NEAR Protocol's v2.13 network upgrade, scheduled for June 2026, introduces two capabilities that address separate but converging risks: dynamic resharding, which automates the splitting of state shards based on demand, and FIPS-204 (ML-DSA) post-quantum cryptographic signing, which replaces Ed255...
"Dynamic resharding is coming soon to NEAR Protocol — unlocking a new level of scalability to power the growth of NEAR Intents and agentic commerce. As more AI agents transact onchain, we'll need more scalable infrastructure. This upgrade allows NEAR to scale to 70+ shards, which would have a higher throughput than Visa." — Illia Polosukhin, Co-founder, NEAR Protocol
NEAR Protocol's v2.13 network upgrade, scheduled for June 2026, introduces two capabilities that address separate but converging risks: dynamic resharding, which automates the splitting of state shards based on demand, and FIPS-204 (ML-DSA) post-quantum cryptographic signing, which replaces Ed25519 keys with lattice-based signatures resistant to quantum attack. Together, the upgrades represent the most technically significant change to NEAR's Nightshade architecture since mainnet launch.
The market response has been immediate. NEAR's token gained 28% in 24 hours following the May 22 announcement, reaching $2.24 on a 161% spike in daily trading volume to $1.02 billion. The Bitwise Near Staking ETP pulled in $7 million in a single week. As of mid-June, NEAR trades around $2.12 with a market capitalization of approximately $2.7–3.2 billion (depending on the data source), circulating supply fully diluted at 1.29 billion tokens.
The upgrades arrive as NEAR Intents, the protocol's intent-based cross-chain execution layer, crossed $20 billion in cumulative transaction volume on June 3, 2026 — double its January 2026 figure and quadruple its November 2025 level. The infrastructure question is no longer hypothetical: NEAR's current 9 shards process approximately 80 TPS under normal load, with a recorded peak of 4,135 TPS. Dynamic resharding is engineered to remove the manual bottleneck that constrained shard expansion to date.
NEAR launched with its Nightshade sharding architecture in 2020. The design partitions blockchain state across multiple parallel shards, with validators producing "chunks" for each shard and assembling them into blocks. The theory was sound. The execution has been incremental.
Phase 1, launched in 2021, introduced basic chunk-only producers. Nightshade 2.0 followed in 2024, introducing stateless validation — a shift that eliminated the requirement for validators to store shard state locally. Instead, validators retrieve "state witness" data from the network to validate state changes. This removed a significant hardware barrier to participation.
The shard count has grown slowly. NEAR expanded from 4 to 6 shards during 2024, adding approximately 50% transaction capacity. In March 2025, the count rose to 8. By late 2025, it reached 9. A May 2025 upgrade pushed block times down to 600 milliseconds with 1.2-second finality.
Each expansion, however, required the same manual process: weeks of validator coordination, a governance vote, and a staged rollout. The human bottleneck limited NEAR's ability to respond to demand spikes. Adding a single shard took weeks. In a network targeting AI agent commerce — where transaction bursts can be unpredictable and sustained — weeks is operationally unacceptable.
The v2.13 upgrade replaces this manual process with deterministic, automated shard splitting. The mechanism works as follows:
Threshold Detection. The protocol continuously monitors each shard's state size. When a shard hits a predefined threshold — whether measured by state size or congestion — the splitting process triggers automatically.
Deterministic Split. The system takes a snapshot of the shard's flat storage, iterates over all entries, and builds two child tries by routing each parent entry to the appropriate child shard. The process completes within a single epoch, meaning state sync of the parent shard, the resharding itself, and catchup of the children shards all occur in one pass.
State Witness Assignment. New validators are assigned to the child shards through the existing state witness mechanism. No manual coordination is required. The split executes in under two seconds of network time.
Merge Capability. When demand subsides, shards can theoretically merge back. This bidirectional elasticity distinguishes the system from simple one-way splitting.
The practical result: NEAR's shard count is no longer gated by governance. The network can scale from its current 9 shards toward 70+ shards without human intervention. At 70 shards, NEAR projects throughput exceeding Visa-level capacity — a claim that remains to be validated under production load but is architecturally plausible given the parallelization model.
This positions NEAR differently from its peers. Ethereum has deferred sharding indefinitely, relying instead on Layer 2 rollups for scaling. Solana processes 2,000–4,000 TPS on a single monolithic chain with parallel runtime execution. NEAR's approach is orthogonal: horizontal scaling through automated state partitioning, with each shard operating as a semi-independent processing unit.
The same v2.13 upgrade bundles NEAR's first post-quantum cryptographic feature. The protocol will integrate FIPS-204 (ML-DSA, formerly known as CRYSTALS-Dilithium), a lattice-based digital signature standard formally approved by the U.S. National Institute of Standards and Technology (NIST) in August 2024.
Current blockchain signatures — including Ed25519 (used by NEAR, Solana, and others) and secp256k1 (used by Bitcoin and Ethereum) — are mathematically vulnerable to Shor's algorithm running on a sufficiently powerful quantum computer. No such computer exists today. The timeline for one is debated, but most estimates place the threat window within the next 10–15 years.
