Jeremy Bradley-Silverio Donato is a versatile leader overseeing daily operations at Zama. With a diverse background in the non-profit, education, and corporate sectors, he has collaborated with many organisations to develop strategies, manage communications, and guide policy. Besides his operational responsibilities, Jeremy is also a skilled author and was named to Business Elite's 40 Under 40 in 2022.
Zama is an open-source cryptography company developing cutting-edge FHE solutions for blockchain and AI. Their technology enables developers to create privacy-focused applications where data remains encrypted during processing, providing secure, composable, and scalable on-chain confidentiality.
We spoke with Jeremy Bradley-Silverio Donato, COO at Zama, at DevConnect Buenos Aires about the resurgence of privacy tech in blockchain and why Fully Homomorphic Encryption (FHE) is set to make confidentiality standard on the decentralised web.
Read more about Zama’s vision for an encrypted blockchain ecosystem in the interview below.
I always like to start with how you got into crypto. What was that journey like for you?
I do not come from a Web3 or crypto background. I was working in edtech before joining Zama, building online learning platforms. This was before COVID, so online learning was also a new thing and a bit intimidating for people. When I joined Zama, it was a similar situation. I came on board right at the start of this journey around FHE and privacy-preserving technologies. Being involved from the ground up feels very exciting to me.
Crypto is often said to be based on the principles of privacy, but that was never really built into the core protocol design. Why do you think it took so long for blockchains to prioritise confidentiality at that level?
If we step back for a moment, I believe we’re seeing a similar situation with artificial intelligence. Everyone agrees AI systems should be more confidential and private, but we haven’t taken significant action yet.
The same thing happened in Web3. Only recently have we reached a turning point where people are seriously considering use cases that truly require confidentiality. When discussing on-chain finance, you need some level of privacy. For example, if you want to run payroll on-chain and don’t want everyone in the company to know each other’s earnings, basic confidentiality becomes necessary.
People have attempted to address this with minimal knowledge, which has its uses, along with one-off apps and private blockchains. Zama is not trying to compete with those technologies. We are exploring how to offer confidentiality on a more institutional level and in a more dynamic way for the industry.
In simple terms, what is FHE? And what breakthroughs made it possible to have FHE on-chain at scale?
People often think FHE is still too slow to be useful. FHE stands for Fully Homomorphic Encryption. It provides true end-to-end encryption and enables computation over encrypted data. The server, the blockchain, or whatever is in the middle can stay unaware of the actual data. This is very helpful for keeping information confidential.
What has helped bring FHE into the spotlight is that people have recognised the limitations of other privacy technologies. ZK has limitations for some use cases, and FHE has limitations for others. The goal is to find the right tool for each specific use case.
As ZK, TEEs, and MPC reach their limits in some areas, people began to ask what else exists. FHE gained momentum partly because of that and partly because companies like Zama heavily invested in making it practical. At Zama, about 50 per cent of all FHE researchers worldwide work with us.
Building on that, what are some limitations of ZK and some limitations of FHE in your view?
ZK is suitable for use cases that don’t require high throughput or compute. It’s ideal when you want to demonstrate something concisely without revealing everything.
Zama’s FHE protocol currently operates slightly above Ethereum’s basic capacity. It is fast enough to handle everything you can do on Ethereum. We’re at around 20 TPS now and aim for 1,000 TPS by 2026. With ASICs, it can reach into the thousands. The good thing is you can combine multiple GPUs and scale horizontally for better performance.
No technology operates in isolation. That’s why I avoid direct comparison battles. All of these technologies have their place. People who value privacy aren’t loyal to a specific technique; they care that the societal importance of confidentiality is upheld.
What breakthroughs made FHE go from an academic idea to something that works onchain at scale?
For a long time, FHE was primarily a mathematics problem. Zama started as an R&D company to address that. We initially believed the main use cases would be machine learning and AI. They will be important later, but we discovered there’s currently a lot of energy and enthusiasm around doing things confidentially on-chain. That has become a very strong use case.
We also found that organisations, governments, and institutions want to move services onto the blockchain. In places like Argentina, where the economy has suffered and legacy systems are broken, blockchain provides a way to achieve rapid societal progress. But to do that, you need basic confidentiality. If you’re going to save or transact at scale, you don’t want everyone to see exactly what you hold or spend.
One of the bigger fears with privacy tech is that it might break composability. What architectural choices allow Zama to preserve composability while keeping everything encrypted?
That is one of the coolest things about the Zama Protocol. It is fully compatible with DeFi. You can perform swaps, lending, staking, and all other activities seamlessly.
The reason is that we helped design a new token standard that makes this possible. It functions like an ERC-20 token but for encrypted tokens. It is composable, decentralised, and easy for developers to use. Without that standard, what you’re describing wouldn’t have been possible. With it, we can maintain composability in an encrypted world.
You mentioned earlier that you are around 20 TPS, targeting 1,000. What bottlenecks did you have to solve to get there, and what do you still see ahead?
FHE was a mathematical problem for a long time. Once our team solved the core math, everything else became simpler. Then it shifted to a compute problem. We often say FHE is no longer a math problem; it is a compute problem. You can solve compute problems with more hardware, which can be costly, or by optimising algorithms.
What I find really interesting about FHE, as Zama implements it, is that we use something called programmable privacy. This feature addresses many regulatory concerns.
People often assume privacy tools are only used by bad actors to hide money, but with programmable privacy, you can grant selective access to certain data elements to specific individuals, such as regulators or payroll personnel. The data owner decides who can see what, while everything remains fully encrypted end-to-end.
Testnet v2 is now the release candidate for mainnet. What were the biggest lessons from running two testnets?
The biggest lesson is to avoid blowing up too quickly. The growth was incredible. Having so many people interested in using the technology is an excellent problem to have. As with any project, you learn to communicate with your community, listen, and incorporate feedback. We also made incremental improvements to the core FHE technology based on that feedback.
I am thankful to the community for sticking with us through the issues. We told them from day one, “Please go and break it.” That is what a testnet is for.
The 55-hour distributed key ceremony was a significant milestone. What did that process prove about your readiness for production?
It shows that we have strong industry partners and that large-scale coordination is possible. We are collaborating with 13 MPC operators and 5 FHE node operators as genesis operators. Zama is one of them.
The involvement of these well-known players significantly boosts trust. This key ceremony required extensive coordination and thorough documentation to ensure everyone clearly understood their role. Distributing the keys among multiple operators creates a strong trust model and sets the tone for how the community will perceive the network and the Zama token. It demonstrates that control rests not with a single individual, but with a group of reputable partners.
Looking ahead one year, five years, ten years, what does a perfect world for privacy look like to you?
I believe Zama’s ultimate goal is to have everything on-chain encrypted. That is an ambitious goal, but I think we will get there. As governments, organisations, and AI systems become more decentralised, they will also need to be encrypted. The vision extends beyond on-chain finance to other digital systems as well.
Before HTTPS, everything on the internet was basically unencrypted. Then, HTTPS introduced a basic level of security and privacy. Zama’s goal is to create something like “HTTPZ” for blockchains, so everything is encrypted by default. End users might not even notice. They’ll just know their data is safe.