BIP Austin digital publishing platform

collapse
Home / Daily News Analysis / Sui tunnels hit 40.6M TPS in live AI agent test

Sui tunnels hit 40.6M TPS in live AI agent test

Oct 11, 2026  Twila Rosenbaum 17 views
Sui tunnels hit 40.6M TPS in live AI agent test

Sui Tunnels Hit 40.6M TPS in Live AI Agent Test

Key facts:

  • Sui's programmable offchain tunnels reached 40,614,180 transactions per second (TPS) in a live stress test.
  • The demonstration took place Wednesday at Sui Basecamp in Singapore.
  • It surpassed a previous Sui tunnel record of about 6.1 million TPS set in July.
  • More than 10,000 tunnels were opened on Sui mainnet during the test.
  • Activity spanned payments, games, and chat applications.
  • CertiK served as an independent auditor and is reviewing the test data.
  • Adeniyi Abiodun, co-founder and chief product officer at Mysten Labs, said agents, not people, drive demand for such throughput.

Sui's programmable offchain tunnels processed more than 40 million transactions per second during a live stress test designed to simulate high-frequency activity between artificial intelligence agents. The layer-1 blockchain said the demonstration peaked at 40,614,180 TPS, a dramatic increase from the roughly 6.1 million TPS recorded in a similar test in July.

The test was conducted Wednesday at Sui Basecamp in Singapore, an event that brings together developers, researchers, and ecosystem participants. During the exercise, more than 10,000 tunnels were opened on Sui mainnet. Activity flowing through those tunnels spanned payments, games, and chat applications, according to the network. The transactions were processed offchain through Sui tunnels, which are programmable channels that settle to the Sui blockchain when they are closed.

The milestone highlights a growing focus on offchain scaling solutions as blockchains prepare for a future in which autonomous software agents may transact with one another continuously. Sui said the tunnels are designed to handle high-frequency activity, including interactions between AI agents, without requiring every transaction to be processed directly onchain. That distinction is important: the 40.6 million TPS figure refers to activity inside the offchain tunnels, not to the base-layer throughput of the Sui blockchain itself.

What Are Sui Tunnels?

Sui tunnels are programmable offchain channels that allow participants to conduct a high volume of transactions away from the main blockchain. Instead of broadcasting every transfer, payment, or interaction to the entire network, the parties involved can transact within a tunnel and then settle the final state to Sui mainnet when the tunnel is closed. This model is conceptually related to payment channels and state channels, which have been explored in the blockchain industry for years as a way to reduce congestion and fees on base layers.

The key difference, according to Sui, is programmability. The tunnels are not limited to simple token transfers. They can be configured to support more complex interactions and application-specific logic. That flexibility could make them useful for use cases that require rapid, repeated microtransactions, such as gaming, machine-to-machine payments, streaming payments, and conversational AI applications.

Programmable tunnels differ from simple payment channels because they can embed logic. A payment channel typically supports two parties exchanging signed balance updates. A programmable tunnel can, in principle, support multi-party applications, conditional transfers, game state, or chat messages. That broader scope is what makes the AI-agent use case plausible. An AI agent might need to pay for data, purchase compute, settle a bet, or exchange messages with another agent. If each of those actions required an onchain transaction, the latency and cost could be prohibitive. Tunnels allow the agents to interact at high speed and settle later.

In the live test, the tunnels were used across payments, games, and chat applications. Opening more than 10,000 tunnels on Sui mainnet during the demonstration also tested the network's ability to coordinate a large number of offchain channels and then settle their activity. The result was a peak throughput of 40,614,180 TPS, which Sui described as a record for its tunnel technology.

Why AI Agents Are Driving the Throughput Race

The test's focus on AI agents reflects a broader shift in how blockchain developers think about demand. For most of crypto's history, scalability discussions centered on human users: retail traders, gamers, collectors, and institutions. Human-driven activity tends to be bursty and limited by human speed. Even the busiest payment networks in the world process transactions at rates that are modest compared with what machines could generate.

Adeniyi Abiodun, co-founder and chief product officer at Mysten Labs, the original contributor to Sui, framed the milestone in those terms. He said:

"No market in the world needs 6 million transactions per second, let alone 40 million: people just don't move that fast. Agents do."

That quote captures the core argument for high-throughput offchain systems. AI agents can operate continuously, respond to market conditions in milliseconds, and interact with other agents without human intervention. If those agents begin to use blockchain infrastructure for payments, coordination, and record-keeping, the resulting transaction volume could dwarf what human users produce. Sui is positioning its tunnel architecture as a way to accommodate that machine-scale activity while still using a layer-1 blockchain for settlement and security.

Industry analysts have increasingly argued that machine-to-machine payments will be a major driver of blockchain adoption. Autonomous agents can negotiate prices, verify services, and execute transactions without human approval. In such a system, throughput and latency become critical. A human may tolerate a few seconds for a payment confirmation; an AI agent optimizing a market strategy or coordinating with other agents may need sub-second finality and the ability to handle thousands of transactions per second. The 40.6 million TPS figure is an attempt to show that the infrastructure can scale to that level, at least within an offchain environment.

