NEAR Protocol's Network Upgrade 2.13, expected to deploy on mainnet in June 2026, introduces two structural changes to the protocol: dynamic resharding and post-quantum cryptographic signing (FIPS-204 / ML-DSA). The upgrade eliminates the manual governance process previously required to add shard...
NEAR Protocol's Network Upgrade 2.13, expected to deploy on mainnet in June 2026, introduces two structural changes to the protocol: dynamic resharding and post-quantum cryptographic signing (FIPS-204 / ML-DSA). The upgrade eliminates the manual governance process previously required to add shards, replacing it with automatic, deterministic shard splitting when state size thresholds are reached. The protocol claims sub-two-second split execution with no validator coordination required.
The upgrade arrives as NEAR positions itself as settlement infrastructure for AI agent commerce. The protocol's NEAR Intents cross-chain layer has processed over $13 billion in cumulative volume, with a fee switch activated in February 2026 directing transaction revenue into automatic NEAR token buybacks at an annualized run rate of approximately $60.6 million as of March 2026. NEAR trades at $2.21 as of June 9, 2026, with a market capitalization of $2.87 billion — up roughly 160% from its February 2026 low of $0.85.
The question facing the protocol is whether infrastructure-level scaling translates into sustained economic activity. TVL sits at approximately $160 million, a fraction of the $500 million peak. Dynamic resharding solves a capacity problem. Whether NEAR has a demand problem is a separate issue entirely.
Until Upgrade 2.13, expanding the number of shards on NEAR required a full protocol governance vote, weeks of cross-team coordination among hundreds of independent validator nodes, and a staged manual rollout. According to the NEAR Foundation, each prior shard expansion was a multi-week engineering project.
Dynamic resharding automates this process. The mechanism works as follows:
The protocol's long-term target architecture envisions scaling beyond 70 shards. At that configuration, according to NEAR Foundation projections, aggregate throughput would exceed Visa's reported peak capacity of approximately 65,000 transactions per second.
The practical impact is that congestion-driven gas fee spikes — a recurring problem on monolithic blockchains during high-demand events — are structurally mitigated. When a shard gets congested, it splits. Traffic distributes. Fees remain low.
Upgrade 2.13 also introduces post-quantum-safe cryptographic signing based on FIPS-204 (ML-DSA), a lattice-based signature standard approved by the U.S. National Institute of Standards and Technology (NIST).
Current blockchain signature schemes — Ed25519 (used by NEAR, Solana, and others) and secp256k1 (used by Bitcoin and Ethereum) — are theoretically vulnerable to quantum computers running Shor's algorithm. While no quantum computer currently exists with sufficient qubits to break these schemes, the cryptographic community has moved toward "harvest now, decrypt later" threat models, where adversaries record signed transactions today for future decryption.
NEAR's approach allows users to rotate their keys to a quantum-resistant standard in a single on-chain transaction. A testnet for the FIPS-204 scheme is expected by end of June 2026. If deployed successfully, NEAR would be among the first major Layer-1 protocols to offer production-ready post-quantum key rotation.
No other top-30 protocol by market capitalization has announced a comparable post-quantum signing implementation at the protocol level, though research is ongoing at the Ethereum Foundation and within the Cosmos ecosystem.
NEAR's sharding design, branded Nightshade, has evolved through three major versions:
Nightshade 1.0 (launch): Modeled the network as a single blockchain where each block conceptually contains all transactions across all shards. Validators were assigned to specific shards but the chain presented a unified view. Shard count was fixed and required manual upgrades to change.
Nightshade 2.0 (2024-2025): Introduced stateless validation, meaning validators no longer needed to maintain local copies of shard state. State witness proofs replaced full state downloads. This reduced hardware requirements for validators and enabled the 1 million TPS benchmark in December 2025.
Nightshade 3.0 (announced at NEARCON 2026): Introduces separation of consensus and execution, atomic cross-shard transactions, and a live private shard. Dynamic resharding, the headlining feature of Upgrade 2.13, is the first mainnet component of the Nightshade 3.0 roadmap.
The architectural progression shows a consistent design philosophy: keep the protocol as a single logical chain (no separate beacon chains or relay chains) while parallelizing execution across an expanding set of shards.
In December 2025, the NEAR Foundation published results of a benchmark test achieving sustained throughput of over 1 million transactions per second across 70 shards. Three Grafana dashboards recorded peak readings of 1,029,497, 1,037,334, and 1,037,495 TPS, followed by sustained performance at approximately 1 million TPS for nearly one hour each.
The benchmark used Google Compute Engine C4D machines and what NEAR described as "realistic workloads" with traffic distributions resembling heavy-load patterns observed on large networks. The Foundation characterized it as "publicly verifiable" with methodology and dashboards available for independent review.
What the benchmark demonstrates: A fully sharded Layer-1 architecture can achieve seven-figure TPS throughput without sacrificing the single-chain abstraction. The result is, by the numbers, the highest throughput demonstration among sharded Layer-1 protocols.
What the benchmark does not demonstrate: That real-world usage will approach these figures. NEAR's current actual throughput, as reflected by its $160 million TVL and approximately $100,000 in daily Intents fees, is orders of magnitude below the benchmark capacity. The gap between theoretical throughput and actual demand remains the central challenge. Visa's 65,000 TPS peak capacity is rarely approached in practice either; daily average throughput runs closer to 1,700 TPS.
