Justin Drake (Ethereum Foundation) - gigagas L1
Justin Drake
Key takeaways
- GigaGas Roadmap: Justin Drake argues Ethereum L1 can reach one gigagas per second, implying roughly 10,000 TPS, by raising the gas limit 3× annually for six years instead of the four-year plan in EIP-7938.
- ZK-Secured L1: Converting Ethereum into a ZK-L1 lets validators verify succinct SNARK proofs instead of executing blocks, enabling stateless nodes that run on a $7 Raspberry Pi Pico while improving decentralisation.
- Real-Time Proving Breakthrough: Four separate teams have already produced single-slot proofs; 13 open-source ZKVM teams aim to hit sub-12-second latency this year, making once “impossible” real-time proving a near-term reality.
- New Hardware Economics: Home-grade GPU clusters capped at $100 k and 10 kW become the reference target. Software and ASIC advances are expected to deliver another 10× improvement each, and collaborative “multi-premise” proving can multiply capacity further.
- Shifting Bottlenecks: As proof verification replaces full execution, scale pressure moves from decentralised nodes to centralised services such as block builders, RPCs, and explorers, which must adopt custom high-throughput databases like QMDB.
- Security via Diversity: At least ten ZKVMs should be proving mainnet blocks by year-end. Validators can run several clients in parallel and accept a four-out-of-five consensus, echoing today’s multi-client strategy.
- Downstream Impact: A stronger L1 with shared sequencing and real-time proofs could boost L2 composability, raise TVL on mainnet, and eventually justify enshrining a ZK-friendly ISA such as RISC-V beneath the EVM.
On GigaGas Scaling and the “Gigas Era”
Justin Drake frames a simple metric—gigagas per second—as Ethereum’s new ambition. By automatically voting the gas limit higher each hard fork, validators could triple capacity yearly and cross one gigagas in six years. That path maintains predictability for application builders while remaining compatible with existing tooling. Crucially, the plan assumes ZK proof verification replaces local execution, so more gas no longer burdens ordinary nodes.
On ZKVMs and Real-Time Proving
Turning Ethereum into a ZK-L1 requires ZK execution clients that output validity proofs every slot. The concept of delayed execution lets the prover use the full 12-second interval between block N and N + 1. Once considered “moon math,” real-time proving is now demonstrated by Succinct, RISC Zero, ZISC, and Airbender, each publishing open-source GPU code. Performance is converging on sub-10-second latency and sub-1 MB proofs with 100-bit security, clearing the bar for mainnet adoption.
On Validator and Prover Hardware Economics
A validator that only verifies proofs and samples data can run on a one-watt Raspberry Pi Pico; proof verification takes mere milliseconds. The heavy lift moves to provers. Drake sets an affordability envelope of $100 k CAPEX and 10 kW, the power draw of a residential EV charger. Current GPU clusters need roughly 10,000× more power than pure execution, but expected 10× software gains plus 10× ASIC gains shrink that gap. If still short, multiple households can cooperate, pooling six or more small clusters to meet one-slot deadlines.
On Shifting Bottlenecks and Database Innovation
With validation lightened, throughput is now limited by data pipelines in block builders, explorers, and RPC providers. Traditional key-value stores like LevelDB and MDBX struggle above 100 million gas. Coinbase’s in-house engine and LayerZero’s QMDB show 10× faster read-write speeds by tailoring storage layouts to the Ethereum state tree. Migrating to such purpose-built databases becomes the next scaling prerequisite.
On ZKVM Diversity, Governance, and Risk Management
Nineteen open-source ZKVM projects are active; ten already prove full L1 blocks. Diversity is not luxury but risk containment: every complex prover is assumed buggy. Operators can imitate the Vouch model, running five distinct ZKVMs and accepting the majority output, while pairing each ZKVM with a different execution client to preserve today’s EL diversity. Selecting the canonical five will hinge on performance, open licensing, proof size, and post-quantum resilience.
On Broader Architectural and Ecosystem Implications
A high-throughput ZK-L1 enhances the economic gravity of mainnet, attracting TVL that feeds rollups. Shared sequencing between L1 and L2s becomes feasible, unlocking synchronous composability without trusted bridges. Looking three to six years out, Vitalik Buterin’s idea to enshrine a ZK-friendly ISA, with RISC-V the leading candidate, could simplify the stack and cut gas fees tenfold by bypassing legacy EVM quirks, though LLVM monoculture risks must be mitigated.
Key actionable insights
- Track real-time proving benchmarks; sub-12-second, sub-1 MB proofs with open-source GPU stacks will signal readiness for mainnet integration.
- Monitor adoption of Ethereum-specific databases; migrations by major RPC providers will indicate whether data throughput keeps pace with rising gas limits.
- Watch for governance proposals linking specific ZKVM–execution-client pairs and multi-client verification rules, as these choices define future consensus safety margins.