Sei has published its Giga Manifesto, outlining a multi-stage overhaul of the Layer 1’s execution, storage and consensus architecture for higher-capacity onchain markets. According to the official Giga Manifesto, the central objective is to remove the single-proposer bottleneck that constrains how much transaction traffic can enter consensus at once. Sei is positioning that architecture around markets that generate continuous orders, cancellations and price updates rather than simple asset transfers.
Giga is not a separate blockchain and its headline performance targets are not currently available across Sei mainnet. The upgrade is arriving incrementally through changes to the existing network. Sei v6.6 introduced the first Ares execution and Eidos storage components on August 4, while additional improvements to both systems remain on the roadmap. The defining Autobahn consensus layer has not yet reached mainnet.
Autobahn Replaces the Single-Proposer Bottleneck
Autobahn is specified through SIP-6, which remains officially classified as Draft. The proposal gives every validator in the active committee its own transaction lane, allowing multiple validators to continuously propose batches instead of relying on one block producer for transaction data. Consensus would order compact references to certified lane blocks while availability, ordering and execution operate as overlapping pipelines.
The architecture also changes the network’s censorship boundary. A transaction could be submitted through several validator lanes, reducing the ability of a single proposer to exclude it. SIP-6 explicitly describes this as resistance to single-proposer censorship rather than complete censorship immunity. Byzantine validators can still censor transactions submitted only to their lanes, and Autobahn does not eliminate MEV or provide a universal fair-ordering guarantee.
Performance claims require a similar distinction. The Giga whitepaper targets more than 200,000 TPS, equivalent to roughly 5 gigagas per second, with sub-250 millisecond finality across a decentralized validator set. Sei says an internal 40-node devnet has already exceeded 5 gigagas per second and demonstrated ordering finality below 250 milliseconds. Those results are development benchmarks, while 200,000 TPS remains a target rather than demonstrated production throughput. SIP-6 itself states that performance depends on workload, topology, payload size, execution costs and deployment conditions.
That distinction is increasingly important as Layer 1 networks compete over latency and execution capacity. The recent Solana Alpenglow mainnet consensus handoff illustrates how protocol targets ultimately need production measurements before they can be treated as sustained network performance.
Tokenized Markets Drive the Giga Roadmap
Sei is framing the upgrade around a wider transition in which tokenized assets are moving beyond issuance into market infrastructure. Creating an onchain representation of a stock, fund or Treasury does not by itself create a functioning market. Execution quality, liquidity, custody, settlement and reliable transaction inclusion become increasingly important once tokenized assets begin trading actively.
That transition is already producing measurable activity elsewhere. Coinbase-linked tokenized stocks on Base have surpassed $4 billion in cumulative DEX turnover, while the broader expansion of tokenized equities is creating a competition over who controls their trading infrastructure. Giga is Sei’s attempt to compete at the execution layer of that emerging market rather than treating token issuance itself as evidence of adoption.
Sei says tokenized products associated with Apollo, BlackRock, Brevan Howard, Hamilton Lane, Laser Digital and Ondo are already represented on the network through partners including KAIO and Securitize. Dinari also plans to bring more than 700 tokenized U.S. equities to Sei. Those deployments establish potential demand for market infrastructure, but they do not demonstrate that Giga-scale order-book activity already exists on the network.
Autobahn must still progress through its governance and deployment process before that architecture reaches production. The roadmap calls for a Giga testnet before mainnet consensus activation, while SIP-6 remains subject to further specification and implementation work. Sei therefore already has pieces of Giga operating through Ares and Eidos, but the multi-proposer consensus system underpinning its 200,000 TPS thesis remains an unfinished stage of the roadmap.