NEAR Protocol

What it does

NEAR Protocol is a sharded proof-of-stake Layer 1 blockchain co-founded by Illia Polosukhin, one of eight co-authors of “Attention Is All You Need” (2017), the Google Brain paper that introduced the transformer architecture underpinning GPT, BERT, and every modern large language modelLLMLarge Language Model. A neural network trained on vast amounts of text to predict the next word in a sequence. Modern LLMs (GPT, Claude, Llama, Qwen, DeepSeek) generate human-quality text and are the foundation of most modern AI products.Like an autocomplete that read every book ever written. It has no memory of individual texts but it has absorbed the patterns of language so deeply that it can generate paragraphs that sound human. The skill is statistical, not conscious.LLM in the glossary →. That paper has been cited over 140,000 times. It’s not an exaggeration to say it changed the course of technology. Polosukhin is now using that credibility to position NEAR as the AI-blockchain convergence platform.

The base layerL1Layer 1. A base blockchain that runs its own consensus mechanism, executes transactions, and settles its own state. Bitcoin, Ethereum, NEAR, and Solana are all L1s. Anything built on top of an L1 is technically a Layer 2 or higher.Like the foundation of a building. Nothing else can exist on top until the foundation is solid. Different L1s make different tradeoffs for what kind of building they can support.L1 in the glossary → is a Nightshade-sharded PoSProof of StakeA consensus mechanism where validators earn the right to create new blocks by staking tokens as collateral. If they misbehave, the network slashes their stake. Proof of Stake replaced energy-intensive mining on most modern chains.Like being a licensed auctioneer. You post a bond to get the license, you earn fees for every auction you run, and you lose the bond if you rig an auction. The bigger the bond, the more auctions you get to run.Proof of Stake in the glossary → chain. Validators process transactions in parallel across multiple shards while producing a single block containing all shard chunks. Nightshade 2.0, shipped in May 2025, brought block times down to 600ms with 1.2-second finality and average throughput of roughly 80 TPS (testnet peaks at 4,135 TPS). The chain has run continuously since its mainnet launch on 13 October 2020 with no consensus failures. 397 validators secure the network, with the top 100 producing blocks and chunks. Stake is comparatively well spread for an L1: the largest holds 5.5% and it takes 10 validators to reach a third of stake.

The AI pivot has three pillars. First, NEAR Intents: a chain abstraction layer that lets users swap assets across chains without managing multiple wallets or bridges. According to DefiLlama, it has processed $34.4 billion in cumulative swap volume, with $53 million in gross fees over the trailing year. Second, NEAR AI: a research lab building IronClaw (hardware-secured AI agent runtime using Intel TDX and NVIDIA Confidential ComputingConfidential ComputeHardware-enforced computation where data and code are encrypted in memory and only the authorised application can access them. The machine's operator cannot read what the application is doing even though they own the machine.Like renting space in a bank vault. The bank owns the building and runs the security, but what you put in the vault is invisible even to the bank staff. Only you have the key.Confidential Compute in the glossary → TEEs), Private Chat (encrypted AI conversations), and a Confidential GPUGPUGraphics Processing Unit. Originally designed to render video game graphics, GPUs turned out to be exceptionally good at the massively parallel math that AI models need. Modern AI training and inference runs almost entirely on GPUs.Like a factory with 10,000 workers doing the same simple task in parallel, versus a CPU which is more like 10 workers each doing different complex tasks. AI training involves doing simple math a million times per second on a million numbers, which is exactly what the GPU factory is designed for.GPU in the glossary → Marketplace. Third, near.com: a consumer super-app launched at NEARCON in February 2026 that unifies multi-chain asset management, AI agents, and confidential transactions.

The perps product shows how far near.com’s confidentiality reaches. Its homepage headline is “Trade perps, confidentially”, but the FAQ describes a dedicated account on the Hyperliquid platform and says only that deposits to and withdrawals from that account are confidential. The privacy covers the funding legs, and the position itself sits on Hyperliquid, an exchange outside NEAR.

Chain Signatures allow NEAR accounts to sign transactions on any blockchain (Bitcoin, Ethereum, and others) using MPCMPCMulti-Party Computation. A cryptographic technique where multiple parties jointly compute a function over their inputs without any party revealing its input to the others. Useful for shared computation on private data.Like a group of friends calculating the average of their salaries without anyone revealing their actual salary to the others. Each person contributes a piece of the answer, and the pieces combine into the result, but nobody learns anything except the average itself.MPC in the glossary → threshold signatures. Aurora provides EVM compatibility. Smart contracts run in WebAssembly (Wasm), primarily written in Rust.

The co-founders are Illia Polosukhin (CEO) and Alexander Skidanov (former CTO). Polosukhin’s background is Google Brain (2014-2017), working on TensorFlow and machine translation. Skidanov is a former MemSQL engineer and ICPC gold medalist. The entities are NEAR Foundation (Zug, Switzerland), Pagoda (formerly NEAR Inc, San Francisco, core protocol development), and NEAR AI (research lab). Team size is approximately 100.

Value proposition

The pitch rests on a unique convergence: the only L1 blockchain whose founder has verifiable, peer-reviewed AI research credentials at that level, combined with proven chain infrastructure and a chain abstraction stack that actually works at scale.

NEAR Intents is the strongest proof point. DefiLlama data shows $34.4 billion in cumulative volume and $53 million in gross fees over the trailing year ($7.6 million in the last 30 days). That’s production infrastructure handling cross-chain value at scale. The near.com super-app is built on top of it, aiming to be the consumer interface that makes chain abstraction invisible.

One incident worth flagging: on 25 April 2026, NEAR Intents took roughly $600,000 of exposure when a Litecoin MWEB zero-day reorg exploit (Litecoin’s first major privacy-layer incident) reversed a swap of 11,000 LTC into BTC. The bug was on the Litecoin side, not NEAR; the Intents protocol functioned as designed. The team pledged user reimbursement and the LTC chain reorg has since been confirmed by The Block and Crypto Times. It’s a reminder that the chain-abstraction layer inherits the security properties of every chain it touches, NEAR’s own chain included.

The second incident was NEAR’s own code. On 1 October 2026 NEAR Intents said a bug in how its Omni deposit and withdrawal infrastructure interacted with the Intents contract cost about $3.8M. It stopped the service, patched the contract and paused deposits and withdrawals on 11 networks while Omni was fixed.

DeFiLlama’s hacks database records it as a $3.8M bridge logic flaw on BSC. On 2 October Illia Polosukhin said the funds had been returned in full; the team promised a detailed report, which we had not found on 4 October 2026.

The developer community is healthy. 2,500 monthly active developers (Electric Capital 2025), 399 public repositories in the near GitHub organisation and 76 in the nearai organisation (GitHub API, 24 September 2026), and an $800 million grants programme announced in October 2021 (distributed over time through Proximity Labs, MetaWeb Ventures, and Foundation direct grants). The Hacken audit of nearcore scored a perfect 10/10 across security, code quality, documentation, and test coverage.

The counter-narrative is equally clear. NEAR was a general-purpose L1 for four years before pivoting to AI in 2024-2025. The timing coincides with the AI hype cycle, not a longstanding technical direction. NEAR AI Cloud and Private Chat launched only in December 2025; the near.com super-app launched in February 2026. These products are weeks old and haven’t been battle-tested. HuggingFace presence is zero: the near-ai organisation exists with three team members but doesn’t have any public models or datasets.

