Near began in 2017 as an AI project, then pivoted to blockchain, and is now circling back to its origins. ForkLog spoke with protocol co-founder Ilya Polosukhin about recent upgrades and the vision for what comes next.\u00a0<\/p>\n
ForkLog (FL): <\/strong>Why didn\u2019t AI work out at the start, and what does blockchain for AI mean today?<\/strong><\/p>\n Ilya Polosukhin (I. P.): <\/strong>I have worked in machine learning since 2007\u20132008, including at Google Research. One of our projects was the Transformers system, which underpins ChatGPT, Midjourney, Gemini and so on. In 2017 I left Google and we teamed up with Alex Skidanov to teach computers to program. Today this is called vibe-coding; back then it sounded like science fiction.\u00a0<\/p>\n At the time, graphics cards weren\u2019t good enough to deliver a reliable product. We also ran into the problem of paying students worldwide who helped us collect data. We turned to blockchain as a coordination and payment system that would let us pay people without intermediaries.\u00a0<\/p>\n In 2024 we linked our conceptual work in this direction to AI. The original Near AI vision is: you talk to a computer, it writes all the code itself, applications are created on the fly and delegate tasks to other services in the user\u2019s interest rather than corporations\u2019.<\/p>\n To earn users\u2019 trust in such AI, we are building confidential-computing infrastructure and decentralised machine learning. Together this forms our view of the future internet.<\/p>\n FL: In 2022, when LUNA and <\/strong>UST<\/strong> were hyped, you launched USN. Then came tap-to-earn apps, <\/strong>STEPN<\/strong> \u2014 you partnered with Sweatcoin; later AI boomed and you returned to the original concept. Is your focus constantly shifting?<\/strong><\/p>\n I. P.: <\/strong>Near is a blockchain on which you can build anything. Here Wallet, for example, initially used \u201ctap-to-earn\u201d as a go-to-market strategy, and now it is a network-abstracted wallet that works with 140 blockchains. The Sweatcoin app is more than ten years old with over 120m installs \u2014 the partnership brought Near a new audience.<\/p>\n Our goal is not to chase hypes but to expand access to sovereignty and self-custody of cryptocurrency. All collaborations are channels to deliver technology to millions of users. With private AI it\u2019s the same: the word \u201cAI\u201d itself matters less than using it to bring our capabilities to people.<\/p>\n FL: <\/strong>How secure is account abstraction from a technical standpoint?<\/strong><\/p>\n I. P.: <\/strong>Originally, to recover a Near Wallet account the seed phrase was sent straight to email. It wasn\u2019t the best, but it worked. Since 2023 Near has been moving to passkeys supported by Apple, Google and Microsoft. They are stored in the device\u2019s Secure Enclave, and recovery happens via iCloud or a Google account. This approach is considered more reliable.<\/p>\n Near\u2019s main advantage is protocol-level account abstraction. You can link several keys (devices, custodians) to one account, set permissions and use two-factor authentication. Chain Signature technology blocks abuse even if a seed phrase leaks.\u00a0<\/p>\n FL: <\/strong>Tell us about the latest protocol upgrades.<\/strong><\/p>\n I. P.: <\/strong>We completed the original sharding design with an improvement, Stateless Validation<\/span>. In the old model we needed \u201cfishermen\u201d to check shards, but it was slow, required a separate crypto-economy and did not scale well. We found a more pragmatic approach, close to zero-knowledge verification, where validators receive a part of the block with all necessary information and verify it without access to shard state.<\/p>\n The second big update is halving block time to 600 ms. This gives 1.2-second finality \u2014 ten times faster than Solana and 300 times faster than Ethereum. The goal is 200 ms.<\/p>\n In parallel we are expanding the number of shards; we are now up to nine. This increases the network\u2019s compute power without losing accessibility. Near keeps scaling while remaining runnable even on ordinary computers, which other blockchains cannot yet offer.<\/p>\n FL: Many blockchains<\/strong> have a core positioning. What is Near oriented towards?<\/strong><\/p>\n I. P.: <\/strong>In the future, everything \u2014 from ordering food and hailing a taxi to transactions and working across networks \u2014 will be executed through a personal AI agent that runs on the user\u2019s side. To enable this, we are building infrastructure that connects everything to everything, and in parallel creating an AI layer where data remain private.