NEAR Protocol is betting its future on autonomous AI agents that transact at machine speed, positioning itself as the settlement layer for a coming agent economy. A June 2026 upgrade called dynamic resharding is the centerpiece of this strategy, designed to automatically scale the blockchain to handle unpredictable bursts of activity. However, the thesis faces a real complication: NEAR's active user numbers are falling, suggesting the agent economy has not yet arrived.
The AI-agent thesis
NEAR's strategy rests on a specific vision: an on-chain economy where AI agents buy compute, settle payments, and execute trades automatically, without human intervention. These agents would generate transaction bursts that would overwhelm conventional blockchains, causing fee spikes and slow confirmations. NEAR aims to be the infrastructure built to absorb that load, with its token branded as "the currency of agents" and the network as "a unified commerce layer."
The requirement for automatic, instant scalability is the technical heart of the bet. NEAR's co-founder, who co-authored the 2017 paper introducing the transformer architecture behind today's large language models, has framed the protocol as fundamental infrastructure for AI-driven commerce.
Dynamic resharding: the June upgrade
The centerpiece of NEAR's bet is dynamic resharding, part of network release 2.13. Sharding splits a blockchain into parallel partitions called shards, each processing transactions independently. Until now, adding a new shard required weeks of validator coordination and a governance vote. Dynamic resharding removes that human bottleneck: when a shard fills past a defined threshold, it automatically splits into more shards, adding capacity in real time without human intervention.
NEAR's leadership says the upgrade will let the network scale to many dozens of shards, with throughput exceeding major payment networks. The same upgrade also adds post-quantum-secure signatures, allowing users to rotate to quantum-safe keys, a forward-looking security measure.
The supporting pieces
Beyond dynamic resharding, NEAR has assembled other components around the AI-agent thesis. NEAR Intents provides a cross-chain settlement system, letting agents settle activity across multiple blockchains. Privacy tooling and tokenomics that tie usage to the token's value round out the strategy. Together, these pieces form a coordinated effort to become essential infrastructure for the agent economy.
The complication: falling usage
Despite the ambitious narrative, NEAR's on-chain usage tells a different story. Active users have been declining, indicating that the agent economy has not yet materialized. This gap between the soaring narrative and actual usage is a serious consideration for any observer. The thesis is coherent and the technology is real, but the market has not yet embraced it at scale. NEAR's bet is a concrete expression of the AI-crypto crossover, and its success depends on whether the agent economy arrives as predicted.