TVLTVLTotal Value Locked. The sum of all assets currently deposited in a protocol's smart contracts. TVL is the standard measure of how much capital a DeFi or DeAI protocol is custodying.Like the assets under management of a hedge fund. AUM tells you how much money the fund has been trusted with, which is a rough proxy for how much business it's doing. TVL plays the same role for crypto protocols.TVL in the glossary → is surprisingly thin relative to market cap on any reading, and through mid-August 2026 the DeFiLlama series swung about 40% between daily reads (see the score log). The TVL-to-mcap ratio sits well below comparable L1s like Ethereum, Solana, or Arbitrum. The 46 million monthly active users figure demands scrutiny; L1 user metrics are notoriously inflatable through bot activity, airdropAirdropDistributing tokens for free to eligible wallets, usually to reward early users, bootstrap a community, or decentralise token ownership away from a small group of insiders at launch.Like a supermarket handing out free samples to people who already shop there. The samples cost the supermarket nothing to print. The goal is to convert casual shoppers into loyal customers by giving them something tangible to talk about.Airdrop in the glossary → farming, and low-value transactions. Without transaction value analysis, the headline number is impressive but unverifiable.

The fundraising history is both a strength and a cautionary tale. $607 million raised across six rounds, one of the largest in crypto. But the investor list includes Three Arrows Capital (co-led the $150 million Series B in January 2022, collapsed in June 2022), Alameda Research, and FTX Ventures (both collapsed November 2022). The forced liquidation of NEAR holdings from bankruptcy estates has likely contributed to sustained sell pressure.

Tokenomics

NEAR tokenomics in brief: the NEAR token pays gas and storage, secures validator seats through staking, and since July 2026 can be staked to the NEAR AI pool for inference credits instead of yield. Two value mechanisms are live on chain: 70% of every gas fee is burned (the whole fee once protocol 87, live only on testnet, reaches mainnet), and the protocol’s slice of NEAR Intents fees moves NEAR into a buyback multisig that holds it rather than burns it. Against that, 2.5% annual issuance, about 89,000 NEAR a day, goes to stakers, and the buyback account has collected only about a tenth of the NEAR issued since it opened in March 2026. Genesis vesting is complete, so issuance is the only dilution (how NEAR AI staking compares).

NEAR is the native L1 tokenTokenA digital unit of value or access rights tracked on a blockchain. Tokens can represent ownership in a project, a right to use a service, a share of future revenue, or simply a tradable asset with no underlying claim.Like a physical poker chip a casino issues. The chip itself has no value. What makes it worth something is what it lets you do at the casino, what the casino has promised, and how much other people will pay you for it.Token in the glossary →. No max supply; it’s inflationary by design with a partial burnBurnPermanently removing tokens from circulation by sending them to an address that no one controls. Burns reduce total supply, which (all else equal) makes each remaining token worth more of the network's value.Like a company buying back its own shares and shredding them. The company's total value stays the same, but each remaining share now represents a slightly bigger slice of that value.Burn in the glossary → mechanism. Total supply: approximately 1.3 billion. The circulating shareCirculating SupplyThe number of tokens currently in circulation and tradeable on the open market. Differs from total supply (which includes locked or unvested tokens) and max supply (the upper limit, if there is one).Like the number of cars on the road today versus the number ever produced. Some are in showrooms, some in junkyards, some still at the factory. Only the ones on the road count toward what people are actually driving.Circulating Supply in the glossary → is disputed: aggregators disagree by about 4% of supply, and no holder of that gap has been identified. All initial vestingVestingA schedule that locks up tokens allocated to insiders, investors, and team members, releasing them gradually over months or years. Vesting prevents insiders from dumping on public buyers immediately after launch.Like a new employee's stock options at a startup. You don't get all the shares on day one. They unlock over four years so you stick around and do the work rather than cashing out and leaving.Vesting in the glossary → is complete.

Distribution (of original 1B genesis allocation):

• Community Grants and Programmes: 17% (distributed over time via network grants)
• Seed/Venture Investors: 17.6% (176M NEAR across pre-seed at $0.0375, seed at $0.05, and venture rounds at $0.25-$0.32; 12-24 month linear vesting; fully vested)
• Core Contributors: 14.5% (145M NEAR to founding team and early contributors; fully vested)
• Community Sale (CoinList): 12% (120M NEAR sold in August 2020; three pricing tiers: $0.40 immediate, $0.34 with 12-month lock, $0.29 with 24-month lock; raised $40M; fully vested)
• Early Ecosystem: 11.7% (bounties, grants, and programmes)
• Operations (Pagoda/NEAR Inc): 11.4% (operational treasury)
• NEAR Foundation: 10% (network stewardship)
• Foundation Endowment: 5.8% (long-term reserve)

Those are genesis allocations, and they are the only Foundation figures this review has ever carried. What the Foundation holds today is not measurable from outside it. Checked August 2026: the two accounts anyone can name, foundation.near and nf-payments.near, hold 63,966 NEAR between them, 0.0049% of supply. foundation.near has zero staked across all forty pools it has ever delegated to and no lockup contract at all.

No public disclosure names a Foundation account set, and its own transparency reports give aggregate treasury without account IDs. More than half of the treasury it reports is fiat in bank accounts, which no chain read can reach. So treat the percentages above as what was allocated at genesis, never as what is held now, and treat any current Foundation treasury figure as the Foundation’s own claim.

InflationEmissionsNew tokens created and distributed by a blockchain protocol over time as rewards to validators, stakers, or miners. Emissions fund network security and participation at the cost of diluting existing holders.Like a company that pays employees partly in newly printed shares. Every year the total number of shares goes up, which means existing shareholders own a slightly smaller slice of the same company unless the company grows faster than the printing.Emissions in the glossary → was 5% annually from genesis. On 30 October 2025, the network completed a halvingHalvingA protocol event that cuts the rate of new token emissions by half. Halvings are scheduled in advance, happen automatically at fixed intervals, and are a core mechanism for enforcing declining token supply growth over time.Like a savings account where the interest rate is contractually cut in half every four years. You still earn interest, but the rate drops on a known schedule, and the issuer can't change it without breaking the contract.Halving in the glossary → upgrade reducing maximum inflation to 2.5%, cutting annual issuanceInflationThe annual rate at which new tokens are created and added to the circulating supply. Most networks use inflation to pay validators, stakers, and infrastructure providers from freshly minted tokens rather than real revenue.Like a landlord who raises the rent every year. If your salary goes up at the same rate, you break even. If it doesn't, you get poorer without noticing, because the number on your payslip hasn't changed but the ground under it has shifted.Inflation in the glossary → from approximately 64 million to 32 million NEAR (roughly 89,000 NEAR per day at today’s supply). The halving was implemented after 80% of block-producing validators adopted protocol version 81, though the preceding community governance vote technically failed to reach its required threshold, a controversial decision that drew criticism from Chorus One, who called it “a dangerous precedent.”

Validators and delegators receive 90% of that issuance and the protocol treasury, treasury.near, receives 10%. We measure the split from the chain’s payouts: the treasury’s balance jumps only at epoch boundaries, by 10.0% of the new supply minted there. The protocol’s protocol_reward_rate setting reads 0, but it doesn’t reflect what the chain pays.

