Base announced on May 4 that its Azul upgrade would integrate Succinct's SP1 zero-knowledge prover alongside its existing TEE proof system, with mainnet activation planned for May 13. The change cuts withdrawal finality from a multi-day challenge window to roughly one day — and makes Base the largest single Ethereum operator to put ZK proofs in production.
Base announced on May 4 that its Azul upgrade would migrate the chain from a fault-proof optimistic rollup to a multi-proof system combining trusted-execution-environment proofs with zero-knowledge proofs generated by Succinct's SP1 virtual machine. Mainnet activation is planned for May 13 if testnet validation completes on schedule. Base secures roughly $7.4 billion of deposits today, which makes it the largest single Ethereum L2 operator to put ZK proofs in production.
The technical mechanic is the part that matters for finality. Optimistic rollups settle with a multi-day challenge window during which any party can dispute a state root and force a fraud-proof exercise — a window that, for Base today, runs to seven days. Until that window closes, withdrawals to Ethereum mainnet are not final. Azul replaces that with a multi-proof layout: when a TEE proof and a ZK proof both back the same state proposal, the chain treats finality as cryptographic rather than economic, and the challenge window collapses to roughly one day. That is a structural shift, not an optimisation.
It is also a quiet repudiation of the assumption Base launched on in 2023. The chain shipped as part of the Optimism Superchain, sharing a fault-proof system with OP Mainnet and inheriting the optimistic-rollup design Coinbase publicly endorsed at launch. Two and a half years on, the same team has chosen to layer cryptographic validity over the top — partly for finality, partly for security, and partly because the ZK side of the L2 stack has matured faster than the optimistic side has fixed its withdrawal lag. The Optimism alignment has not been cancelled; it has been quietly downgraded.
Succinct's SP1 is the more interesting pick. SP1 is a general-purpose zkVM that lets developers prove arbitrary Rust programs rather than constrained DSL circuits, which is what makes it a credible substitute for the bespoke proving stacks that dominated 2024. Base is also not stopping at one prover — the Azul design accepts multiple proof types in parallel and treats agreement between them as the security property. That structure is meant to dilute the risk that a single proving stack carries an undiscovered bug and breaks finality for everyone. It also lets Base swap provers later without another forklift upgrade.
The framing from inside Base is direct. Brian Trunzo, the L2's chief growth officer, called the SP1 selection "the single largest vote of confidence that ZK is indeed the endgame for Ethereum scaling" — a line that sounds like an architecture pronouncement but is also a competitive shot at zkSync, Linea and Scroll, the chains that have been calling themselves zkEVMs for two years. Base now claims the largest TVL of any chain running a zero-knowledge prover in production, and it does so without ever rebranding itself as ZK-anything. The label was always less interesting than the deposit base.
For users the immediate change is the withdrawal experience. A seven-day challenge window has been the binding constraint on Base bridge UX since launch — it is the reason every meaningful flow back to Ethereum routes through a third-party fast-withdrawal aggregator that fronts the liquidity and charges a fee. Cutting the window to roughly 24 hours does not eliminate fast bridges; it does meaningfully compress their margin. It also shifts the security question. With cryptographic finality, the bridge UX gets cheaper. Without it, the only thing standing between user funds and a malicious sequencer is the social-coordination layer of the fault-proof game — a layer that, as the Kelp DAO incident demonstrated, can absorb a $292 million exploit but cannot prevent one.
There is a competitive subtext to the timing. Coinbase pushed Base hard through 2024 and 2025 as the default settlement layer for its retail product set; the chain's TVL grew faster than any optimistic rollup in the cohort and overtook OP Mainnet in mid-2025. The Azul upgrade is the first independent network upgrade Base has executed without coordinating it through the broader Superchain release process. The implication, deliberate or not, is that Base is now setting its own scaling roadmap rather than inheriting Optimism's.
What Azul does not do is solve the bigger question hanging over every Ethereum L2 — sequencer centralisation. Base still runs a single sequencer operated by Coinbase, and the cryptographic finality the upgrade introduces does not change that. It changes how quickly users can exit if Coinbase ever pauses the sequencer, which is a real reduction in trust assumption, but it leaves the live-ordering monopoly untouched. The roadmap document attached to the upgrade promises decentralised sequencing in a future phase. The roadmap document said the same thing eighteen months ago.
What ships on May 13, if it ships, is the largest single migration of Ethereum L2 deposits onto ZK validity in the network's history.
