NEAR Protocol's Network Upgrade 2.13, scheduled for mainnet deployment in June 2026, introduces dynamic resharding — the capacity for the network to automatically split congested shards without human intervention. The upgrade also bundles FIPS-204 post-quantum-safe signing into the same release. ...
"The users of blockchain will be AI agents, not humans." — Illia Polosukhin, Co-founder, NEAR Protocol (CoinDesk, March 2026)
NEAR Protocol's Network Upgrade 2.13, scheduled for mainnet deployment in June 2026, introduces dynamic resharding — the capacity for the network to automatically split congested shards without human intervention. The upgrade also bundles FIPS-204 post-quantum-safe signing into the same release. NEAR's native token appreciated 115% over the 90-day window preceding late May 2026, with daily trading volumes peaking above $1 billion.
The protocol's cross-chain execution layer, NEAR Intents, crossed $20 billion in cumulative transaction volume on June 3, 2026, up from $10 billion in January 2026. In March 2026, Brave Browser integrated NEAR Intents natively, exposing the swap infrastructure to its 110 million users. Combined with an October 2025 inflation halving from 5% to 2.5% and a February 2026 fee-switch activation that routes 100% of Intents revenue into open-market NEAR purchases, the protocol is engineering a supply-contraction mechanism tied directly to usage growth.
The question now is whether the infrastructure can sustain the load that autonomous AI agents — NEAR's stated target user base — will generate at scale. Dynamic resharding is the protocol's answer.
Blockchain sharding divides a network into parallel processing units, each handling a subset of transactions independently. Most sharded networks operate with a fixed shard count — Ethereum 2.0 targets 64 shards, Polkadot supports a fixed set of parachains via slot auctions. NEAR's approach differs: the shard count adjusts programmatically based on demand.
Prior to this upgrade, NEAR operated with a manually managed shard count. The network increased from 6 to 8 shards in March 2025 and climbed to 9 shards later that year. Each adjustment required coordination among validators and governance overhead. Dynamic resharding eliminates that bottleneck.
The mechanism monitors the state size of each shard. When a shard crosses a predetermined capacity threshold, the protocol splits it deterministically — within two seconds, according to NEAR's technical documentation — and assigns validation to state witnesses without operator intervention. Co-founder Illia Polosukhin stated the system can scale the network beyond 70 shards, which he described as delivering "higher throughput than Visa."
The inverse also applies: underutilized shards can be merged, reclaiming validator resources. This bidirectional elasticity is architecturally distinct from any production Layer-1 network currently operating.
NEAR's sharding design, called Nightshade, has been rolling out incrementally since 2021. The core principle is that each shard generates and stores blocks as segments ("chunks") of a single blockchain, rather than operating as an entirely separate chain. Cross-shard reads and writes are handled natively at the protocol level, not through bridges or relay chains.
Dynamic resharding extends Nightshade by adding three capabilities:
Automated splitting. A shard hitting its state-size ceiling triggers a hierarchical split. The split is deterministic — every validator computes the same boundary — so no consensus round is needed for the split decision itself.
State witness validation. New shards are validated by state witnesses, nodes that verify shard state transitions without storing full shard history. This reduces the hardware requirements for supporting additional shards.
No governance overhead. Previous shard increases required proposals, votes, and coordinated node updates. Dynamic resharding removes the human coordination layer entirely.
NEAR's observed testnet throughput during load testing reached 4,135 transactions per second (TPS). The protocol's long-term claim — exceeding 1 million TPS with sufficient shards — remains unverified in production conditions. The gap between testnet peaks and mainnet sustained load is a standard caveat across all sharded architectures.
Network Upgrade 2.13 bundles a second significant change: the introduction of FIPS-204 (ML-DSA, formerly CRYSTALS-Dilithium) post-quantum-safe digital signatures. ML-DSA is a lattice-based algorithm standardized by NIST in August 2024 as part of its first batch of post-quantum cryptography standards.
NEAR's implementation allows any account holder to rotate their keys to a quantum-safe scheme in a single transaction, with no address migration required. This is enabled by NEAR's human-readable account model, which decouples account identity from the underlying cryptographic key pair.
The testnet for FIPS-204 signing is expected by end of June 2026. NEAR is among the first production Layer-1 networks to ship post-quantum signing as an opt-in feature, though it does not yet mandate quantum-safe keys for all transactions. A separate report on this site covers the broader post-quantum migration race across multiple blockchains.
NEAR Intents, the protocol's cross-chain execution layer, has become the primary driver of network economic activity. The system allows users or AI agents to express a desired outcome — "swap token A on chain X for token B on chain Y" — and solvers compete to execute the optimal path.