70% of transaction fees are burned, creating deflationary pressure proportional to network usage. That share rises to the whole fee once mainnet reaches protocol 87, which removes the developer rebate covered below.

A governance-forum proposal posted on 30 September 2026 would cut maximum issuance from 2.5% to 1.6%, stepped down every epoch over 24 months, keeping the 90/10 split. It became a formal House of Stake draft on 7 October. It still needs a House of Stake vote, and validators would then have to adopt it.

On 23 February 2026, NEAR activated the Intents fee switch: fees generated through the cross-chain Intents layer now buy NEAR on the open market, with the bulk shared to integration partners and the remainder retained by the protocol. Per DeFiLlama, gross Intents fees came to $53 million over the trailing year, of which $8.5 million is the protocol-retained slice that creates durable buy pressure. NEAR said in February 2026 that Intents revenue had already funded over 1M NEAR of buybacks, which we can’t confirm: the buyback account we traced on NearBlocks, buybacks.multisignature.near, holds about 1.77M NEAR funded from intents.near, but it was created on 18 March 2026, after that statement. That NEAR sits in a multisig rather than being burned, so it doesn’t reduce supply.

StakingStakingLocking up a cryptocurrency to help secure a blockchain network, usually in exchange for rewards. The locked tokens act as a security deposit that can be taken away if the staker misbehaves.Like putting down a large rental deposit for an apartment. You get the money back if you behave, you earn interest while it's locked, and the landlord takes it if you trash the place.Staking in the glossary → yields about 4.9% a year nominal before validator commission, down from ~9% pre-halving. The arithmetic is simple: stakers’ 90% of the 2.5% issuance, 2.25% of total supply, is spread across whatever share is staked, and 599.6 million NEAR is staked across the active set. There is no fixed validatorValidatorA computer that runs the full blockchain protocol, verifies transactions, and proposes new blocks. Validators are the workers that keep a Proof of Stake network running, and they earn rewards for doing the work correctly.Like a notary public who witnesses and stamps legal documents. Validators witness transactions, check they follow the rules, and stamp them into the permanent record. A notary who commits fraud loses their license. Validators work the same way, except the license is staked tokens that get slashed on misbehaviour.Validator in the glossary → minimum: the threshold is a seat price set by the active set each epoch, currently 9,826 NEAR against a protocol floor of one 62,500th of total stake, so it moves as stake moves. Unstaking takes 2-3 epochs, and with Nightshade 2.0 running blocks at roughly 600ms an epoch is 7.1 hours rather than the 12 hours the older documentation assumes, so the wait is under a day.

On 31 July 2026, NEAR added a “Staking for NEAR AI” sink: the staking yield on staked NEAR converts into AI-inference credits on NEAR AI Cloud, including Anthropic, OpenAI and Google models, without the tokens being consumed. NEAR’s docs calculator sets the credit at stake × NEAR price × staking APY, one dollar of credit per dollar of yield, accruing per second with unused credit carried forward. Agent hosting keeps a fixed ratio of staked NEAR ÷ 100 a month in dollars (how the three designs compare). Either way, no card or cloud billing account is needed.

Funding totals $607 million across six rounds: $810K pre-seed (September 2017), $2.88 million seed (October 2019), $21.6 million venture (January 2020), $40 million CoinList public sale (August 2020), $150 million Series B Tranche 1 (a16z/3AC co-leads, January 2022), and $350 million Series B Tranche 2 (Tiger Global/Dragonfly co-leads, April 2022).

NEAR is listed on Binance, Coinbase, OKX, Kraken, KuCoin, and Revolut. Daily volume is substantial. Bitwise’s single-asset NEAR ETF, filed with the SEC on 6 May 2025, had its registration declared effective on 24 September 2026, and NYSE Arca certified the listing under the ticker NRR on 28 September (SEC EDGAR). Its prospectus sets a 0.75% annual fee and a secondary objective of earning more NEAR through staking.

Two other routes sit behind it. Grayscale filed on 20 January 2026 to convert its NEAR Trust into a spot ETF on NYSE Arca under the ticker GSNR, with Coinbase Custody as custodian and BNY Mellon as administrator; its last amendment was in June 2026, and the trust still trades over the counter. Bitwise also filed multi-asset Strategy ETFs including NEAR on 30 December 2025, allocating up to 60% directly to the token. The ATHATHAll-Time High. The highest price a token has ever reached. ATH is usually quoted as a reference point for how far the current price has fallen (or risen) since the peak.Like the record lap time on a racetrack. It tells you what the car has been capable of at its absolute best, not what it will do today. Whether that record gets broken again depends on conditions that may or may not come back.ATH in the glossary → of $20.44 was hit on 16 January 2022; the token was about 80% below that peak on 24 September 2026 (CoinGecko).

How to participate

Stake NEAR. The most accessible entry point. Delegate to any of the active validators through NEAR WalletWalletSoftware that stores the private keys needed to control tokens on a blockchain. A wallet does not actually hold any tokens. The tokens live on the chain. The wallet holds the keys that prove you own them.Like the key to a safe deposit box. The key doesn't contain your valuables. The valuables sit in the bank's vault. The key is what proves you're allowed to open the box and take them.Wallet in the glossary →, the near.com app, or third-party wallets. No minimum for delegation. About 4.9% a year nominal before commission. Unstaking takes 2-3 epochs, roughly 14 to 21 hours at the current block rate.

Run a validator. Permissionless, with entry set by the current seat price rather than a fixed minimum; read it from the active set before committing. Hardware requirements: 8-core CPU, 24GB RAM, 1TB SSD, 1Gbps network. Top 100 validators by stake produce blocks and chunks; additional validators validate chunks only.

Build on NEAR. Deploy smart contracts in Rust or JavaScript/TypeScript via NEAR SDKs. Access chain abstraction (Intents, Chain Signatures), AI infrastructure (NEAR AI, IronClaw), and Aurora EVM compatibility. 30% of transaction fees generated by smart contracts are rebated to the contract, which the live protocol config still confirms, though House of Stake approved a proposal to remove that rebate in June 2026 and it has not yet reached the chain. The removal ships in nearcore 2.14.0 at protocol 87, still a release candidate: on 4 October 2026 testnet ran 87 with no rebate, while mainnet ran 86 with the 30% intact. Treat it as a benefit on notice. Grant funding available through the Foundation.

Use the network. The near.com super-app provides multi-chain asset management, AI agents, and confidential transactions. NEAR Intents enables cross-chain swaps. NEAR AI Private Chat offers encrypted AI conversations. DeFiDeFiDecentralised Finance. Financial services like lending, trading, and yield farming built on smart contracts instead of traditional banks or brokerages. DeFi protocols are usually permissionless and global.Like a vending machine that can give you a loan, swap your currencies, or invest your savings. Nobody is behind the counter, the rules are written into the machine itself, and anyone with money in the right format can use it.DeFi in the glossary → on NEAR includes Ref Finance and Burrow.

Trade perps. Since September 2026 near.com has also offered perpetual futures: per the perps FAQ, trading runs “through a direct integration of Hyperliquid”, with 50+ markets at up to 40x, margin funded from any chain through NEAR Intents, and it’s closed to US persons.

Governance. Lock NEAR as veNEAR (vote-escrowed NEAR) in the House of Stake governance system for stake-weighted voting. Read the scope before assuming it is protocol governance: the approved proposals on the vote.dao contract are House of Stake’s own operating budget and constitution, funding programmes, and one protocol parameter. Proposals pass a five-account council before reaching a vote, at most three can be active at once, and protocol upgrades still ship by validator adoption of a nearcore release.

