NEAR Protocol has marked a major advancement in blockchain performance after reporting that it surpassed one million transactions per second in a publicly verifiable benchmark. The test, which employed actual NEAR core code and realistic usage patterns, was conducted using cost-efficient Google Compute Engine C4D machines distributed across 70 shards. This demonstration has been presented as proof that the network’s sharding design can scale to extremely high transaction volumes without compromising decentralization or cost efficiency.
Project developers noted that the result strengthens confidence in NEAR’s capacity to support the expanding ecosystem of applications built on the network. This includes high-throughput decentralized applications and emerging on-chain AI frameworks that require predictable block production, low latency, and stable operational costs. The team emphasized that the performance milestone illustrates that a blockchain can meet global workloads while retaining accessibility for validators and affordability for users.
A central factor in NEAR’s performance is its sharded architecture, which distributes execution, storage, and state across multiple parallel shards. This model contrasts with monolithic blockchain designs that concentrate all activity into a single chain. NEAR’s approach enables throughput to scale linearly as additional shards are added. It also supports decentralized participation, since validators are able to run nodes with modest hardware requirements without being overwhelmed by network load.
Nightshade 2.0 Upgrade and Deterministic Accounts Enhance Capability
The network’s performance has been elevated by a recent upgrade known as Nightshade 2.0, which introduced stateless validation and delivered significant speed improvements. The mainnet currently operates with nine shards, a block time of approximately 600 milliseconds, and a finality time of roughly 1.2 seconds. These elements have been described as key to ensuring consistent and efficient execution.
NEAR has also implemented a Deterministic Accounts system, designed to streamline shard execution for smart contracts. This mechanism supports use cases such as cross-chain DeFi activity and large-scale e-commerce systems, where predictable routing and high-volume processing are essential.
This 1M TPS milestone ensures scalable infrastructure for NEAR Intents and proves NEAR's sharded design is the future-proof architecture for high-load applications and a high-volume AI economy.
— NEAR Protocol (@NEARProtocol) December 8, 2025
The benchmark test was intentionally structured to match real-world conditions as closely as possible rather than rely on artificial lab setups. Each shard managed one million accounts, and the flow of traffic across shards was designed to replicate typical network patterns. Only native token transactions were processed, and all entries were inserted directly into the mempool. Simulated traffic was distributed across three separate Google Cloud regions to reflect geographically diverse usage.
Verifiable Performance Demonstrates Readiness for AI and DeFi
During testing, NEAR continued to process more than one million transactions per second across 70 shards using C4D virtual machines powered by 8-core CPUs. These machines cost under one thousand dollars per month, underscoring the affordability of the infrastructure used. Developers indicated that several optimizations contributed to the performance, including improvements in execution speed, signature verification processes, database and tree structure efficiency, and consensus pathways.
The scripts used in the benchmark have been made available as open source resources, allowing independent parties to verify the results. This level of transparency is intended to reinforce confidence in the network’s capabilities as it moves toward supporting more complex, future-focused applications.
The milestone highlights NEAR’s readiness for next-generation blockchain demands, including on-chain AI computing, high-speed decentralized finance activity, and enterprise-level software ecosystems. According to project contributors, future research will explore dynamic re-sharding, advanced sharded smart contract systems, and further performance enhancements. These efforts aim to maintain scalability and affordability while preserving the network’s decentralized structure as demand for blockchain-based applications continues to rise.