NEAR is making a specific wager: that the future of crypto is autonomous AI agents transacting at machine speed, and that they will need a blockchain built to handle them. A June upgrade is the centerpiece. Here is the thesis, the technology, and the one number that complicates it.
Summary
- NEAR is betting that AI agents will need a blockchain built for machine-speed transactions.
- Dynamic resharding is the June upgrade designed to scale capacity automatically.
- NEAR Intents gives agents a way to settle activity across multiple chains.
- The thesis is coherent, but falling active users show the agent economy has not arrived yet.
NEAR Protocol has spent 2026 rebuilding its pitch to the market around a single, specific bet: that the future of crypto belongs to autonomous AI agents, software that transacts on its own at machine speed, and that those agents will need a blockchain engineered to handle them.
The wager is sharp and unusual in a field full of vague AI branding, because NEAR is pointing at a concrete use case, an on-chain economy where AI agents buy compute, settle payments, label data, and execute trades automatically. Those agents could generate bursts of transactions that would paralyze a conventional blockchain, and NEAR is positioning itself as the infrastructure built to absorb that load.
A major network upgrade in June 2026, introducing automatic scaling, is the centerpiece of the bet, and NEAR’s leadership has branded the token “the currency of agents” and the network “a unified commerce layer.” The thesis is deeply interesting, the technology is real, and there is one number that complicates the whole story.
This piece works through NEAR’s bet in full: the AI-agent thesis and why a blockchain for agents would need to be different, the June upgrade called dynamic resharding and what it actually does, the other pieces NEAR has assembled around the thesis including its cross-chain settlement system and its privacy tooling, the tokenomics that tie usage to the token’s value, and the honest complication, a gap between NEAR’s soaring narrative and its actual on-chain usage that every serious observer should weigh.
The goal is to explain what NEAR is trying to become and to assess the bet clearly, neither dismissing a real and ambitious technical effort nor accepting the narrative uncritically. NEAR is one of the more concrete expressions of the AI-crypto thesis, and understanding it illuminates where that whole idea stands.
The bet: a blockchain built for AI agents
To understand NEAR’s strategy, you have to understand the specific future it is betting on, because the entire technical effort follows from a particular vision of how crypto will be used.
The vision is an on-chain economy populated by autonomous AI agents, software programs that act on their own to accomplish goals, transacting with each other and with services at machine speed and scale. In this future, an AI agent might need to buy computing power on one blockchain, settle a payment on another, and store data on a third, all automatically.
A swarm of such agents reacting to an opportunity, a profitable arbitrage, or a large data-labeling task, could suddenly generate hundreds of thousands of transactions in a short span. This is a fundamentally different usage pattern from human-driven crypto, where transactions arrive at human pace and human scale.
Agents operate at machine frequency, in unpredictable bursts, and at volumes that would overwhelm a blockchain designed for human users, which is the problem NEAR has decided to solve. That makes it another AI-crypto crossover, but one focused on transaction infrastructure rather than identity.
NEAR’s co-founder, who notably co-authored the 2017 research paper that introduced the transformer architecture underlying today’s large language models, has framed the protocol as fundamental infrastructure for exactly this AI-driven commerce.
Why would AI agents need a different blockchain instead of using existing ones? The answer is about handling unpredictable, machine-speed demand without breaking.
On a conventional blockchain, a sudden explosion of transactions causes congestion: fees spike, confirmations slow, and the network becomes expensive and sluggish for everyone. That is fatal for AI agents that need to transact cheaply, instantly, and at scale without warning.
A blockchain serving AI agents must be able to absorb sudden, massive surges of activity while keeping fees low and confirmations fast, scaling up its capacity automatically the moment demand spikes. There is no time to wait for human intervention when a swarm of agents starts transacting.
This requirement, automatic, instant scalability to handle unpredictable machine-speed bursts, is the technical heart of NEAR’s bet, and it is what the June upgrade is designed to deliver. NEAR is wagering that whoever builds the blockchain that can handle AI agents at scale will become essential infrastructure for the agent economy, and it is trying to be that blockchain.
The June upgrade: dynamic resharding
The centerpiece of NEAR’s bet is a June 2026 upgrade called dynamic resharding, and understanding what it does, in plain terms, explains why NEAR thinks it can serve AI agents when other blockchains cannot.
The concept rests on sharding, a technique NEAR has used since its launch to scale its blockchain. Sharding splits a blockchain into multiple parallel partitions called shards, each processing transactions independently, like opening multiple checkout lines in a grocery store instead of forcing everyone through a single queue.
More shards mean more transactions processed in parallel, and therefore more capacity. For a basic primer on the ledger model, sharding and scaling explained starts with the blockchain structure that sharding modifies.