Honest assessment

What works

Illia Polosukhin’s credentials are in a class of their own. A co-author of the paper that created the transformer architecture is now building an AI-blockchain platform. No other L1 founder has equivalent AI research pedigree. That is a factual competitive advantage, not narrative-crafting, and it cannot be replicated.

The chain itself is mature. Five-plus years of continuous mainnet operation with no consensus failures. Two Rainbow BridgeBridgeA protocol that lets you move assets from one blockchain to another. Bridges typically lock the asset on the source chain and mint a wrapped version on the destination chain. Bridges are notoriously the most-attacked component in crypto.Like a coat check at a club. You hand over your coat, get a numbered ticket, and the club promises to return the coat when you bring back the ticket. The trust assumption is that the coat check doesn't lose your coat or run away with it.Bridge in the glossary → attack attempts were both automatically detected and thwarted by watchdog systems within seconds. The Hacken audit scored nearcore 10/10, with zero critical, high, or medium findings. Nightshade 2.0 delivers measurable performance: 600ms blocks and 1.2-second finality put NEAR competitive with modern L1s while maintaining an active set of 397 validators.

NEAR Intents is the standout product. DefiLlama data shows $34.4 billion in cumulative volume and $53 million in gross fees over the trailing year ($7.6 million in the last 30 days). Volume reached $4.8 billion in the last full month, with stablecoin swaps dominating activity across roughly 35 chains, which points to substantive cross-chain economic activity rather than speculative churn. The chain abstraction thesis, that users shouldn’t need to know or care which chain their assets are on, is sound, and Intents is delivering it at scale. The fee switch activation in February 2026 directs Intents fees into open-market NEAR buybacks, a concrete value accrual mechanism most L1s lack, though only the protocol-retained slice, $8.5 million over the trailing year, is durable buy pressure.

The inflation halving in October 2025 cut issuance from 5% to 2.5%, a meaningful economic change. However, the process raises governance concerns. The community governance vote technically failed to reach the required threshold, but the upgrade was implemented anyway once 80% of validators adopted the new protocol version. Chorus One, a major staking provider, criticised this as “a dangerous precedent.” The outcome was positive for tokenomics; the process was questionable for governance credibility.

Open source commitment is strong. 475 public repositories across the near and nearai GitHub organisations. nearcore has 2,620 stars and 795 forks. IronClaw, the TEE-based AI agent runtime, has 12,631 stars on the nearai org (GitHub API, 24 September 2026). The NEP process is public and active. Multiple audits from Hacken, Sigma Prime, Blaize, and BlockSec cover core protocol and infrastructure.

The AI layer landed its first named customer in May 2026. The Government of Bermuda signed NEAR AI for a prototype assistant that lets public servants process sensitive citizen data inside hardware-enforced enclaves, where the host OS, GPU operator, and NEAR AI itself are cryptographically locked out. Bermuda’s own Royal Gazette and ETHNews both reported it, so this is a verified government reference customer rather than a NEAR press release.

Brave is a second named consumer of the stack. The browser runs confidential LLM inference for its built-in AI on NEAR AI Cloud TEEs, so a user’s prompts are processed inside an attested enclave the model host can’t read. Brave documents the integration on its own engineering blog, which makes it a third-party reference rather than a NEAR announcement, though neither party has published request volumes.

What doesn’t work

The AI pivot is still early. NEAR AI Cloud and Private Chat launched in December 2025, the near.com super-app in February 2026. IronClaw and the Confidential GPU Marketplace are announced with minimal usage data. The Bermuda pilot is the first dent in the no-traction story, but it remains a single prototype: NEAR AI Cloud and the Confidential GPU Marketplace publish no revenue or compute-hour metrics, and the near-ai org has zero public models or datasets on HuggingFace. The research lab is well credentialed; the products stay pre-traction on the numbers.

TVL tells a stark story: about a third of the $607 million raised on DeFiLlama’s September 2026 7-day median, for a chain claiming 46 million monthly active users. That’s a DeFi depth problem. Either the users aren’t bringing capital, the DeFi applications aren’t compelling enough, or the user count is inflated. Probably all three to varying degrees.

Then there’s the collapsed backer problem. Three Arrows Capital co-led a $150 million round in January 2022 and collapsed five months later. Alameda Research and FTX Ventures were also backers. Forced liquidation of NEAR holdings from these bankruptcy estates has created sell pressure for years. Whether that overhang is fully resolved isn’t clear.

Foundation influence remains significant despite emerging governance. The AI pivot, near.com launch, and major strategic decisions appear to be Foundation and Pagoda-driven rather than community-governed. The House of Stake system is new, and broader token-holder participation rates are still emerging.

One more transparency concern: the 2022 wallet seed phrase incident. A security firm discovered that seed phrases were being leaked to Mixpanel when users set up email/SMS recovery. The bug was fixed the same day, but NEAR didn’t publicly disclose it until the security firm forced the issue two months later. No funds were lost, but the delayed disclosure undermines trust.

The risk

L1 competition is the existential challenge. NEAR competes with Ethereum, Solana, Avalanche, Aptos, Sui, and others for developer and user mindshare. Each has advantages in specific dimensions. The AI pivot must translate into concrete, measurable differentiation beyond a narrative advantage. If NEAR AI products fail to attract users beyond the crypto-native audience, the chain risks being a technically capable but competitively squeezed L1 with an AI marketing overlay.

TEETEETrusted Execution Environment. A hardware-secured region of a CPU or GPU where code runs in isolation, so even the machine's operator can't read what's happening inside. TEEs give decentralised AI inference privacy guarantees.Like a bank vault inside a bank. The bank owns the building, staffs the lobby, and runs the security cameras. But what's inside the vault is invisible to everyone, including the bank staff, unless the customer opens it.TEE in the glossary → trust assumptions matter. NEAR AI’s privacy guarantees rely on Intel TDX and NVIDIA Confidential Computing. TEEs have had historical vulnerabilities (Spectre, Meltdown, SGXSGXIntel Software Guard Extensions. The first widely-deployed TEE technology, introduced in 2015. SGX creates encrypted memory regions (enclaves) where code and data are protected from the operating system and the machine's owner.Like a safe deposit box at a bank. The bank owns the safe room and can see who comes in and out, but they can't see what's inside the boxes. SGX gives applications a private box on a shared computer.SGX in the glossary → attacks). Hardware-based trust isn’t equivalent to cryptographic guarantees. If a TEE vulnerability is discovered, the privacy promise of Private Chat and IronClaw would be immediately compromised.

The price performance creates a perception problem. Roughly 80% below ATH, with the peak driven by a $500 million Series B round from investors who subsequently went bankrupt. Regardless of technical progress, that chart doesn’t make it easy to attract new holders.

My position

I hold NEAR. The founder credentials are unmatched, the chain infrastructure is proven, and NEAR Intents is a standout product. The AI pivot is early and I’m watching execution closely: TVL needs to grow if the 46 million monthly users translate to on-chain value, NEAR AI products need published models and usage data, and the Foundation needs to keep ceding control to the House of Stake governance system. The bones are excellent. Whether the execution follows is what determines whether this holding grows.

