Conversation with Zama CEO: FHE Will Reshape Blockchain Privacy

In this episode of WuTalk Podcast, Rand Hindi, CEO of Zama, delves into the disruptive potential of Fully Homomorphic Encryption (FHE) in the blockchain industry. He emphasizes that FHE will provide a crucial balance between privacy protection and scalability. Zama’s protocol adds a native privacy layer to existing public blockchains like Ethereum and Solana, enabling truly encrypted and usable on-chain transactions.

Rand also explains why FHE stands out among privacy technologies—compared to Zero-Knowledge Proofs (ZK) or Multi-Party Computation (MPC), FHE is better suited to support composable confidential token transfers and DeFi operations. Additionally, he shares Zama’s latest technical progress and the upcoming confidential token auction.

Audio transcription provided by GPT may contain errors. Please listen to the full podcast on Xiaoyuzhou, YT, etc. The views expressed are those of the author and do not represent WuTalk. Please strictly comply with local laws and regulations.

Xiaoyuzhou:

YouTube:

Background on Zama and CEO Rand Hindi

Ehan: Welcome to WuTalk Podcast. If you want to get the latest updates on blockchain and crypto, subscribe to our YouTube channel and follow us on Twitter to stay in sync and join the discussion.

Today, we’re excited to welcome Dr. Rand Hindi, CEO of Zama. Welcome! Could you briefly introduce yourself and tell us how you started Zama?

Rand: Hi everyone, I’m Rand, currently CEO of Zama. I started coding at age 10 and launched my first company as a teenager in the ‘90s. At 21, I pursued a PhD in AI—over 20 years ago now. After that, I created one of Europe’s earliest AI companies, which was acquired by Sonos in 2019. In 2020, together with my co-founder Pascal Paillier—one of the inventors of FHE—we started building Zama.

I’ve been in crypto since 2013, having experienced four full cycles. I’m also an investor in over 100 companies across biotech, AI, and crypto.

Ehan: Was there a key moment when you realized “privacy” would become a core limitation for AI and blockchain?

Rand: My early interest in privacy stemmed from my first company—a social platform I started as a teenager. Seeing user data constantly flowing into our servers felt wrong to me; just because I was the founder didn’t mean I should see all that private information.

Later, during my PhD, I worked on AI for genomics, dealing with a lot of medical data like DNA. That reinforced my belief that privacy is fundamental for AI to thrive.

When I entered the blockchain space, I found it strange: in the real world, you’d never show your bank account to your neighbor, but on-chain everything is public by default. That’s not normal.

So for me, privacy isn’t just an issue for AI or blockchain; it’s a foundational capability that all digital products should have.

FHE vs. Other Privacy Technologies

Ehan: For newcomers, what’s the fundamental difference between FHE (Fully Homomorphic Encryption), ZK, and MPC?

Rand: In the on-chain privacy tech stack, there are four main types: FHE, MPC, ZK, and TEE. Only FHE offers all three properties: security, composability, and verifiability.

FHE’s encryption is strong enough to resist future quantum attacks, meaning data encrypted and put on-chain today won’t be broken by quantum computers tomorrow. In contrast, TEE has recently been shown to have fundamental flaws—researchers found major issues in Intel’s implementation, compromising all privacy chains relying on TEE. Intel themselves have stated decentralized protocols are outside their security guarantees.

MPC is great for key management—we use it at Zama for balance decryption—but as a computation layer, it struggles with scalability and verifiability.

ZK is powerful, but its biggest limitation is non-composability. You can transfer a ZK privacy token, but it’s hard to stake or swap it while maintaining privacy, limiting many applications.

FHE doesn’t have this problem. Whether it’s staking, swaps, lending, or other DeFi operations, everything can be done privately on the same chain. FHE also scales directly: more compute power equals better performance. Just add hardware and it keeps getting faster.

How Zama Uses FHE, MPC, and ZK to Enable Privacy on Existing Chains

Ehan: How do ZK, MPC, and FHE coexist in the modern cryptography stack?

Rand: At Zama, we use all three. Zama isn’t a new L1 or L2, but a layer on top of existing public blockchains (like Ethereum or Solana), enabling native confidential tokens and DeFi without bridges. Think of it as “HTTPS for blockchains.”

FHE handles all encrypted-state computation—balances, amounts, state updates all in ciphertext. MPC is used for secure decryption, distributing keys across nodes to avoid single points of control. ZK handles scaling—we’re also developing ZK-FHE, running FHE-encrypted states on ZK Rollups.

Ehan: How does the integration work in the EVM ecosystem?

Rand: For developers, it’s almost the same. You still write Solidity, just add a few lines to specify which fields are encrypted and who can decrypt, then deploy as usual. Users still use their wallets. Ideally, users won’t even notice Zama is there—the experience is unchanged.

Ehan: Will Zama become the default privacy layer for multiple Layer2s?