Background on Sui and Mysten Labs

Sui is a layer-1 blockchain developed by Mysten Labs, a company founded by former engineers who worked on Meta's Diem project. The network launched its mainnet in 2023 and uses the Move programming language, which was originally created for Diem. Sui's design emphasizes parallel execution, an object-centric data model, and low-latency transaction processing. Those features are intended to allow the network to handle many independent transactions at once rather than processing them in a single sequential line.

Mysten Labs has positioned Sui as infrastructure for applications that need speed and scalability, including gaming, payments, and consumer-facing services. The network has also pursued innovations such as sponsored transactions, zkLogin, and programmable transaction blocks. The tunnel project extends that strategy by moving high-frequency interactions off the main chain while preserving the ability to settle on Sui.

Sui's mainnet is secured by a delegated proof-of-stake consensus mechanism and validators. The network has faced the same market pressures as other layer-1 blockchains, including competition for developers, users, and liquidity. By focusing on performance and novel architectures, Sui has sought to differentiate itself. The tunnel test is part of a broader narrative that Sui can serve as settlement layer for high-frequency applications, even if those applications run offchain for most of their operations.

The previous tunnel test in July reached about 6.1 million TPS. The jump to more than 40 million TPS in a matter of months suggests that the team has been able to increase the capacity of the offchain channel system significantly. However, the comparison also underscores how early the technology is. Moving from a lab-style test to production-scale use will require sustained reliability, security, and adoption.

Offchain TPS vs. Onchain TPS

It is important to distinguish between offchain throughput and onchain throughput. When a blockchain advertises a TPS number, it usually refers to transactions processed directly by the base layer and validated by its consensus mechanism. Those transactions compete for blockspace and are subject to the network's fee market and confirmation times. Offchain tunnels, by contrast, allow participants to transact without touching the base layer for every transaction. The base layer is used primarily for opening and closing tunnels, or for resolving disputes.

This architecture can achieve very high throughput because it bypasses the consensus bottleneck for most transactions. But it also introduces different trust assumptions. Participants may need to lock funds or collateral, and the system must ensure that the final settlement reflects the correct state of the tunnel. Security depends on the design of the tunnel protocol, the underlying blockchain's ability to enforce settlements, and the availability of participants or watchtowers to monitor for fraud.

Measuring TPS is also not standardized. Some tests count simple transfers, while others include more complex operations. Some report peak throughput, while others report sustained throughput. The 40,614,180 TPS figure is a peak number recorded during a live stress test. Without the full methodology and audit, it is difficult to compare directly with other networks' TPS claims. The upcoming CertiK report may provide more clarity on how transactions were counted, what hardware was used, and whether the test can be replicated.

Sui's test was designed to demonstrate what the tunnel system can do under controlled conditions. The network said the transactions were processed offchain, and more than 10,000 tunnels were opened on mainnet. That combination shows both the offchain capacity and the onchain coordination required to manage a large number of channels. It does not mean that Sui's base layer can process 40.6 million TPS, nor does it mean that every transaction in the test would have been settled individually onchain.

Independent Audit and What Comes Next

CertiK served as an independent auditor during the demonstration and is reviewing the test data. A report is expected in the coming days. Independent verification is a critical step for any high-profile performance claim. Stress tests can be affected by hardware configurations, network conditions, software optimizations, and the specific workload used. An auditor's review can help clarify what was measured, how the tunnels were configured, and whether the results are reproducible.

The audit may also shed light on the security properties of the tunnel system. Offchain channels have long been praised for their efficiency, but they have also faced challenges related to liquidity, capital efficiency, and user experience. If Sui's tunnels can support programmable interactions across payments, games, and chat applications, they may open new possibilities for applications that need both speed and blockchain-based settlement.

The involvement of CertiK suggests that Sui wants the result to be taken seriously by the security community. Independent audits are common in crypto for smart contracts and blockchain protocols, but auditing a live stress test is less common. The report could validate the architecture or raise questions about its limitations. If the results hold up, they could encourage more developers to experiment with programmable tunnels for AI, gaming, and payment applications. If not, the milestone may be seen as a marketing exercise rather than a technical breakthrough.

For developers, the milestone is a signal that Sui is investing in infrastructure for a future of automated, high-frequency activity. For investors and observers, it is a reminder that scalability claims should be evaluated carefully, especially when offchain systems are involved. The 40.6 million TPS figure is striking, but its significance will depend on how the technology performs outside a controlled test and whether it can attract real usage.

The broader context is a rapidly evolving competition among layer-1 and layer-2 networks to support AI-driven economies. As artificial intelligence agents become more capable, they may need payment rails, identity systems, and coordination mechanisms that operate at machine speed. Sui's tunnel test is one attempt to prepare for that scenario. Whether it becomes a widely used standard or remains a specialized tool will depend on adoption, developer interest, and the results of independent reviews like the one underway from CertiK.

Regardless of the audit outcome, the test reflects a larger trend: blockchains are preparing for a world in which software agents, not just humans, are the primary users. That shift will place new demands on scalability, privacy, and interoperability. Offchain tunnels are one response to those demands. Their success will depend on whether they can provide the speed of centralized systems while preserving the security and verifiability of a blockchain. The coming days should bring more data on how close Sui's tunnel system is to that goal.


Source:Cointelegraph News


Share:

Your experience on this site will be improved by allowing cookies Cookie Policy