Freedom Score: 62/100

NEAR Protocol scores 62/100 (C grade). Full methodology at Freedom Score Methodology.

Infrastructure decentralisation (14/20): 397 validators secure the network, with the top 100 producing blocks and chunks via Nightshade sharding. Validator entry is permissionless, gated by a seat price that moves with the active set rather than a fixed minimum. 599.6 million NEAR is staked and the spread is good by L1 standards: the largest validator holds 5.5%, the top 10 hold 35.9%, and 10 are needed to reach a third of stake. Nightshade 2.0 delivered 600ms blocks and 1.2-second finality. The chain has operated continuously since October 2020 with no consensus failures. Rainbow Bridge has successfully defended against multiple attack attempts. Network upgrade 2.13 activated on mainnet on 20 July 2026, bringing dynamic resharding (shards split automatically as state grows, without a coordinated validator vote each time) and post-quantum-safe signing (FIPS-204 / ML-DSA-65), which places NEAR among the first chains running a NIST-approved post-quantum signature scheme in production (reported by LiveBitcoinNews and BSCN, 21 July, both citing NEAR’s mainnet announcement). Geographic distribution is reasonable though concentrated in cloud providers. Pagoda runs some core RPC infrastructure, and the Foundation has outsized influence on development priorities.

Governance decentralisation (10/20): House of Stake governance launched late 2025 with veNEAR enabling stake-weighted voting, and the contracts are genuinely live: venear.dao holds 496 accounts and 5.3 million veNEAR against a 1.3 billion supply. The structure is more gated than the announcements suggest. The vote.dao contract carries a 5-account council and an equally sized reviewer set that decide which proposals reach a vote, caps active proposals at 3, and is itself owned and upgradeable by a Sputnik DAO. Its one approved protocol-parameter decision, removing the developer gas rebate, was passed in June 2026 and has still not reached the chain. The NEP process provides structured protocol upgrade pathways. However, the inflation halving (October 2025) was implemented despite the community governance vote failing to reach the required threshold. Validators adopted the upgrade anyway once 80% ran the new protocol version. Chorus One called this “a dangerous precedent.” The Foundation retains significant influence over strategic direction, treasury allocation, and core development priorities. The AI pivot, near.com launch, and major product decisions appear Foundation-driven rather than community-governed. Governance mechanisms exist but have been overridden when the outcome did not match the desired direction.

Token distribution fairness (7/15): The CoinList public sale (12% of genesis, three pricing tiers) provided broad retail access. All vesting is now complete, eliminating future vesting pressure. However, venture rounds gave early investors 10x pricing advantage over public sale participants ($0.0375 pre-seed versus $0.40 public). The collapses of 3AC and Alameda/FTX likely resulted in forced selling of NEAR holdings, contributing to the long price decline from the 2022 peak. Inflationary modelModelA trained neural network that takes inputs (text, images, audio) and produces outputs (more text, classifications, generated content). In DeAI the model is the thing that actually does the work.Like a very experienced apprentice who has spent years watching thousands of masters make furniture. They can't explain how they know when a joint is right, but they can make a chair that looks and functions like a Chippendale. The training is invisible. The output is what matters.Model in the glossary → has minted over 300 million additional NEAR since genesis, diluting initial holders.

Censorship resistance (11/15): NEAR is a permissionless L1 where anyone can create accounts, deploy contracts, and transact. 397 active validators provide strong block productionBlockA batch of transactions added to a blockchain at a set interval. Each block cryptographically links to the previous one, creating an append-only chain that can't be rewritten without redoing all the work since.Like a page in a ledger. Every page has a fixed number of entries, every page references the previous page, and once a page is filled and signed off it can't be edited without visibly invalidating every page that came after. The chain is just a very long series of these sealed pages.Block in the glossary → redundancy, though redundancy is not the number that matters for censorship: 10 colluding validators hold more than a third of stake, which is enough to halt the chain. Chain Signatures enable cross-chain operations without centralised bridges. Rainbow Bridge has proven resilient against attacks. Confidential transactions (February 2026) improve financial privacy. However, RPC node concentration could create access-layer censorship vectors, and the Foundation could influence application promotion in the near.com super-app.

Data sovereignty (8/15): Standard NEAR transactions are fully public, typical for L1s. NEAR AI Private Chat runs in TEEs with encrypted conversations the platform cannot access, and IronClaw provides a hardware-secured agent runtime; NEAR AI Cloud also exposes a public attestation endpoint that returns the TEE measurement registers and Intel TDX quote for any request, with an open-source verifier, so the confidentiality claim is checkable by anyone rather than taken on the operator’s word. Confidential Intents enable restricted-visibility cross-chain execution, and Universal Send (May 2026) extends confidential payments to any asset across supported chains from the near.com app. NEAR reports that close to half of near.com swap volume now routes through the confidential path, an early adoption signal that is not yet independently audited. These features are recent (December 2025 to May 2026) and their security rests on TEE hardware trust assumptions, which carry a history of side-channel vulnerabilities. NEAR accounts are pseudonymous, not anonymous, and the near.com app collects user data for AI features, though confidential mode is opt-in.

NEAR AI Cloud now also issues attestation tokens signed by Intel Trust Authority from a public endpoint, /v1/attestation/ita-token, documented in NEAR AI’s API reference. A token we fetched on 24 September 2026 verified against Intel’s published signing keys, so a third party can now check the gateway enclave without relying on NEAR’s own verifier. The same token grades the platform’s trusted computing base as OutOfDate, dated May 2025 and listing ten open Intel security advisories, and it applies no NEAR-specific measurement policy.

Open source and transparency (12/15): 475 public repositories across near and nearai GitHub organisations. nearcore is GPL-3.0 with 2,620 stars and 795 forks, and IronClaw has 12,631 stars; both were pushed on 24 September 2026, the day of our GitHub API read. Hacken audit scored 10/10. Multiple audits from Hacken, Sigma Prime, Blaize, and BlockSec. NEP process is public. Smart contracts are deployed and verifiable. The wallet seed phrase incident in 2022 was patched the same day but not publicly disclosed for two months, a transparency gap in incident response.

Path to improvement

Three changes would materially increase NEAR Protocol’s score:

1. Demonstrate AI product traction with verifiable data. Publish usage metrics for NEAR AI products: active users of Private Chat, compute hours on IronClaw, GPU Marketplace volume. Put models and datasets on HuggingFace. The AI pivot narrative is credible because of the founder; it’ll only be convincing because of the data. Without measurable traction, NEAR AI remains a well-credentialed research lab, not a production platform.
2. Grow TVL relative to market cap. The TVL-to-mcap ratio is low compared to peers (check DeFiLlama for current figures). Either incentivise DeFi migration to NEAR, build native financial primitives that use chain abstraction, or acknowledge that DeFi depth is not the growth vector. The current ratio undercuts the 46 million MAU claim.
3. Respect governance outcomes. The inflation halving delivered a good economic outcome, but implementing it after the community vote failed to reach threshold undermines governance credibility. Future protocol changes must follow the governance process to completion, even when the outcome is inconvenient. Extend House of Stake governance to cover treasury allocation, AI product strategy, and Foundation expenditure. Demonstrate that governance votes are binding, not advisory.

Returns Score: 73/100

NEAR scores 73/100 (B grade). Full methodology at Returns Score Methodology.