Rand: Exactly, that was our goal from the start. Zama is inherently multi-chain, not competing with any L1 or L2, but adding confidentiality to Ethereum, Solana, Base, Tron, BNB, and more. Wherever users are, we want privacy to be the default.

Simplifying Encrypted Smart Contract Development and Scaling FHE

Ehan: What changes do developers need to make to build encrypted smart contracts with Zama’s FHE?

Rand: Basically none. You still write Solidity and deploy to Ethereum. Our library lets you mark which fields are encrypted and who can decrypt. For a confidential stablecoin, just make balances encrypted integers and set decryption permissions—users see their own balances, and if you need compliance, auditors can view specific data. Zama doesn’t restrict development; it just provides tools so you can implement privacy and compliance as needed.

Ehan: FHE used to be considered slow and expensive. What breakthroughs has Zama made?

Rand: Before Zama, that was true. When we started, a confidential transfer took 10 minutes. Now, it’s nearly 1,000 times faster, thanks to deep R&D—our team includes 37 PhDs and one of FHE’s inventors.

FHE is now faster than Ethereum. On GPU, we reach 500–1,000 TPS per chain (Solana, Base, Tron, etc.). We’re building dedicated FHE ASICs, which will allow a single server to reach about 100,000 TPS with lower energy consumption.

FHE is no longer a math problem—we solved the math; now it’s just about compute power: more power, faster FHE, broader use cases.

Ehan: What’s the biggest technical bottleneck now?

Rand: Hardware. Building an FHE ASIC is like building a Bitcoin miner ASIC—it takes $30–50 million. That’s a lot, but worth it to enable fully encrypted global payments. The main challenges now are time and funding, not science.

Ehan: How far are we from FHE supporting apps with tens of millions of users?

Rand: We can already support Ethereum-scale—hundreds of millions of users aren’t a problem. Heavier apps like perpetual swaps or confidential AI need more robust hardware and ASICs. But most million-user scenarios are covered today; ultra-high-throughput apps will be fully feasible within four years.

Ehan: What’s the path to “real-time encrypted computation”?

Rand: Still hardware. FHE is already very secure—even quantum computers can’t break it, and it supports all kinds of computation: DeFi, stablecoins, transactions, AI, you name it. The only thing left is to make it faster.

Ensuring Compliance and Auditability in Private Computation

Ehan: How does Zama ensure private computation doesn’t undermine security or regulatory transparency, balancing privacy and auditability?

Rand: First, everything Zama does is for legal, legitimate use cases. We want banks, financial institutions, companies, and startups to build compliant, privacy-centric apps. Zama’s protocol doesn’t encrypt anything itself—we provide tools so developers and token issuers can build confidential tokens or apps, deciding what to encrypt and who can access it.

For example, if you issue a stablecoin with confidential transfers, balances and amounts are encrypted. Users see their own balances, but you can add permissions for issuers to view certain transactions for compliance audits like AML.

It’s like the real world: you see your bank account, the bank sees it, but your neighbor can’t. We want to bring this model on-chain.

Ehan: Will encrypted computation affect MEV ordering or on-chain fairness at L1 or L2?

Rand: Not at all. We’re built on top of L1 and L2, not replacing any part. We don’t change ordering, consensus, or system architecture. All rules about “what’s encrypted” and “who can decrypt” are still defined on the underlying L1.

Think of Zama as “an encryption coprocessor for public chains.” It adds privacy without changing the chain itself. The chain doesn’t even have to integrate us—just deploy the Zama contract and any developer or user can use it directly.

FHE Applications in Crypto: Confidential Payments, Token Distribution, and Composable Private DeFi

Ehan: Which crypto use cases will adopt FHE first?

Rand: We see many, but “payments” is the most obvious. If you want to use stablecoins as a bank account, privacy is essential. Projects like Ray Cash are building fully on-chain, non-custodial “on-chain banks” with encrypted stablecoins—balances are private, but you can still stake, swap, spend, transfer cross-border. In places where banks fail or governments freeze funds, on-chain encrypted assets provide huge peace of mind.

Another clear use case is token distribution. We’ll distribute a confidential Zama token version—everyone gets their share without knowing how much others get. This is crucial for fairness and avoiding unnecessary conflict.

Then there’s trading. Today, as soon as a whale moves tokens, social media explodes. Confidential swaps, deposits, and trades via FHE can eliminate this market noise completely.

So the question isn’t “what can FHE do,” but “what won’t need FHE in the future?” I can hardly think of exceptions.

Ehan: Why are crypto auctions particularly well-suited for FHE?

Rand: Auctions and privacy are a natural fit. Google’s IPO used a sealed Dutch auction for fair allocation and real price discovery. We brought the same mechanism on-chain using FHE.

For Zama’s token auction, users bid with encrypted stablecoins—price tiers are public, but your bid amount is always private. Everyone settles at the same clearing price, and differences are refunded automatically. The entire auction is on Ethereum mainnet, powered by Zama technology.