NEAR has scaled this way for years, but until now, adding a new shard was a slow, manual process requiring weeks of validator coordination, a governance vote, and a staged rollout. That is the equivalent of needing a committee meeting every time the store wanted to open another checkout line.
That manual bottleneck is exactly the problem for AI agents, because when a surge of agent activity hits, there is no time to convene a vote and coordinate validators over weeks. The capacity has to appear immediately or the network congests.
Dynamic resharding removes the human bottleneck entirely. With the upgrade, when a shard fills up past a defined threshold, it automatically splits into more shards, deterministically and without any human intervention, adding capacity in real time exactly when and where it is needed.
In the grocery-store analogy, the store now automatically opens new checkout lines the moment the existing ones get crowded, with no manager required. NEAR’s leadership says the upgrade will let the network scale to many dozens of shards, with throughput exceeding that of major payment networks.
They frame it as foundational to the AI-agent vision: when a swarm of agents suddenly floods the network, dynamic resharding isolates that surge into newly created shards, absorbing it while keeping fees flat and confirmations fast for everyone else.
The same upgrade also adds post-quantum-secure signatures, cryptographic protections designed to resist future quantum computers, letting users rotate to quantum-safe keys. That is a forward-looking security measure that signals NEAR’s ambition to be durable infrastructure.
The upgrade, part of NEAR’s network release numbered 2.13, is the technical delivery of the AI-agent bet: automatic, instant scaling built precisely for the unpredictable machine-speed demand that agents would generate.
The pieces around the bet
Dynamic resharding is the centerpiece, but NEAR has assembled several other pieces around the AI-agent thesis, and seeing them together shows that the bet is a coordinated strategy, not a single feature.
The most important supporting piece is NEAR’s cross-chain settlement system, called Intents, which addresses a problem specific to AI agents operating across multiple blockchains. Rather than requiring an agent to hold tokens on every chain and navigate the complexity of moving between them, the Intents system lets an agent simply express what it wants to accomplish, and specialized participants called solvers figure out the optimal path across chains to make it happen.
For an AI agent that needs to buy compute on one chain, settle on another, and store data on a third, this abstraction is exactly what makes operating across a fragmented multi-chain world practical. The Intents system has processed a large volume of cross-chain activity, generating tens of millions of dollars in fees and settling transactions across many dozens of blockchains.
It is central to NEAR’s pitch as a “unified commerce layer” for agents, the connective tissue that lets agents transact across the whole crypto ecosystem through one interface.
NEAR has also leaned heavily into privacy, the second supporting pillar, on the reasoning that AI-driven commerce and confidential finance require privacy guarantees. The protocol’s infrastructure powers products offering confidential on-chain treasuries, private multisig, payroll, and balance management for organizations that need to manage funds without exposing everything publicly.
Separately, NEAR’s AI division rolled out automatic anonymization of personal information in prompts sent to closed AI models, scrubbing sensitive data before it reaches the inference infrastructure. That addresses enterprise concerns about data leakage when using AI.
Together with dynamic resharding, these pieces, cross-chain settlement through Intents and a suite of privacy tools, form a coordinated thesis: NEAR is trying to be the scalable, cross-chain, privacy-capable settlement layer that AI agents and confidential finance need. It is assembling the specific capabilities that an agent economy would require instead of just adding a generic AI label.
That strategy is coherent, which is part of what makes the bet credible enough to take seriously.
The tokenomics: tying usage to value
For investors, the question is how NEAR’s technical ambitions connect to the value of the NEAR token, and the protocol has restructured its tokenomics to forge that link, which is worth understanding.
NEAR made two important tokenomic changes that tie network usage to token value. First, it cut its inflation rate, reducing the maximum annual issuance of new tokens significantly, which matters because the token supply is now fully unlocked, so lower issuance means less dilution of existing holders.
Second, and more directly tied to the AI-agent thesis, NEAR activated a fee mechanism on its Intents settlement system, under which the fees generated by cross-chain settlement activity are used to buy NEAR tokens on the open market. This creates a direct feedback loop: more usage of the Intents system generates more fees, and those fees translate into more buying pressure on the token.
That means if AI-agent and cross-chain activity grows, the growth flows through to demand for NEAR. The design is meant to ensure that the token captures value from the network’s actual usage instead of relying purely on speculation, aligning the token’s value with the success of the AI-agent thesis.
The proof-of-stake base matters too, because staking is how networks like NEAR secure themselves while issuing rewards and aligning validators. That is NEAR’s proof-of-stake foundation, and it sits underneath the scaling and usage story.