Token utility (16/20): NEAR has the strongest token utility of any project in this review. It’s the native gasGasThe fee paid to a blockchain to process a transaction. Gas is denominated in the chain's native token and varies with network demand. Sending a transaction without enough gas means the transaction fails and the gas is still consumed.Like the petrol that powers a car. You need to put petrol in to make the engine run. The amount of petrol you need depends on how far you're driving and how much you're carrying. If you run out, the car stops.Gas in the glossary → token for a proven L1 that has run continuously since October 2020. Staking secures an active set of 397 validators. Governance via veNEAR is live but has not yet decided a protocol parameter that has shipped. Storage staking locks NEAR for on-chain data. Every transaction, every smart contractSmart ContractA program stored on a blockchain that runs automatically when its conditions are met. Smart contracts are how blockchains do anything beyond just transferring tokens — DeFi, NFTs, DAOs, and DeAI infrastructure all run on smart contracts.Like a vending machine. You put in the right input and it produces the expected output, no human operator required. The rules are fixed in the machine itself, anyone can use it, and nobody can stop a transaction in the middle.Smart Contract in the glossary → deployment, every cross-chain Intent burns NEAR through the 70% fee burn mechanism. This is textbook L1 utility. The token is structurally necessary for the chain to function.

Value accrual (15/20): Two complementary mechanisms now operate. The 70% base gas burn creates a direct link between on-chain usage and token scarcity. The Intents fee switch, activated 23 February 2026, directs cross-chain Intents fees into open-market NEAR purchases. DeFiLlama shows gross Intents fees of $53 million over the trailing year ($7.6 million in the last 30 days), of which $8.5 million is protocol-retained; the balance reaches integration partners who can resell, so the durable buy pressure is the retained slice. The bought NEAR is held in a multisig rather than burned, and the buyback account had collected about a tenth of the NEAR issued since it opened in March 2026, so the mechanism is far from offsetting the ~32 million NEAR a year of new supply. It’s still a concrete, verifiable, revenue-driven value accrual mechanism, which most L1s lack entirely.

Supply dynamics (14/20): No maximum supply remains a structural negative. However, the inflation halving in October 2025 cut annual issuance from 5% to 2.5% (~32 million NEAR per year), materially reducing dilution. The 70% base gas burn and the Intents buybackBuybackUsing protocol revenue to purchase tokens on the open market, usually to burn them or return them to a treasury. Buybacks convert business income into upward pressure on the token by reducing circulating supply.Like a public company using profits to repurchase and retire its own shares. The cash leaves the company's balance sheet, the share count drops, and every remaining shareholder owns a slightly bigger slice of the same business.Buyback in the glossary → mechanism lower net issuance, but nowhere near deflation: together they offset a small fraction of new supply, and the bought NEAR is held rather than burned. All genesis vesting is complete, which removes insider sell pressure from vesting and leaves no remaining team or VCVCVenture Capital. Private investors who fund projects at an early stage in exchange for equity or token allocations. VC rounds are typically pre-launch, at steep discounts to any future public price, with multi-year vesting.Like angel investors in a startup who buy shares before the company goes public. They take more risk because the company might fail, so they get a better price. Once the company IPOs they can sell, and the public market pays whatever price it thinks is fair.VC in the glossary → cliffs. This is a significant advantage over most projects in this space. The circulating share itself is disputed: aggregators differ by about 4% of supply, and we haven’t identified who holds that gap. The halving was economically positive but governance-negative. It was pushed through despite the community vote failing to reach threshold, implemented by validator software adoption rather than governance mandate.

Revenue sustainability (14/25): NEAR Intents has processed $34.4 billion in cumulative volume per DeFiLlama. Gross Intents fees came to $53 million over the trailing twelve months, with $7.6 million in the last 30 days. Volume reached $4.8 billion in the last full month, dominated by stablecoin swaps across roughly 35 chains, which points to substantive cross-chain activity rather than speculative churn. This is production-grade revenue at scale, independently verifiable against DeFiLlama rather than self-reported.

The accrual nuance matters. Of the gross Intents flow, $8.5 million over the trailing year is protocol-retained revenue, with the balance distributed to integration partners who can sell into the market, so the slice that creates durable NEAR demand is the retained portion. The remaining concern is the AI pivot: NEAR AI Cloud and the Confidential GPU Marketplace publish no revenue metrics, so those streams stay unproven. The Intents revenue alone justifies a meaningful sustainability score.

LiquidityLiquidityHow easily a token can be bought or sold without moving the price. High liquidity means you can enter or exit large positions quickly at the quoted price. Low liquidity means even small trades can swing the market.Like the difference between selling a house and selling a share of Apple stock. The house might be worth more on paper, but finding a buyer at that price takes weeks. The Apple share converts to cash in one click.Liquidity in the glossary → and access (14/15): Tier-1 CEX listings across Binance, Coinbase, OKX, Kraken, KuCoin, and others, 89 venues in all, the widest coverage of any project we track. Turnover runs near an eighth of the market cap daily, so seven-figure positions move without meaningful slippageSlippageThe difference between the expected price of a trade and the price you actually get when the trade executes. Slippage usually goes against the trader and gets worse with bigger trades or thinner markets.Like trying to buy 1000 bananas at the corner shop. The first few are at the marked price, but by the time you've bought them all you've moved the price up because there are no more bananas left at the original level. The shop has to restock at higher cost.Slippage in the glossary →. One qualification we should state because we hold NEAR: Binance leads at about a sixth of reported flow and KuCoin at a tenth, but four of the next six places are second-tier venues, so a good deal of the volume behind that headline is not on the exchanges the score rests on. US access moved from filings to a listing in September 2026. Bitwise’s single-asset spot NEAR ETF went effective with the SEC on 24 September and NYSE Arca certified it to list as NRR on 28 September, with a 0.75% fee and staking inside the fund (SEC EDGAR). The score waits on NRR showing assets and volume once it trades. Grayscale’s GSNR conversion is still at S-1 stage, and Bitwise also includes NEAR in 11 multi-asset strategy ETFs filed in December 2025. A Bitwise NEAR Staking ETP already trades on Deutsche Börse Xetra; Bitwise’s CEO reported it at roughly $36 million AUM in May 2026 after a week of net inflows. The roughly 80% decline from ATH remains a sentiment headwind, but the liquidity and institutional access infrastructure is among the strongest in crypto.

Path to improvement

Three changes would materially increase NEAR Protocol’s returns score:

1. Burn what the buyback buys. NEAR bought through the Intents fee switch sits in a multisig. Burning it, or committing it to a published policy, would turn a held balance into a supply reduction. Months where burns exceed the ~2.7 million NEAR monthly issuance are still far off, so this is the step that would make the fee switch count.
2. Demonstrate AI-driven fee growth. Publish monthly fee revenue broken down by source: base chain transactions, Intents, NEAR AI Cloud, Confidential GPU Marketplace. The Intents revenue is strong and verified. Show whether the AI pivot is generating incremental fees or just rebranding existing activity.
3. Grow TVL to validate the user base. Even on the most generous reading of the DeFiLlama series, locked value per claimed monthly active user is a few dollars at best, which strains credibility. Either incentivise meaningful DeFi migration that brings TVL in line with the user base, or provide granular user activity data that explains the disparity.