\u00a0<\/p>\n Centralised companies today use user logs, and leaks only confirm the risk. Blockchain offers an alternative internet where privacy and confidentiality are built into the architecture.<\/p>\n Product quality matters: as long as centralised solutions are more convenient, users choose them. The task is to make decentralised products better and to seek niches where privacy is especially valuable, for example in medicine or finance.<\/p>\n FL: The <\/strong>NEAR token\u2019s price doesn\u2019t react much to announcements and upgrades. Why?<\/strong><\/p>\n I. P.: <\/strong>You need more buyers than sellers. There are several systems that use the token: staking, future governance and various DeFi applications with network abstraction. We are at an early stage, but activity is steadily growing.<\/p>\n There are protocols that generate revenue in NEAR, for example NEAR Intents, which aggregates liquidity and buys back some tokens to burn. These are only first steps, so fees are minimal \u2014 the goal at this stage is not monetisation but user growth and adoption.<\/p>\n The same goes for AI: the focus is on expanding the audience and popularising the product.<\/p>\n FL: <\/strong>In January 2022 you raised $150m in a round led by Three Arrow Capital, which later <\/strong>went bankrupt<\/strong> due to the <\/strong>Terra<\/strong> collapse. What\u2019s happening with the tokens they held?<\/strong><\/p>\n I. P.: <\/strong>All NEAR belonging to investors. Only Near Foundation assets remain, which will be unlocked over about a year.\u00a0<\/p>\n When NEAR rallied last year, some investors exited: some funds were ending, others sold all their assets. Many assets passed to other companies and were then liquidated \u2014 partly via OTC, partly on the market. Essentially, everyone who wanted to or had to sell has already sold.<\/p>\n The key task now is to attract new users, developers, projects and supporters who want to join the ecosystem and help develop it further.<\/p>\n FL: <\/strong>In a 2023 interview you spoke about reaching 1bn users in the NEAR ecosystem. Have you hit those numbers?<\/strong>\u00a0<\/p>\n I. P.: <\/strong>Not a billion yet. We currently have around 50m monthly active users. The goal was to reach a billion per month.\u00a0<\/p>\n There are no specific deadlines for this goal; we continue to grow on all fronts.\u00a0<\/p>\n FL: <\/strong>In Ukraine the first reading passed a <\/strong>tax bill<\/strong><\/a> with an 18% rate. What do you think?<\/strong><\/p>\n I. P.: <\/strong>The question is what exactly is considered capital gains and taxed. In the US, for example, even converting ether to StakeETH or NEAR to StakeNEAR can be considered a taxable operation, although it\u2019s nowhere clearly spelled out. There are too many instruments and ways to work with crypto, and in the end nobody understands how to act correctly.<\/p>\n Clear rules would benefit everyone. An example is the GENIUS Act<\/a> on stablecoins in the US, which opened the gates for innovation: companies began launching their own blockchains and engaging actively. That shows regulation can stimulate growth.<\/p>\n For Ukraine the issue is especially important: the country needs any financial support, but it\u2019s hard to define the rules correctly, especially for the general public. We want to believe they will be workable and applicable rather than a formality everyone will bypass.<\/p>\n FL: <\/strong>Does cyclicality in the crypto market persist? Will we see another hype cycle in autumn\u2013winter?<\/strong><\/p>\n I. P.: <\/strong>Cycles are always obvious in hindsight, and it\u2019s hard to say precisely what has changed. There are expectations now that rate cuts in the US will attract more institutional capital \u2014 regulated and able to hold crypto \u2014 which didn\u2019t exist before.<\/p>\n New investment avenues have appeared, such as access to NASDAQ via ETFs and Digital Asset Treasuries<\/a>, which were previously absent, opening capital inflows into the ecosystem.<\/p>\n But regardless of market swings, cycles come and go. The main thing is to keep building, improve products, attract users and strengthen the ecosystem without relying on short-term market waves.<\/p>\n FL: <\/strong>What\u2019s in your portfolio?<\/strong>\u00a0<\/p>\n I. P.: <\/strong>Mainly <\/strong>NEAR, a bit of bitcoin and many memecoins on Near.<\/p>\n