ETHPrague 2026 (Prague, Czech Republic, May 9–12, 2026) is the most ideologically distinctive Ethereum event of the 2026 cycle. 117 talks across tracks spanning governance, AI agents/security, network states, ZK proving, encrypted computing, decentralization, public-goods funding, and orbital compute. The Prague edition functions as the cypherpunk-Ethereum counterweight to the institutional EthCC[9] (April 2026) — a venue where the failure modes of decentralization are interrogated rather than glossed.

The conference’s center of gravity

ETHPrague’s signal-to-noise ratio is unusually high because it operates outside the institutional-DeFi attractor. Six themes dominate:

  1. Neocypherpunk consolidation as CROPS — Polar (Cypherpunk is Good Business) crystallised the 2025–2026 cypherpunk renaissance into a four-letter regulative ideal: Censorship resistance, Open source, Privacy, Security. The framing absorbed Vitalik’s “Make Ethereum Cypherpunk Again,” Pcaversaccio’s Ethereum Cypherpunk Manifesto, Shawe’s DeFi Punk, the Trustless Manifesto, and the (controversial) Ethereum Foundation mandate. Decentralization explicitly demoted from meta-principle to a means-toward-an-end of censorship resistance. → Neocypherpunk & CROPS

  2. EZ / synchronous L2 composability — Jordi Baylina’s headline project, debuted in a Vitalik fireside chat. EZ is the most credible proposal for making L2s atomically composable with Ethereum L1 — a single transaction can synchronously call across base, Arbitrum, and L1, with each rollup defining its own state-transition function and trust set of proof systems. Real-time ZK proving (sub-12s) is the technological precondition. First mainnet experimental smart contracts targeted for Devcon Berlin (late May 2026). → L2 Synchronous Composability

  3. AI agents have killed the audit cycle, both attacking and defending — April 2026 was the highest-loss hacking month on record, almost entirely AI-augmented. Cost-of-intellect halves every 1.5 months (Eigor Gulamov / 7chat); state-of-the-art models now find PhD-level cross-domain bugs. Riptide/Greg AI’s continuous AI audits already collect public bounties on Lido, Chainlink, Aave, Uniswap. Audit cycles measured in months are economically obsolete. → AI Smart Contract Auditing

  4. Encrypted computer and obfuscation — Janmajaya Mall (Phantom Zone) reported on a year of progress moving from FHE+threshold committees toward direct circuit obfuscation via the local-mixing approach (reversible circuits + shuffle/shoot/lock/compress); blow-up reduced from FHE’s billion× to ~100–2000×. Machina (Ethereum Foundation team) implemented Diamond IO from lattice primitives. The path to FHE-free private smart contracts is opening. → Encrypted Computing — FHE & Obfuscation

  5. Network states getting concrete — Balaji Srinivasan (remote) reported on Network School in Malaysia, sub-$18K/yr cost-of-living, alongside Burning Man as a 40-year governance case study (Marian Goodell). The dominant frame: history is running in reverse — the postwar centralization peak is over; expect 600+ jurisdictions emerging from the unwinding American empire. → Network States & Pop-up Cities

  6. Decentralization theatre admitted in public — A rare honest panel (“The Decentralization Stack: Where Are We Still Lying to Ourselves?”) and Migle Rakitaite’s “Decentralisation Theatre” forced a public admission: October 20, 2025 (the AWS outage that took down “half the blockchain”) proved that Ethereum’s social-coordination stack runs on GitHub + Discord + Twitter + Infura/Alchemy + Wallet Connect — every layer above pure L1 is centralized.