NEAR's implementation takes advantage of an architectural feature most chains lack: human-readable account names with rotatable access keys. On Bitcoin or Ethereum, migrating to a new signature scheme means migrating to a new address — a coordination nightmare across exchanges, contracts, and counterparties. On NEAR, account holders can upgrade their signing key via a single transaction without changing their account identity. The old key is replaced; the account name persists.
Near One, the protocol's development arm, deployed ML-DSA signing to the testnet during Q2 2026. Mainnet rollout will follow security audits and community coordination. The upgrade is opt-in at the account level — users can upgrade when ready, and applications require no modification.
NEAR is not the only chain pursuing post-quantum protections, but it appears to be among the first to deploy a NIST-standardized algorithm to a production-adjacent environment. The practical value today is limited. The strategic value — eliminating a future migration crisis — is potentially significant.
The technical upgrades exist in the context of accelerating demand on NEAR infrastructure. NEAR Intents, the protocol's cross-chain execution layer, has emerged as the primary driver.
The volume trajectory:
The system has processed over 25 million swaps. Cumulative fee revenue reached approximately $32 million, with recent weekly annualized rates suggesting $36–58 million in annual revenue depending on the measurement period.
NEAR Intents operates on an intent-based architecture: users (or AI agents) express a desired outcome — swap token A on Chain X for token B on Chain Y — and competing solvers determine the optimal execution path. The system supports 25–70+ chains depending on direct versus indirect connections, with wallet compatibility extending to hardware devices like Ledger.
The AI agent angle is where NEAR's scaling story converges with its demand story. NEAR co-founder Illia Polosukhin has publicly stated that "the users of blockchain will be AI agents" and characterized blockchain as "the back end" to AI's "front end." The protocol has branded $NEAR as "the currency of agents" and launched Shade Agents — autonomous agents capable of executing DeFi strategies or booking real-world services using confidential cross-chain infrastructure.
Intent-based architectures are structurally suited for AI-driven interactions because agents can express desired outcomes without encoding the mechanical steps of cross-chain execution. Whether this theoretical fit translates into sustained transaction volume from AI agents remains unproven. The $20 billion in Intents volume to date is primarily human-driven.
NEAR's token economics shifted materially in Q4 2025 when a "Halving Upgrade" reduced maximum annual inflation from 5% to 2.5%. Combined with fee burn mechanisms, this moves the protocol closer to — though not yet at — a deflationary threshold.
Key metrics as of mid-June 2026:
The token trades at roughly 90% below its all-time high, which provides context for the percentage gains following the upgrade announcement. A 28% move on a $2 token is $0.50, not a fundamental re-rating.
Institutional interest is measurable but modest. The Bitwise Near Staking ETP accumulated $7 million in weekly inflows around the announcement period. This is directionally meaningful — it signals institutional appetite for staking exposure — but small relative to institutional flows into Bitcoin or Ethereum products.
Untested at scale. Dynamic resharding has not operated under sustained production load with dozens of shards. The mechanism's behavior during extreme demand spikes, network partitions, or adversarial conditions remains theoretical.
TVL disconnect. At $157.6 million, NEAR's TVL is modest for a network with 46 million monthly active users and 8+ million daily transactions. This suggests much of the activity may be low-value or bot-driven, which could undermine the economic sustainability argument.
Post-quantum timeline ambiguity. While NEAR is deploying ML-DSA to testnet, the mainnet timeline is undefined beyond "after audits." Quantum computing threats remain distant, and no chain faces immediate risk. The upgrade's value is prophylactic rather than responsive.
AI agent demand is speculative. The narrative that AI agents will become primary blockchain users is plausible but unconfirmed. Current Intents volume is human-generated. If AI agent adoption fails to materialize at scale, the resharding capacity may exceed actual demand.
Competitive landscape. Ethereum's Layer 2 ecosystem handles an order of magnitude more value. Solana's monolithic approach offers comparable speed at lower complexity. NEAR's sharding advantage is real but does not exist in a vacuum.
NEAR's v2.13 upgrade addresses two distinct infrastructure challenges: scaling capacity without governance friction, and future-proofing cryptographic security against quantum threats. The dynamic resharding mechanism is the more immediately consequential feature, removing a manual bottleneck that has constrained shard expansion since mainnet launch.
The protocol's bet is that AI agent commerce will generate unpredictable, bursty demand that requires elastic infrastructure. The $20 billion in Intents volume provides evidence of growing cross-chain activity, though the AI agent thesis remains forward-looking. NEAR's current TVL-to-activity ratio suggests the network's economic density — value captured per transaction — has room to improve.
What is measurable: NEAR is shipping production-grade infrastructure changes on a defined timeline. Dynamic resharding moves from theory to deployment in June 2026. Post-quantum signing has reached testnet. Whether the demand materializes to justify the capacity is a separate question — one the market has not yet definitively answered.