Beyond the Blockchain: Grigore Roșu on the Future of Verifiable Systems
In Episode 7 of Exponential, Grigore Roșu — professor, computer scientist, and founder of Pi Squared — joins the show to explore a bold thesis: blockchains as we know them are only a first step.
The future of web3, Grigore argues, lies in universal computation, massively parallel verification, and an evolution beyond the bottlenecks of ordered chains.
Pi Squared, a Nexus partner, is building toward that future. With its FastSet protocol and a new approach to trust, the company is rethinking how verifiability and decentralization scale — while keeping verifiability at the core.
“Blockchains gave us a great run. But we have to transcend them. We need a paradigm shift.”
From proofs to protocols
Grigore journey spans from theoretical mathematics in Romania to teaching programming languages and distributed systems at the University of Illinois Urbana-Champaign.
Along the way, he helped pioneer formal verification for mission-critical software at NASA and later founded Runtime Verification, a company that brought those techniques to industry. Pi Squared is the next logical step — built around a deceptively simple idea: if you claim something is true, you should be able to prove it.
“Mathematical proof is the best argument we have as a species for correctness,” Grigore said. “Pi Squared starts with that and adds a cryptographic layer on top. It’s a proof of proof.”
The result is a stack that can execute and settle computational claims across any programming language, verify them automatically, and generate cryptographic attestations of correctness. That’s where the name comes from: proof (π) of proof (π²).
Order as a bottleneck
While cryptographic verification is becoming more powerful and accessible thanks to zero-knowledge proofs, Grigore believes the next breakthrough lies in rethinking how networks coordinate.
At the heart of today’s blockchains is the concept of strong consistency: every node maintains the same global state, updated in a strict sequence of blocks. But that design imposes a hard ceiling on throughput.
“Strong consistency is like having a million people go through a single door. No matter how organized you are, it will be slow.”
FastSet, Pi Squared’s consensus protocol, takes a different route. Instead of total ordering, it uses strong eventual consistency — a model in which nodes can process transactions in different orders, so long as the final state converges.
More than 90% of web3 transactions, Grigore claims, don’t actually require global ordering. Payments, escrows, voting — all can happen in parallel.
“Validators don’t even need to talk to each other. Transactions finalize independently. There are no blocks.”
This architecture yields performance gains. FastSet already delivers over 100,000 transactions per second, with sub-100ms finality. But the biggest innovation is invisible: every transaction remains fully verifiable, just like on a blockchain.
Post-blockchain doesn’t mean post-web3
Grigore is quick to clarify that his critique isn’t about rejecting web3. Digital property rights, decentralized systems, and cryptographic trust still matter. What needs to evolve is the infrastructure.
“We don’t care how things work under the hood,” he said. “When you switch from gasoline to electric cars, you don’t have to relearn how to drive. That’s the kind of upgrade we’re building.”
FastSet retains everything people expect from blockchain UX: cryptographic guarantees, transparency, auditability. What changes is the execution model. Transactions aren’t lined up in blocks — they’re settled directly, independently, and verifiably. And when ordering is needed, it can be applied selectively, by application-level logic.
The bigger picture of verifiability
For Grigore, verifiability is more than a technical goal — it’s the defining principle of a trustworthy internet.
“Without verifiability, web3 doesn’t exist,” he said. “It’s the only way we can truly own our assets and trust our systems.”
He sees zero-knowledge proofs as the great enabler, compressing massive computational traces into short, verifiable signatures.
But he also cautions against viewing ZK as a silver bullet. It’s one layer of a broader proof system — one that must also scale with real-world demands and developer ergonomics.
“We should be able to write smart contracts in any language — Java, Python, C—and get proofs automatically. That’s what we’re building.”
What comes next
Despite the technical complexity, Grigore is optimistic. The core conceptual pieces — parallel consensus, programmable proofs, universal languages — are falling into place. The remaining challenges, he argues, are engineering ones.
And with both crypto and AI accelerating, he sees an inflection point coming fast. “We don’t have a shortage of liquidity,” he said. “We have a shortage of use cases. But the moment is here. AI will need verifiability. And only crypto can match its speed.”
“Soon, we won’t even say web3 anymore. It’ll just be the background. As natural as owning your own assets.”
You can find Exponential on Spotify, Apple Podcasts, and YouTube. Whether you’re a builder, investor, or simply someone fascinated by the future of computation, we hope you’ll subscribe and join the conversation.