This tokenomic structure is what makes the AI-agent bet an investment thesis and not just a technical one. If NEAR succeeds in becoming the settlement layer for AI agents, the resulting surge in transaction activity would generate fees that buy NEAR, and the reduced inflation would mean that demand is not offset by heavy new issuance.
The logic is clean: usage drives fees, fees drive token buying, and lower inflation preserves the effect, so the token is engineered to benefit if the agent economy materializes on NEAR. The caveat, which the next section develops, is that this entire mechanism depends on real, growing usage.
The fee-to-buyback loop only generates meaningful demand if the Intents system and the broader network are actually being used at scale. A clever tokenomic design that ties value to usage is only as valuable as the usage it captures, and that is precisely where NEAR’s story meets its complication.
The structure rewards success, but it cannot manufacture it.
The number that complicates the story
Here is the honest complication that any serious assessment of NEAR must confront, because it is the gap between the narrative and the reality, and it is the single most important thing for a skeptical observer to weigh.
NEAR’s token has rallied substantially on the AI-agent thesis, surging on the announcement of dynamic resharding and the broader AI narrative. The story is compelling, the technology real, the strategy coherent.
But the on-chain usage tells a more sobering story. The number of daily active users on the NEAR network fell dramatically over 2026, dropping from nearly three million earlier in the year to a small fraction of that, a steep decline that analysts have flagged as a warning sign precisely because it diverges so sharply from the soaring price and narrative.
This is the gap that complicates everything: NEAR’s price and story point to a thriving AI-agent future, while its actual usage, measured by active users, has been falling, not rising. The narrative describes a network about to be flooded with AI-agent activity, while the data shows fewer humans actually using it.
That disconnect between price action and on-chain usage is exactly the kind of signal that should make an observer cautious.
This does not mean the bet is doomed, but it means the bet is unproven and largely ahead of its evidence. Some of NEAR’s rally has been driven by factors other than fundamental adoption, including short squeezes that force bearish traders to buy back positions and amplify upward moves, and by the powerful pull of the AI narrative itself, which can lift a token’s price faster than real usage justifies.
The crucial open question is whether the AI-agent thesis will translate into actual, sustained on-chain activity: whether the fees, the usage, the agent transactions, and the revenue capture will genuinely grow enough to justify the renewed market attention and the token’s price.
The technology may work as advertised and the strategy may be sound, but the agent economy NEAR is betting on has not yet arrived at scale on its network. The falling user count is a reminder that the thesis remains a wager on the future, not a description of the present.
An honest assessment holds both truths: NEAR has built coherent, interesting infrastructure for a plausible future, and that future has not shown up in the usage data yet, leaving the bet credible but unproven.
How to weigh the bet
For anyone trying to assess NEAR, the situation comes down to weighing a real and coherent technical bet against an unproven thesis and a worrying usage trend, and a few principles clarify the judgment.
The case for taking NEAR seriously is real. The AI-agent thesis is plausible, a future of autonomous agents transacting on-chain is a credible direction for crypto, and NEAR has built a coherent, technically ambitious set of tools for it: automatic scaling through dynamic resharding, cross-chain settlement through Intents, privacy infrastructure, and tokenomics that tie usage to token value.
This is not vague AI branding bolted onto an unrelated chain; it is a focused, multi-year effort to build specifically for the agent economy, led by a team with deep AI credentials. If the AI-agent future arrives and NEAR captures even a meaningful share of it, the network’s design positions it to benefit substantially, and the tokenomics would channel that benefit to the token.
For an investor who believes in the AI-agent thesis and in NEAR’s execution, the bet has a clear logic. Agents would use the code agents would transact through, and NEAR is trying to make that code scale across chains and bursts of activity.
The case for caution is equally real and rests on the gap between narrative and reality. The thesis is unproven, the agent economy has not arrived at scale, the on-chain usage has been falling, not rising, and part of the price strength has come from market mechanics like short squeezes and the momentum of the AI narrative instead of from fundamental adoption.
An investor should weigh that the bet is precisely that, a bet on a future that may or may not materialize on NEAR specifically, in a competitive field where other blockchains are also pursuing scalability and AI use cases. Automatic scaling, if it proves valuable, could be matched by competitors.
The disciplined reading is to treat NEAR as a high-conviction bet on a specific and unproven future, sized to the reality that the thesis is ahead of the evidence. Watch the actual usage data, the fees, the active users, and the agent activity, because those are the real tests of whether the narrative is becoming reality.
That discipline matters especially against the market backdrop for altcoins, where strong narratives can still run into a difficult macro and liquidity environment. The technology and strategy are real; the adoption is the open question, and watching it, not the price, is how to know whether the bet is paying off.
None of this is investment advice; it is a frame for assessing one of crypto’s more concrete and ambitious AI bets with appropriate clarity about what is proven and what is hoped.