Score change log

Date	Score	Change	Reason
2026-10-13	Data	N/A	Validator set re-measured on-chain and against NearBlocks: the halt coefficient is now 9, not the hand-typed eight, so the censorship line is bound to the live figure; the score evidence’s stale 698 (staking pools) is now the measured 414. Censorship 11/15 held.
2026-10-13	Data	N/A	Base-chain revenue finding re-anchored: DeFiLlama now counts only the burned 70% of gas as revenue, so the trailing year fell about 30% on basis alone while gross gas held flat (NearBlocks agrees), and the “falling steadily” trend no longer holds. Scores unchanged.
2026-10-09	Data	N/A	Corrected the treasury split: the chain pays treasury.near 10% of issuance at each epoch boundary, so stakers get 90%, not the whole 2.5% we wrote from 13 August off a config field that reads 0. Split and staking yield now measured and bound; scores unchanged.
2026-10-08	Data	N/A	NEAR AI now documents the inference-credit formula (stake × NEAR price × APY, rolled over), replacing our “no published rate” reading of 13 August; agent hosting stays at staked NEAR ÷ 100. Scores unchanged.
2026-10-07	Data	N/A	Turnover restated from near a tenth to near an eighth of market cap (7-day median 12.5%, early October). Scores unchanged.
2026-10-06	Data	N/A	”Approaching the deflationary threshold” withdrawn: Intents buybacks are held in a multisig, not burned, and have collected about a tenth of issuance since March 2026. Tokenomics now opens on demand, value capture, live status and dilution. Scores unchanged; Value Accrual 15 and Supply Dynamics 14 flagged for review.
2026-10-04	Data	N/A	Added the 1 October NEAR Intents exploit: about $3.8M lost through an Omni and Intents contract bug, patched, and returned in full on 2 October (DeFiLlama hacks record plus NEAR’s own posts). Scores unchanged.
2026-10-04	Data	N/A	Treasury rate re-read at 0 on protocol 86, and the 30 September 1.6% issuance proposal added as pending; its “90/10 split” wording doesn’t match the chain. Scores unchanged.
2026-10-04	Data	N/A	The developer rebate removal now has a release, nearcore 2.14.0 (protocol 87, live on testnet only); mainnet still pays the 30% rebate and burns 70%. Scores unchanged.
2026-10-04	Freedom	Governance 11 → 10 (Freedom 63 → 62, C held)	House of Stake is live, but a five-account council gates every vote and its one approved protocol decision has not reached mainnet.
2026-09-29	Data	N/A	Bitwise’s single-asset spot NEAR ETF (NRR, filed May 2025) went effective 24 Sep and NYSE Arca certified the listing 28 Sep; corrects the April framing of Bitwise as multi-asset only. Liquidity & Access 14/15 held.
2026-09-24	Data	N/A	Sweep: ATH drawdown 93% to ~80%, supply 1.289B to ~1.30B (bound), validators bound, GitHub counts, TVL-to-raise ratio refreshed; the false “raise dwarfs market cap” line dropped; the buyback claim now attributed to NEAR with our trace. Scores unchanged.
2026-09-24	Data	N/A	NEAR Intents volume, gross fees and retained revenue now bound to a weekly DeFiLlama refresher; the hand-typed ~$40.4M / ~$2.9M / ~$4.6M / ~$2.0B from August were stale ($48.3M / $6.2M / $6.6M / $2.8B). Scores unchanged.
2026-09-24	Data	N/A	The Supply Dynamics rationale still said “99.99% circulating” after the 17 August log dropped it; it now says the circulating share is disputed by about 4% of supply. Scores unchanged.
2026-09-24	Data	N/A	Added near.com perps (a Hyperliquid integration, funding legs confidential only) and refreshed cumulative Intents volume from $24.4B to $30.7B per DeFiLlama. Scores unchanged.
2026-09-24	Data	N/A	NEAR AI Cloud’s Intel Trust Authority tokens verify against Intel’s keys, but the same token grades the platform TCB OutOfDate with ten advisories. Data Sovereignty 8/15 held.
2026-09-23	Data	N/A	Integrity records re-cut: 5 claims now quote NEAR’s own pages verbatim and 8 are our own measurements. Scores unchanged.
2026-09-07	Data	N/A	WS3b consensus rail: the research JSON still carried the 698 validator count that the page corrected in August, so an agent reading our dataset got a figure 68% above the set. Synced, and the halt coefficient added: eight colluding validators exceed a third of stake. Censorship 11/15 held. We hold NEAR.
2026-09-06	Data	N/A	WS3b liquidity rail: all five tier-1 listings confirm across 89 venues, the widest in the corpus, and the hardcoded $100M volume floor is replaced with turnover near a tenth of market cap. Added the composition qualification: four of the eight largest venues are second-tier. Liquidity & Access 14/15 held.
2026-08-20	Data	N/A	Freshness banding paused on the chain_tvl verdict until 2026-09-19 while DeFiLlama’s Near adapter flaps, so it stops routing a spurious ~40% drift each time a refresh lands on the low state. The pause is time-boxed and warns once it expires.
2026-08-19	Data	N/A	DeFiLlama’s NEAR chain TVL is unreliable: it swung 92.3M/54.6M/92.5M/54.8M across 15-19 August. Traced to Rhea Lend, whose supplied leg drops ~$36M in and out of the index while its borrowed leg holds steady. Absolute TVL removed from all four remaining surfaces; the 7d median that produced the 2026-08-17 figure smooths exactly this fault.
2026-08-19	Data	N/A	Correction. Chain TVL was still stated as $111M in five places, against the $92.3M the 2026-08-17 read settled (DeFiLlama 7d median). All five re-dated to the current figure, and the per-user locked value recomputed.
2026-08-17	Data	N/A	Attempted to measure the Foundation treasury; the answer is that it cannot be: the two nameable accounts hold ~64K NEAR, foundation.near has zero staked across all 40 pools it ever used and no lockup. Genesis percentages now labelled as genesis, not current holdings.
2026-08-17	Data	N/A	Dropped the “99.99% circulating” claim. It came from CoinGecko, which models no lock-up at all and reports a circulating figure larger than its own total; NearBlocks puts ~54M NEAR (4.1%) non-circulating. On a held position the flattering source had gone unchecked. Supply Dynamics flagged for September.
2026-08-17	Data	N/A	Chain TVL corrected from $53.7M to $92.7M. The old figure was a DeFiLlama index dropout, captured on 16 August and wrongly confirmed by re-reading the same endpoint. The stored figure is now a trailing-7d median.
2026-08-15	Data	N/A	Mechanism audit: “Staking for NEAR AI” is an ordinary staking pool (nearai.pool.near) with reward_fee_fraction 100/100, not protocol-direct routing. The stake behind the credits reads ~678,000 NEAR across five delegator accounts, so the obligation we called unsizeable is on-chain.
2026-08-15	Data	N/A	House of Stake read from vote.dao and venear.dao: a five-account council gates which proposals reach a vote, 483 accounts hold veNEAR, and no inflation proposal exists, so the halving was not a veNEAR decision. Token Utility and Governance evidence corrected.
2026-08-15	Data	N/A	vote.dao approved removing the developer gas rebate in June 2026, but the live protocol config still returns the 30% rebate. Recorded in the build section as a benefit on notice.
2026-08-15	Data	N/A	Validator entry is a per-epoch seat price, not the fixed 25,500 NEAR minimum we carried. With 600ms blocks an epoch runs ~7.4 hours, so unstaking is roughly 15-22 hours rather than 24-36.