Track structure (~11 tracks)

  • Main Stage — Vitalik & Jordi Baylina fireside (EZ debut); Martin Hansen (Ethereum kernel funding); Network School / Balaji (remote).
  • Encrypted Computer / Privacy — Privacy Pools v2 (McCabe); Phantom + Diamond IO update (Mall); ZK passport personhood (Viki Val / X511); Open Signer; Recovery as key to sovereignty (van Wijk / Nilium).
  • AI Track (Tooling / Audit / Agentic) — AI Killed the Audit Cycle (Riptide); Beyond Human Review (Gulamov / 7chat); Benchmarking AI for SC Security; Hardware Wallets in the Age of AI; The Agentic Risk; Your AI Agent Works for OpenAI (Pavlín / LMAO).
  • DAOs / Governance — Lessons from L2 DAOs (Urbański); Decentralisation Theatre (Rakitaite); Governance Security Checklist (Steidl); Designing for Participation; Burning Man fireside.
  • Network States / Pop-up Cities — Pop-up Cities (Jimin Lee); Network State experiment (Stancel); Digital economic zones (Poli); From Temporary Gatherings to Durable Institutions (Fritsch).
  • ZK / Proving — ZisK Pushing Limits (Baylina); From ZK rollups to ZK Ethereum (Shemyakov); zk-Based Consensus Proofs (Kux); POD recursive client-side proofs (Knight); Keccacheck.
  • Stablecoins / DeFi — $330B Stablecoin Blind Spot (TokenBrice); TradFi on Stablecoin Rails; When CoWs Meet RWAs; Risk Ratings; The Future of DeFi.
  • Cypherpunk / Philosophy — Cypherpunk is Good Business (Polar); Dark Agorist Cyber Guerilla (Taaki); Decentralize or Die (Almond); How to Pirate a $60 Trillion Institution (Hope); The Decentralization Stack panel.
  • Orbital / Space — Orbitport (Yahalom); Confidential Compute for Orbital AI (Bar); Space Fabric (Rezabek).
  • Public Goods / Funding — Ethereum Kernel (Hansen); Mapping Dependencies of Publicly Funded Open Source (Sukhinina); Sourcing Funding for Ethereum Growth (Mehta); Diversifying Funding via FRCs (Romanutti).
  • Solidity / Dev Tooling — What’s Next for Solidity (Jacob / Argo); Plank (Philogy); Wake 5.0 fuzzing (Prevratil); Past/Present/Future of EVM Fuzzers (Grieco); Distributed Frontend Deployment (Losev); Unfucking the RPC Layer (Lukas); EIP-8141 Frame Transactions (Forshtat).

What ETHPrague adds to the wiki

Compared to EthCC[9] (institutional DeFi, ZK production, agents-as-replacement-for-dapps), ETHPrague extends coverage into:

Cross-conference resonances

  • Privacy by default crystallizing: Privacy Pools v2 (ETHPrague) + Kohaku SDK (ECC2/EthCC9) + Cloaked wallet + Pcaversaccio’s roadmap = the “privacy as wallet feature, not standalone app” thesis is now consensus. → Privacy as UX Design
  • AI in audit pipeline: ETHPrague’s audit-cycle-is-dead framing complements EthCC[9]‘s shift-left security and behaviour-based detection. → AI Smart Contract Auditing, Smart Contract Security (2026 State)
  • Account abstraction’s endgame: EIP-8141 Frame Transactions is the most ambitious native-AA proposal yet, scoped for Hegota (~2027). → Account Abstraction
  • ZK at L1: ETHPrague’s ZK-EVM track confirms EthCC[9]‘s direction — optional proofs in Hegota, mandatory proofs as the long-term endgame. → ZK Proving Infrastructure

Connections

Open Questions

  • Does EZ survive contact with the realities of multi-rollup proof-system coordination, or does it collapse into “every chain joins one proof system anyway”?
  • The cost-of-intellect halving every 1.5 months — does that hold as Anthropic/OpenAI absorb the compute frontier, or do per-token costs flatten when state-of-the-art models hit hardware limits?
  • Local-mixing obfuscation is not yet provably secure. The challenge release (mid-May 2026) is the inflection point — if break attempts succeed, FHE remains the only viable path. If they don’t, this is the most important cryptography development of the cycle.
  • Privacy Pools v2 launches with the ASP gatekeeper preserved — is the “compliant privacy is default” thesis a permanent equilibrium, or a temporary regulatory accommodation that will fragment as MiCA / Clarity Act enforcement details settle?