to happen. So, you need FHE or something similar to enable confidential AI on public infrastructure. I think deep confidentiality is necessary for AI. But even without running AI itself, imagine, for example, you want to enable payments between agents — you still want those payments to be confidential.
So, something like the X402 protocol for agent payments needs to be encrypted with FHE if you want it to be widely adopted. Don’t think of FHE as something you’re using just within a specific application. Think of FHE as a confidentiality layer that’s just put everywhere by default. Just like HTTPS when you connect to a website — the communication is encrypted, the data is encrypted. You don’t think about it because everybody does it. Now, when I send you a message on Signal, Telegram, or WhatsApp, the message is encrypted. We don’t think about it because it’s by default. We want to do the same thing for blockchain. Privacy should be by default. It should not be something people have to worry about.
Ehan: Could FHE solve the trust problem with AI agents making financial decisions?
Rand: I think if you want to trust an AI model, you have to give it more personal data. And if you give it more personal data, you want that data to be confidential. I believe confidentiality is needed if you want a truly personalized AI experience. It’s not going to happen without it.
Zama’s Organizational Growth, Compliance Approach, and Vision for Making Blockchain Privacy the Default
Ehan: As privacy regulations tighten, how will FHE fit into compliance frameworks?
Rand: Zama doesn’t define compliance — we simply provide the tools. Developers and token issuers decide what to encrypt and who can access it. Our role is to supply the infrastructure so they can implement whatever compliance model they need.
Ehan: Tell us more about your team and how Zama operates.
Rand: We have around 100 people, including 37 PhDs, making us the largest research team in cryptography and FHE. Our co-founder Pascal Paillier invented the Paillier FHE scheme, and we also work with leading academics like Nigel Smart.
We’ve raised over $150 million and reached a $1.2 billion valuation, backed by Multicoin, Pantera, and Protocol Labs. This gives us the resources and runway to push FHE into mainstream blockchain infrastructure.
Ehan: Zama also raised new money this year — how is your relationship with investors?
Rand: Our investors backed us from day one, knowing it would take years before commercialization. They supported us through the deep research phase because they understood the potential impact on blockchain and AI. Now the protocol works and is scaling quickly — none of this would have been possible without their long-term commitment and experience across multiple crypto cycles.
Ehan: How do you see blockchain privacy evolving in the next three to five years?
Rand: I think privacy on blockchain will follow the same path as HTTPS or encrypted messaging. Adoption is slow at first, then suddenly becomes the default. Once people realize they can have privacy, they won’t go back. Our goal is to make Zama the technology that powers that transition.
Ehan: How do you align a research-heavy team with execution?
Rand: Research can take years, so we separate short-term and long-term projects. Some ideas validate quickly; others require long horizons but deliver massive impact. What matters is that everyone understands we’re not just a research lab — we have real users depending on our protocol, and we must keep improving what they rely on.
Ehan: What organizational challenges did you face scaling from research to a real company?
Rand: Starting during COVID was the toughest part. Doing advanced research remotely — without whiteboards or in-person debates — is very difficult. We had to build strong frameworks and processes for distributed research collaboration. It was challenging, but essential for scaling.
Upcoming Events: Zama’s Mainnet Launch and Token Auction
Ehan: Are there any upcoming events from Zama that the community should be aware of in the next few months?
Rand: Well, the two most important events right now are that we’re launching our mainnet before the end of the year, and then in January, we’re going to be doing the auction to sell Zama tokens. Essentially, anyone who wants to use Zama tokens — whether they want to be an operator, a validator, or help secure the protocol, or any developer who needs tokens — will be able to get some.
I really want to emphasize that we’re doing an auction to sell tokens for an existing protocol that’s already built and already on the mainnet. This isn’t for a future project. This is a sale for people who want to use the token in the Zama protocol. I don’t think many people in crypto have done this before, holding a public auction and sale for something that already exists. And we’re using our own technology to run this sale, which I think is something really new.
Ehan: Zama has made rapid progress from FHE to encrypting applications. What’s the next major step in turning FHE into a mainstream blockchain primitive?
Rand: The next step is 1,000 TPS on GPU, then 10,000 TPS on GPU, and finally 100,000 TPS on ASIC. That’s it. The goal now for Zama is very simple: make it faster, make it cheaper, and make it more widely available on more blockchains. That’s our focus.
Follow us
Twitter: https://twitter.com/WuBlockchain
Telegram: https://t.me/wublockchainenglish