2026-08-13	Data	N/A	Worth recording about our own source: revenue.near.org says its figures are “Pulled directly from onchain sources” and then, in the same footer, that “Figures shown are illustrative”. That hedge is why we reconcile the net figure against two independent DeFiLlama adapters rather than resting on NEAR’s dashboard.
2026-08-13	Data	N/A	Read the live protocol config on three independent RPC endpoints: the 10% protocol-treasury slice we described is gone (protocol reward rate 0), so stakers take the whole 2.5%. Corrected the treasury split and the staking yield to ~5.3% nominal.
2026-08-13	Data	N/A	Enumerated the validator set on-chain: 414 active validators, not the 698 we carried (that figure counts staking pools). Nakamoto coefficient is 8. Corrected across the page and migrated NEAR onto the verdict spine, including a match on its own revenue dashboard.
2026-08-04	Data	N/A	DeFiLlama refresh of NEAR Intents metrics (verified first-hand): cumulative volume to $24.4B, gross fees ~$40.4M annualised (Aug 2026, 30d ~$2.9M), protocol-retained ~$4.6M TTM, monthly volume ~$2.0B (was $19B/$36.9M/$4.0M/$1.9B, Jun). Dropped the stale “~2.1M bought back” fact-strip row. TVL untouched. No score change.
2026-08-03	Editorial	N/A	Added the 31 July 2026 “Staking for NEAR AI” launch (staked NEAR → AI credits across 43 models) to staking prose. Resolved the deferred Infrastructure 14→15 as HELD: 2.13’s improvements do not touch the dimension’s structural deductions (Pagoda RPC, Foundation influence, cloud concentration). No score change.
2026-07-25	Data	N/A	Network upgrade 2.13 went live on mainnet 20 July 2026, activating dynamic resharding and post-quantum-safe signing (FIPS-204 / ML-DSA-65); corrected the stale “not live on mainnet yet” body line. Verified via LiveBitcoinNews + BSCN (21 July). No score change (Infrastructure 14 → 15 deferred to monthly).
2026-06-28	Data	N/A	DeFiLlama refresh + accrual correction (NEAR-only). Refreshed the NEAR Intents fee run-rate ~$34.6M → ~$36.9M annualised gross (June 2026, 30d ~$3.0M) across the description, frontmatter (valueAccrual, supplyDynamics, revenueSustainability), highlights, pillars, fact strip, body prose, and the Value Accrual + Revenue Sustainability dimensions. Corrected the “100% of Intents fees into NEAR buybacks” framing: the gross flow buys NEAR on the open market, but only ~$4.0M/yr is protocol-retained since the Feb-2026 fee switch (the rest reaches integration partners who can resell), so the durable buy pressure is the retained slice. Verified via DeFiLlama near-intents (gross $36.9M / protocol revenue $4.0M), now wired into meta.data_sources[] and auto-tracked by refresh:primary (drift-guarded). Removed the low-trust “2.1M NEAR bought back” figure (single low-trust source, inconsistent with other reports). Added: NEAR AI Cloud public attestation endpoint (cloud-api.near.ai/v1/attestation/report returns TEE measurement registers + Intel TDX quote with an open-source verifier, so confidentiality is independently checkable) to Data Sovereignty; Brave as a second named confidential-inference consumer (Brave engineering blog, no published volumes). No score change. Deferred candidate strengthened: Freedom Data Sovereignty 8 → 9 now supportable on verifiable-attestation grounds (decide at monthly).
2026-05-31	Editorial	N/A	May 2026 review (NEAR-only). Added: Government of Bermuda x NEAR AI prototype assistant (14 May, first named NEAR AI customer, citizen data in hardware-enforced enclaves; verified via Bermuda Royal Gazette + ETHNews); Universal Send confidential payments live on near.com (late May); confidential path now ~half of near.com swap volume (NEAR-reported, attributed, not independently audited); network upgrade 2.13 dynamic resharding + post-quantum FIPS-204 / ML-DSA signing expected June 2026 (CoinDesk + NEAR). Refreshed Bitwise NEAR Staking ETP from $28.3M / 3.08% to ~$36M AUM (Bitwise CEO, May 2026). Checked the $VVV-on-NEAR-Intents claim and deliberately omitted it: VVV is genuinely swappable via Intents (confirmed on the chaindefuser token registry, Base contract 0xacfe6019…21bf), but Intents routes 180+ tokens including memecoins and tokenised equities, so a single token being swappable is not DeAI-significant. The NEAR AI Cloud private-inference relationship with Venice is already documented and scored on the Venice page (TEE enclave operator since March 2026, attested), so nothing new accrues to the NEAR page from it. No score change: AI products gained a named pilot but still publish no revenue or usage metrics, so “AI revenue unproven” stands and nothing crosses a grade boundary. Deferred candidates: Freedom Data Sovereignty 8 to 9 if confidential-swap adoption is independently audited; Returns Revenue Sustainability if NEAR AI publishes usage/revenue.
2026-05-26	Data	N/A	DeFiLlama refresh of NEAR Intents metrics: cumulative volume $15.2B → $19.17B (+26%); annualised fee run-rate ~$33M → $34.61M (30d fees $2.84M); 24h DEX volume $92.3M. Updated frontmatter (description, valueAccrual, supplyDynamics, revenueSustainability), body prose (NEAR Intents sections, Value Accrual and Revenue Sustainability dimensions), tldr, highlights, fact strips. Removed unverified $25M / $27M cumulative-fees figures from prose (DeFiLlama exposes annualised + 30d only; cumulative fees not directly fetched, hold pending direct verification). Removed stale “24-hour fees ~$135K” line. No score change; trajectory continues but no grade-boundary trigger. Research JSON Intents fields also refreshed.
2026-05-02	Editorial	N/A	Added note on 25 April 2026 Litecoin MWEB reorg exploit: NEAR Intents took ~$600K exposure on a 11,000 LTC→BTC swap reversed by a chain-side reorg (Litecoin bug, not NEAR). Team pledged reimbursement. No score change; cross-chain risk inheritance flagged.
2026-04-11	Returns	70 → 73	Intents fee run-rate verified at ~$33M annualised per DeFiLlama (cumulative fees $27.56M, 30d $2.78M, 24h $134,855). Grayscale filed Form S-1 on 20 January 2026 to convert NEAR Trust into spot ETF on NYSE Arca (ticker GSNR, Coinbase Custody, BNY Mellon administrator). Value Accrual 14→15, Revenue Sustainability 13→14, Liquidity & Access 13→14.
2026-04-11	Data	N/A	Verified Intents fee run-rate and cumulative fees against DeFiLlama APIAPIApplication Programming Interface. A structured way for one piece of software to talk to another. In DeAI, APIs let applications request inference from a model without running the model themselves.Like a waiter in a restaurant. You don't walk into the kitchen and cook your own meal. You tell the waiter what you want, they tell the kitchen, the kitchen cooks it, and the waiter brings it back. The API is the waiter.API in the glossary →. Grayscale S-1 filing verified via SEC coverage. Bitwise ETF framing clarified (multi-asset Strategy ETFs, not single-asset spot).
2026-03-24	Data	N/A	Intents volume $13B → $15.2B. Removed unverifiable $45M annualised fees. Verified against DeFiLlama.
2025-03-03	Both	N/A	Initial publish. Freedom 62/100, Returns 70/100.