Alice Liu: Welcome to House of ZK! Today, I'm thrilled to sit down with Teemu Päivinen, Founder and CEO of Gevulot. Teemu, thanks so much for joining me.
Teemu Päivinen: Thank you, Alice. It’s great to be here and chat with you.
Alice: Let’s start with your journey. Can you share how you first got involved in the Web3 space and the world of zero-knowledge technology?
Teemu: Of course. My story begins back in 2011 when a friend of mine introduced me to Bitcoin. He had discovered it on an obscure forum, and the idea of native internet money was captivating. At the time, I was pursuing a career in traditional finance, but I quickly became disillusioned with the inefficiencies and misaligned incentives in the industry. Bitcoin offered a fresh sandbox for experimentation - a way to build entirely new models.
We started day trading Bitcoin and soon realized how hard it was to buy and sell in Europe. Inspired by Coinbase in the U.S., we launched Coinmotion, a Bitcoin exchange, in 2012. Coinmotion implemented full KYC from day one, which made growth more challenging, but it’s now the largest brokerage service in the Nordics. After selling the exchange to a competitor, I explored other ventures, including a video company and advising projects like Dapper Labs during the early days of CryptoKitties. These experiences gave me a unique perspective on building and scaling Web3 technologies.
Eventually, I co-founded Equilibrium in 2018, a company focused on low-level engineering for blockchain systems. This led to collaborations with top projects like the Ethereum Foundation, StarkWare, and Aleo. Gevulot, which I now lead, was spun out of Equilibrium to address decentralized compute needs.
Alice: That’s a fascinating trajectory. With Gevulot, you’re tackling decentralized compute. How do you see the big-picture role of ZK technology in shaping the future, both in and beyond blockchain?
Teemu: Zero-knowledge technology is transformative because it enables provable correctness without revealing sensitive details. In blockchain, this is critical for scalability and privacy. Beyond blockchain, ZK has potential applications in areas where trust and transparency are paramount, such as digital elections or proving computations in cloud environments.
To realize this potential, we need to lower ZK’s computational overhead, which is currently a significant bottleneck. If the overhead were zero - even though that’s an idealized scenario - it would unlock use cases across industries. Every AWS computation could include a proof of correctness, eliminating trust in cloud providers. While we’re not there yet, progress in academia and engineering is moving us closer.
Alice: You’ve written about the “ZK endgame.” Could you elaborate on what that means for Gevulot and the broader ZK ecosystem?
Teemu: The “ZK endgame” envisions a digital world where trust is replaced with verifiable proofs. For Gevulot, this means building the infrastructure to make ZK-based systems scalable, cost-effective, and accessible. Our focus is on creating a decentralized cloud where computations can be proven correct without centralization risks. We’ve already solved the file storage problem with technologies like Filecoin; now it’s time to address compute.
We’re also working on making proof generation efficient. For example, Gevulot’s ZK Cloud allows computations to be performed on a single node with minimal overhead, competing with centralized cloud providers like AWS. By abstracting the complexity, end-users can benefit from censorship-resistant, permissionless systems without needing technical expertise.
Alice: That’s an ambitious vision. Can you walk us through some practical use cases for ZK technology that you’re excited about?
Teemu: In the near term, ZK technology is essential for scaling blockchain through rollups. Rollups reduce costs, enable privacy, and allow different execution environments. Longer-term, I’m excited about ZK’s role in proving L1 blocks and enabling stateless co-processors. These developments could transform Ethereum and other blockchains into “world computers,” where validators verify proofs instead of re-executing transactions.
Beyond blockchain, I see potential in applications like algorithmic calculations and ZK machine learning. While proving AI models isn’t commercially viable yet, it’s a promising area for future research. For now, even simple statistical calculations performed off-chain can significantly enhance blockchain’s capabilities.
Alice: What about Gevulot’s specific contributions? Can you explain the relationships between components like ZK Cloud, Firestarter, and ZK Boost?
Teemu: ZK Cloud is our overarching vision for decentralized compute. Firestarter, launching soon, is the production-ready iteration of this system. It’s a stable environment for onboarding production workloads with hundreds of GPUs and CPUs. Users can deploy containerized applications, and the platform handles everything from scaling to cost optimization.
ZK Boost, on the other hand, is an abstraction layer for proof generation. It standardizes the API for interacting with different proving options, from centralized providers to decentralized networks. This simplifies integrations and ensures compatibility across the ecosystem.
Alice: It sounds like Gevulot is addressing both immediate and long-term needs. For those who want to get involved, what opportunities are available?
Teemu: There are several ways to get involved. Developers can experiment with our Devnet or apply for a free tier on Firestarter when it launches. Node operators can join the network to contribute compute resources. We’re also actively engaging with the community to refine our tools and APIs, so feedback is always welcome. Follow us on Twitter or check our documentation at docs.gevulot.com for the latest updates.
Alice: Thank you, Teemu. It’s been incredibly insightful to hear your thoughts on ZK technology and Gevulot’s role in the space. I look forward to seeing how your vision unfolds.
Teemu: Thank you, Alice. It’s been a pleasure. Let’s do this again sometime!Follow @teemupai and @gevulot_network on X.
