Blockchain technology has long faced the challenge of the “scalability trilemma,” the difficulty of achieving high throughput, low latency, and robust decentralization all at once. Traditionally, blockchains like Ethereum have processed transactions sequentially, which creates bottlenecks, leads to high fees, and slows down confirmation times when networks are congested. To address these issues, the industry is increasingly turning to parallelism, the ability to process multiple transactions or blocks simultaneously. This shift is revolutionizing blockchain scalability, and the Sei network’s Giga upgrade stands out as a prime example of how parallelism can be fully realized in practice.
At its core, parallelism in blockchain can be implemented at several levels. Transaction-level parallelization involves grouping independent transactions, such as unrelated token swaps or NFT transfers, and processing them at the same time. This approach can dramatically increase transactions per second (TPS) and reduce confirmation times, though it requires advanced conflict detection to prevent issues when two transactions try to modify the same data. Block-level parallelization, meanwhile, allows multiple blocks to be proposed and validated concurrently, distributing the workload across the network and preventing congestion. However, this method demands more from node hardware, which can pose risks to decentralization. Parallelism can also be applied to smart contract execution, where contracts are run in parallel, often using optimistic rollups or zero-knowledge proofs. This enhances the responsiveness of decentralized applications (dApps) and reduces on-chain load, though it presents challenges in ensuring correctness and compatibility with existing developer tools.
Compared to the traditional sequential approach, parallelism offers significant advantages. While legacy blockchains like Ethereum typically achieve 15–30 TPS with confirmation times ranging from minutes to hours, parallel architectures such as those used by Solana and Sei can reach tens of thousands of TPS and deliver sub-second finality. This results in lower, more predictable gas fees and enables horizontal scaling, where adding more nodes increases capacity, rather than relying on ever-larger individual nodes.
Sei Network was among the early adopters of parallelism, utilizing optimistic concurrency control (OCC) to process transactions in parallel and only rolling back those that conflict. Research has shown that about 65% of Ethereum transactions can be parallelized without issues, making this approach highly efficient. Solana’s Sealevel engine takes a similar approach, leveraging GPU-powered parallelism to execute thousands of smart contracts at once and synchronize transactions efficiently. Ethereum is also moving in this direction, with Layer 2 rollups and sharding on its roadmap to enable parallel execution and exponential scaling.
The Sei Giga upgrade, launched in 2024, represents a quantum leap for blockchain parallelism. It introduces a multi-proposer consensus protocol, known as Autobahn, which allows multiple blocks to be proposed and processed in parallel. This eliminates the single-threaded bottleneck of traditional blockchains and results in up to 70 times faster block production and 50 times higher throughput, with the network capable of reaching up to 200,000 TPS and finality in under 400 milliseconds. The upgrade also features a parallel execution engine based on optimistic parallelism, where transactions are executed concurrently across CPU cores. If a conflict is detected, only the affected transactions are rolled back and retried, maximizing efficiency since about 65% of transactions do not require rollback.
A key innovation of Sei Giga is its custom EVM-compatible execution engine, built from scratch for parallelism and smart dependency tracking. This allows developers to deploy Ethereum dApps on Sei without rewriting their code, instantly benefiting from higher throughput and lower fees. The upgrade further enhances scalability with asynchronous state commitments, decoupling state proofs from transaction processing to reduce latency, and a flat key-value storage model with cryptographic accumulators. This storage design allows for faster parallel reads and writes, and tiered storage ensures that recent data is kept on fast SSDs while historical data is offloaded to distributed databases, lowering hardware costs for validators.
The performance improvements brought by Sei Giga are dramatic. While legacy EVM chains might handle around 100 TPS with confirmation times measured in seconds or minutes, Sei Giga achieves up to 200,000 TPS and sub-400ms finality, all while maintaining EVM compatibility. This brings blockchain performance closer to that of traditional Web2 infrastructure, making high-volume applications in DeFi, gaming, and consumer sectors viable on-chain. Since the Giga upgrade, Sei has processed billions of transactions and seen rapid adoption from developers and users alike.
Parallelism is no longer a theoretical improvement, but the foundation for the next generation of blockchains. Sei’s Giga upgrade demonstrates how combining multi-proposer consensus, parallel execution engines, and advanced storage solutions can deliver orders-of-magnitude improvements in throughput, latency, and cost, without sacrificing decentralization or security. As more networks adopt similar strategies, the blockchain industry is poised to support real-world, high-volume applications, fulfilling the original promise of decentralized technology.