NEAR Intents is the protocol's cross-chain routing layer, enabling users to express transaction "intents" (desired outcomes) that solvers fulfill across multiple chains. The system has processed over $13 billion in cumulative swap volume and currently handles roughly $3 billion in monthly activity, according to NEAR Foundation data.
Fee switch economics: In February 2026, NEAR activated a fee switch on the Intents layer. All Intents fees are now routed into automatic NEAR token purchases on the open market. Key metrics as of recent reporting:
| Metric | Value | |---|---| | Annualized fee revenue (March 2026) | ~$60.6 million | | Monthly buybacks (recent) | ~$3 million | | Cumulative buybacks/burns | 2.1 million NEAR tokens | | Intents TVL | >$80 million | | Daily fees | ~$100,000 | | Monthly volume (May 2026) | ~$2 billion |
Deflation threshold: Following the Halving Upgrade in Q4 2025, NEAR's maximum annual inflation rate dropped from 5% to 2.5%. At current prices and the channel-mix-weighted fee rate, the deflationary threshold — the point at which buyback pressure exceeds new token emissions — sits at approximately $177 million in daily Intents volume. Current daily volume runs well below that figure, meaning the token remains net inflationary. Whether Intents volume scales to the deflation threshold is among the protocol's most material open questions.
Launched in February 2026, Confidential Intents adds a privacy execution layer to NEAR's cross-chain infrastructure. The system operates through a dedicated private shard running parallel to mainnet, connected via a Trusted Execution Environment (TEE) bridge.
Unlike Zero-Knowledge proof-based privacy systems (e.g., Zcash, Aztec), which require computationally expensive client-side proof generation, Confidential Intents handles privacy at the execution environment level. Users toggle on "confidential mode" for transfers, deposits, and withdrawals, which hides sender details and transaction routes from public view.
The design targets two use cases: institutional transactions requiring frontrunning protection (MEV mitigation), and AI agent transactions where exposing strategic intent to public mempools creates competitive disadvantages. NEAR explicitly states the system is "not a mixing service" and supports selective disclosure for compliance-aware institutional use.
The approach sidesteps the regulatory risk that has plagued privacy-focused protocols — notably, the Zcash vulnerability discovered by AI in May 2026 that triggered a 50% crash in ZEC — by building compliance hooks directly into the privacy layer.
NEAR's strategic positioning centers on becoming the settlement layer for an "agentic economy" — a future state where AI agents autonomously transact on-chain at machine speed. The thesis rests on several infrastructure components:
The thesis is coherent at the architectural level. The economic question is whether autonomous AI agent activity generates meaningful on-chain fee revenue. As of June 2026, AI agent transaction volume on any blockchain remains negligible relative to human-driven DeFi, NFT, and payment activity. NEAR is building infrastructure for a market that does not yet exist at scale.
Arthur Hayes, co-founder of BitMEX and CIO of Maelstrom, referred to NEAR as part of a "holy trinity" of AI-adjacent crypto assets in late May 2026, contributing to a 54% weekly price surge. The endorsement reflects market enthusiasm for the narrative but does not constitute evidence of product-market fit.
NEAR's approach to scaling differs fundamentally from Ethereum's current strategy:
| Dimension | NEAR (Nightshade) | Ethereum (Danksharding) | |---|---|---| | Sharding type | Execution sharding (transactions processed in parallel across shards) | Data sharding (shards provide data availability for L2 rollups) | | Execution model | Single-chain with parallel shard execution | L1 consensus + L2 execution (Optimism, Arbitrum, etc.) | | Current throughput | ~1M TPS (benchmark); actual usage far below | ~30 TPS L1; L2s collectively handle ~200+ TPS | | Scaling unit | Shard (auto-splits via dynamic resharding) | Rollup (independent L2 teams build/maintain) | | Privacy approach | Confidential Intents (TEE-based private shard) | Application-level (individual L2s implement) |
Ethereum's rollup-centric roadmap has achieved significantly higher actual economic activity: Ethereum L2s collectively hold over $30 billion in TVL versus NEAR's $160 million. However, Ethereum's approach introduces fragmentation — liquidity, composability, and user experience fracture across dozens of L2s with different bridges, standards, and security assumptions.
NEAR's single-chain abstraction avoids this fragmentation but has not yet attracted comparable capital or user activity. The Glamsterdam upgrade, Ethereum's next major release targeting 10,000 TPS at the L1 level, has been delayed to Q3 2026, potentially narrowing the throughput gap if NEAR's dynamic resharding deploys first.
NEAR Protocol's Upgrade 2.13 represents a genuine engineering milestone. Dynamic resharding solves the operational bottleneck of manual shard expansion, and post-quantum signing addresses a threat vector that most protocols have not yet begun to tackle at the implementation level. The Nightshade architecture, after three major iterations, delivers a technically differentiated scaling approach.
The economic picture is less clear. At $160 million in TVL and roughly $100,000 in daily Intents fees, the protocol's infrastructure capacity vastly exceeds its current demand. The fee switch buyback mechanism is an elegant tokenomics design, but the deflationary threshold remains out of reach at current volume levels. The AI agent economy that NEAR is positioning to serve is, as of mid-2026, still largely theoretical.
For the protocol to justify its $2.87 billion market capitalization through fundamentals rather than narrative, Intents volume would need to grow by a significant multiple. Dynamic resharding ensures the infrastructure can handle that growth if it arrives. Whether it arrives is a question of market adoption, not protocol architecture.