Key metrics as of early June 2026:
| Metric | Value | |---|---| | Cumulative volume (all-time) | $20.0 billion | | 30-day volume | ~$2.0 billion | | 7-day volume | ~$611 million | | 24-hour volume | ~$85.6 million | | Total swaps processed | 25+ million | | Chains integrated | 35+ | | Cumulative fees generated | ~$32 million |
The growth trajectory has been steep. NEAR Intents hit $5 billion in cumulative volume in November 2025, doubled to $10 billion by January 2026, and added the second $10 billion in roughly four months. In March 2026, Brave Browser integrated NEAR Intents natively into its wallet, exposing the swap layer to 110 million browser users across Bitcoin, Solana, Zcash, Cardano, Ethereum, Base, Arbitrum, and other EVM-compatible chains.
NEAR Intents is also the mechanism through which AI agents are expected to transact. Rather than requiring agents to hold tokens on specific chains or navigate bridge protocols, the intent system lets an agent express a desired action and delegates execution to specialized solvers. For autonomous agents performing high-frequency cross-chain operations — purchasing compute on one network, settling payments on another, storing data on a third — this abstraction layer reduces the integration complexity per chain to near zero.
NEAR executed two structural changes to its token economics within a six-month window:
October 30, 2025 — Inflation halving. Maximum annual inflation was reduced from 5% to 2.5%, cutting new token issuance from approximately 64 million NEAR per year to approximately 32 million.
February 23, 2026 — Intents fee switch activation. All fees generated through NEAR Intents are now routed into open-market purchases of the NEAR token. This creates buy pressure that scales linearly with transaction volume.
According to analysis by SVRN Research, the deflationary threshold — the daily Intents volume at which fee-driven buybacks exceed new token issuance — sits at approximately $177 million per day at current prices and fee rates. Current 24-hour Intents volume of approximately $85.6 million places the protocol roughly halfway to that threshold.
As of June 2026, NEAR's market capitalization stands at approximately $2.6 billion, with 44.9% of supply staked across 254 validators. The token trades at approximately $2.03, down from intra-period peaks near $2.27 in late May.
| Protocol | Sharding Model | Shard Count | Dynamic Adjustment | Cross-Shard Communication | |---|---|---|---|---| | NEAR | Nightshade (dynamic) | 9 → 70+ (target) | Automatic split/merge | Native protocol-level | | Ethereum 2.0 | Danksharding (data) | 64 (planned) | Fixed | Via rollups | | Polkadot | Parachains | Fixed (auction-based) | No (slot auctions) | Relay chain (XCM) | | MultiversX | Adaptive state | 3 active shards + metachain | Semi-automatic | Built-in |
Ethereum's current scaling strategy has shifted from execution sharding to a rollup-centric model, where the base layer provides data availability and security while rollups handle execution. This is architecturally different from NEAR's approach, which shards execution at the base layer.
Polkadot's parachain model requires projects to win slot auctions or purchase coretime to access shared security, introducing a capital cost to scaling that NEAR's permissionless shard-splitting avoids. MultiversX implements adaptive state sharding with semi-automatic adjustment but operates with a significantly smaller network (3 active shards plus a coordination metachain).
NEAR's dynamic resharding is, on paper, the most aggressive auto-scaling design among production Layer-1 networks. Whether it functions as described under adversarial mainnet conditions remains to be tested.
Unproven at scale. Dynamic resharding has not yet operated on mainnet. The gap between testnet demonstrations and production performance under adversarial conditions is non-trivial.
State growth concerns. Automatic shard splitting could lead to rapid state expansion if demand surges. Validator hardware requirements may increase faster than anticipated.
Centralization pressure. State witness validation lowers the bar for shard verification, but operating a full validator on a 70+ shard network may concentrate validation among well-resourced operators.
Deflationary threshold is distant. At $85.6 million in daily Intents volume versus the $177 million deflationary threshold, the protocol remains net inflationary. The tokenomics design is promising in structure but has not yet delivered on its core value proposition.
AI agent demand is speculative. The thesis that AI agents will become the primary users of blockchain infrastructure is reasonable but unverified. Current Intents volume is driven predominantly by human-initiated cross-chain swaps, not autonomous agent activity.
Competitive risk. Ethereum's rollup ecosystem processes significantly more value and has a larger developer base. Solana's monolithic high-throughput approach avoids sharding complexity entirely. NEAR's differentiation depends on the agentic commerce thesis materializing.
NEAR Protocol is shipping infrastructure designed for a user base that does not yet exist at scale: autonomous AI agents transacting across chains at machine speed. Dynamic resharding addresses the supply-side constraint — network capacity — while NEAR Intents addresses the demand-side interface — expressing and executing cross-chain actions.
The economic design is internally coherent. Inflation reduction tightens supply. Fee-switch activation ties buyback pressure to usage. Dynamic resharding removes the capacity ceiling that would otherwise throttle growth. If the pieces function as designed, the protocol creates a flywheel: more agents generate more Intents volume, which generates more fee revenue, which generates more buyback pressure, which improves token economics, which attracts more validators, which supports more shards.
The conditional word is "if." Dynamic resharding is untested on mainnet. The AI agent economy is nascent. The deflationary threshold has not been reached. What NEAR has built is a machine with the right gears. Whether those gears turn under real load is the next chapter.