A coherent bet, ahead of its evidence
NEAR’s wager is one of the clearest expressions of the AI-crypto thesis in the market: a bet that autonomous AI agents will transact on-chain at machine speed and scale, and that they will need a blockchain built to absorb that load.
The June dynamic resharding upgrade is the centerpiece, delivering automatic, instant scaling designed precisely for the unpredictable bursts an agent economy would generate. Around it, NEAR has assembled a coherent strategy: cross-chain settlement through Intents, privacy tooling, and tokenomics that channel usage-driven fees into buying the token.
Led by a team with deep AI credentials and pointed at a plausible future, the bet is specific, technically real, and worth taking seriously, not the vague AI branding that decorates so many crypto projects.
The complication is the gap between the narrative and the evidence. NEAR’s price and story describe a network on the verge of an AI-agent boom, while its actual usage, measured by a daily active user count that has fallen sharply over 2026, tells a more sobering tale.
Part of the rally has come from short squeezes and the pull of the AI narrative, not fundamental adoption. The agent economy NEAR is betting on has not yet arrived at scale on its network, which leaves the thesis credible but unproven, ahead of its evidence.
The honest assessment holds both: NEAR has built impressive, focused infrastructure for a believable future, and that future has not shown up in the usage data, making NEAR a high-conviction bet on a specific future, not a description of present reality.
Whether dynamic resharding and the Intents system become the rails of a real agent economy, or whether the narrative outruns the adoption, is the question that will define NEAR. The answer lies not in the price but in whether the agents ever actually arrive.
The bet is placed and the infrastructure is built; the economy it is built for has yet to show up.
Frequently asked questions
What is NEAR betting on with the AI-agent thesis?
NEAR is betting that the future of crypto involves autonomous AI agents, software that transacts on its own at machine speed, and that those agents will need a blockchain engineered to handle their unpredictable, high-volume activity. It envisions an on-chain economy where agents buy compute, settle payments, and store data automatically, generating bursts of transactions that would overwhelm conventional blockchains. NEAR is positioning itself as the scalable settlement layer for this agent economy, branding its token “the currency of agents.”
What is dynamic resharding?
Dynamic resharding is a June 2026 NEAR upgrade, part of network release 2.13, that lets the blockchain automatically add capacity when demand spikes. NEAR uses sharding, splitting the network into parallel partitions, or shards, like multiple checkout lines. Previously, adding a shard required weeks of manual validator coordination and a governance vote. Dynamic resharding removes that bottleneck: when a shard fills up, it automatically splits into more shards, with no human intervention, adding capacity in real time, which is essential for absorbing sudden AI-agent surges.
Why would AI agents need a special blockchain?
Because they transact at machine speed in unpredictable bursts. A swarm of agents reacting to an opportunity could generate hundreds of thousands of transactions suddenly, which on a conventional blockchain causes congestion, spiking fees and slowing confirmations for everyone. A blockchain serving agents must absorb these surges automatically while keeping fees low and confirmations fast, scaling capacity the instant demand spikes, because there is no time for human intervention. That automatic, instant scalability is what NEAR’s dynamic resharding is built to provide.
How does NEAR’s token capture value from this?
Through two tokenomic changes. NEAR cut its inflation rate, reducing dilution since the supply is fully unlocked. More importantly, it activated a fee mechanism on its Intents cross-chain settlement system, where fees from settlement activity are used to buy NEAR on the open market. This creates a feedback loop: more usage generates more fees, which buy more NEAR, so growth in AI-agent and cross-chain activity flows through to token demand. The design ties the token’s value to actual network usage rather than pure speculation.
What is the problem with NEAR’s story?
A gap between narrative and reality. NEAR’s price and story describe a thriving AI-agent future, but its on-chain usage tells a different tale: daily active users fell sharply over 2026, from nearly three million to a small fraction of that. This decline diverges from the soaring price, and analysts flag it as a warning sign. Part of the rally also came from short squeezes and AI-narrative momentum rather than fundamental adoption. The thesis is unproven, and the agent economy has not yet arrived at scale on NEAR.
Is NEAR a good investment?
That depends on whether you believe the AI-agent thesis and NEAR’s execution, and it is genuinely unproven. The case for it: a plausible future, coherent and ambitious technology, a credentialed team, and tokenomics tying value to usage. The case for caution: the thesis is unproven, usage has been falling, the agent economy has not materialized at scale, competitors are pursuing similar goals, and price strength has partly come from market mechanics. The disciplined approach watches actual usage data, not price, as the test. This is not investment advice.
As of June 21, 2026. Crypto markets and protocol details change quickly; verify current data before relying on this analysis. This article is information, not investment advice.
No Comment! Be the first one.