Ehan: What applications are only possible with FHE, not ZK?

Rand: Any scenario needing both privacy and composability. ZK enables confidential transfers but hits limits with staking, swapping, lending, or using private identities across contracts. FHE makes all these operations private and composable.

Ehan: How do you make FHE development easier?

Rand: The key is seamless integration with existing developer toolchains. On Ethereum or Base, you still write Solidity, just add the Zama SDK; on Solana, use our Rust library. No new languages or changes in habits required.

Developers just keep doing what they do best: coding, deploying, iterating. Our goal is to ensure this smooth workflow continues.

FHE for Private AI and Confidential Agent Interactions

Ehan: Before Zama, you worked in AI. How do you see FHE enabling “private AI” on and off-chain?

Rand: It’s simple: if everyone can see your prompt, AI can’t be on-chain—it’s totally unworkable. You need FHE or similar privacy tech to run AI on public infrastructure safely. Deep confidentiality is essential for AI.

And it’s not just about running models—if you want AI agents to make payments, you want those payments to be confidential too.

Protocols for agent-to-agent payments (like X402) need FHE to encrypt payment content for real-world use.

Don’t think of FHE as just an internal app component—it should be the “privacy foundation” for all systems, like HTTPS. When you visit a website, the transmission is encrypted; when you use Signal, Telegram, WhatsApp, your messages are encrypted. No one thinks, “Oh, I’m using encryption now”—it’s just default.

We want blockchain to be the same: privacy by default, not something users have to worry about.

Ehan: Can FHE help solve AI agents’ “trust issues” in financial decision-making?

Rand: If you want to trust an AI model, you have to give it enough personal data. Once you provide that data, you want it to remain private. Truly personalized AI experiences require privacy as a prerequisite—it’s impossible otherwise.

Zama’s Organizational Growth, Compliance Strategy, and Path to Industry Standard Privacy

Ehan: With global privacy regulation tightening, how will FHE fit into compliance frameworks?

Rand: Zama doesn’t define “compliance”; we just provide tools. Developers and token issuers decide what data to encrypt, who can access it, and how. Our role is to offer infrastructure so they can implement their own compliance requirements.

Ehan: Can you describe your team size and how Zama operates?

Rand: We’re about 100 people now, including 37 PhDs—one of the world’s largest research teams in FHE and cryptography. My co-founder Pascal Paillier invented the Paillier FHE scheme, and we collaborate with top academics like Professor Nigel Smart.

To date, we’ve raised over $150 million, with a $1.2 billion valuation and backing from Multicoin, Pantera, Protocol Labs, and other top investors. This gives us the resources and runway to make FHE mainstream blockchain infrastructure.

Ehan: Zama also completed a new funding round this year. How’s your relationship with investors?

Rand: Our investors have supported us from day one. They knew this technology would take years to commercialize, but invested for the long term because they saw FHE’s potential impact on blockchain and AI. Now the protocol is live and scaling fast—none of this would be possible without their experience and long-term commitment.

Ehan: How do you see blockchain privacy evolving in 3–5 years?

Rand: I think blockchain privacy will repeat the history of HTTPS and encrypted messaging—slow at first, then rapidly becoming the default. Once users realize “on-chain privacy is possible,” they won’t accept anything less. Zama’s goal is to be the foundational tech driving this transition.

Ehan: How do you keep a research-focused team execution-oriented?

Rand: Research often takes years, so we split topics into short- and long-term. Some directions can be quickly validated, others need long-term investment but have huge impact. The key is for everyone to understand: we’re not just a research lab anymore—we’re a real company with users relying on our protocol, and we must keep improving the product.

Ehan: What organizational challenges did you face scaling from research team to real company?

Rand: The hardest part was founding the company during the pandemic. Doing intense research remotely is tough—a lot of foundational discussion really needs face-to-face, whiteboard sessions. We had to build processes and frameworks for distributed research collaboration. It was hard, but crucial for team growth and long-term development.

Upcoming Milestones: Mainnet Launch and Token Auction

Ehan: What major events should the community look for from Zama in the coming months?

Rand: The two biggest things: we’ll officially launch our mainnet by the end of the year, and in January, we’ll hold a Zama token auction. Anyone needing Zama tokens—operators, validators, protocol security participants, or developers—can acquire tokens in this auction.

One thing I want to stress: the tokens we’re auctioning are for a protocol that’s already built and live on mainnet. This isn’t a “future promise” sale. It’s rare: a public auction for an existing protocol, and we’ll run the auction using our own technology—a first.

Ehan: Zama is advancing quickly in FHE and encrypted apps. What’s the next key goal?

Rand: Simple: reach 1,000 TPS on GPU, then 10,000 TPS, and finally 100,000 TPS on ASIC. Our core focus is making it faster, cheaper, and available on more public blockchains. That’s our priority going forward.