More about our guest today Ans : I entered the crypto space in mid-2017. Prior to that, I worked at IBM for three years as a software developer, primarily in data analytics. I left my 9-to-5 role in February 2018 and have been fully dedicated to crypto ever since. I first learned about NEAR while conducting in-depth project analysis. I was immediately drawn to it as a Layer1 blockchain capable of competing with Ethereum. Key strengths that stood out included its human readable addresses, negligible fees, near-instant finality, and significantly higher throughput compared to competitors at the time. Further research revealed strong institutional backing, with hundreds of millions raised in venture funding. The team also impressed me, particularly co-founder Illia Polosukhin, an ex-Google AI researcher and co-author of the seminal 2017 paper “Attention Is All You Need” (which introduced the Transformer architecture powering modern LLMs like GPT), and Alexander Skidanov, an ex-MemSQL engineer. I view NEAR as one of the most strategically positioned Layer1s for 2026–2030. Rather than competing purely on speed or DeFi TVL, it is building the foundational infrastructure for the agentic internet, where AI agents perform the heavy lifting on behalf of users. If executed successfully, NEAR possesses one of the highest upside potentials among established Layer1s.

$NEAR : Review📜 What if the man who co-authored the research paper that made ChatGPT possible decided that AI shouldn't be owned by corporations and built an entire blockchain to make sure it isn't? Meet NEAR Protocol - a Layer-1 blockchain co-founded by Illia Polosukhin, co-author of "Attention Is All You Need," the 2017 transformer paper that birthed GPT, BERT, and the entire modern AI revolution. 1 million TPS. Chain Abstraction across 35 chains. User-Owned AI with confidential computing. Post-quantum security. Dynamic resharding. The blockchain that's betting the future of AI runs on privacy, ownership, and decentralization. Let's explore how NEAR is building the Agentic Web. 👇 ⚪ NEAR at a Glance NEAR is a public blockchain platform that uses a proof-of-stake consensus mechanism and provides smart-contract functionality. It was founded in 2018 by Illia Polosukhin and Alexander Skidanov, with mainnet launching in 2020. The $NEAR token powers transaction fees, staking, storage, and governance. NEAR surged nearly 30% in May 2026 on the dynamic resharding announcement, with the Bitwise NEAR Staking ETP signaling strong institutional demand. Marketplace Insight: In December 2025, NEAR hit 1 million TPS in a publicly verifiable benchmark using Nightshade 2.0 sharding, with sub-$0.01 fees and 1.2-second finality. On October 30, 2025, NEAR's inflation rate was cut from 5% to 2.5% annually, roughly halving new token issuance from ~64M to ~32M NEAR per year. NEAR led all major blockchains with over 950 GitHub commits in H1 2025, the highest development activity in the industry. When the co-author of the transformer paper builds a blockchain for AI, that's not a narrative pivot. That's the original thesis coming full circle. ⚪ Mission NEAR's mission is to build the Agentic Web: an internet where AI agents operate autonomously on behalf of users, with privacy, ownership, and decentralization guaranteed at the protocol level. NEAR has evolved into a comprehensive AI-driven ecosystem. By prioritizing usability and massive scalability, it serves as the invisible backbone for a faster, more intelligent internet where humans and AI agents interact seamlessly. 🔵 A Brief History NEAR Protocol was founded in early 2017 by Illia Polosukhin and Alexander Skidanov. Initially, the project was established as Near.ai, an artificial intelligence startup focused on program synthesis. Polosukhin was a former engineering manager at Google Research, where he co-authored the seminal 2017 paper "Attention Is All You Need," which introduced the transformer architecture. This paper is the foundation of GPT, BERT, ChatGPT, and the entire modern AI revolution. All eight co-authors have since left Google. Alexander Skidanov holds a Master's in Computer Sciences from Izhevsk State Technical University. He began his career at Microsoft before moving to MemSQL as Senior Software Developer and Director of Engineering. He received the ICPC gold medal in 2008. Skidanov also held a position at OpenAI in the period leading up to ChatGPT's release in 2022. The two founders initially built Near ai as a pure AI company before pivoting to blockchain when they realized no existing chain could meet their scalability needs. They built NEAR Protocol from scratch. NEAR raised over $500M across multiple rounds, including an $800M ecosystem fund announced at NEARCON 2021. Mainnet launched in 2020. The protocol rapidly grew into a top-20 blockchain with its Nightshade sharding architecture. In October 2025, NEAR halved its inflation from 5% to 2.5%. In December 2025, the network achieved 1M TPS in a publicly verifiable benchmark. In February 2026, NEAR launched IronClaw, a security-focused AI assistant operating inside isolated WebAssembly environments. In March, Venice integrated with NEAR AI for verifiably private model inference. The Near com super-app launched in February 2026, abstracting all blockchain complexity with feeless cross-chain swaps across 35 chains, AI assistance, and agent automation in a single interface. In May 2026, NEAR added FIPS-204 post-quantum cryptography for future-proof key management. Dynamic resharding is scheduled for June 2026, allowing the blockchain to automatically add shards as demand grows without human intervention. 🔵 Ecosystem Narrative NEAR's ecosystem has evolved from a scalable smart contract platform into the leading infrastructure layer for AI agents, chain abstraction, and user-owned intelligence. Every major upgrade in 2025-2026 reinforces this thesis. Key dynamics include: ➛ Chain Abstraction hides all blockchain complexity from users. Through Near com, users sign up with email or FaceID and manage assets across 35 chains without bridging, seed phrases, or understanding which chain they're on. NEAR Intents route cross-chain execution automatically. ➛ Nightshade 3.0 Sharding achieves 1M TPS with 1.2-second finality and sub-$0.01 fees. Dynamic resharding (June 2026) will automatically add shards as demand grows with no human intervention. ➛ User-Owned AI with IronClaw confidential computing processes AI inference inside hardware-isolated TEEs. Venice AI integrated for verifiably private model inference. Your prompts stay encrypted even while being processed. ➛ NEAR AI Agent Market enables autonomous agents to handle payments, trading, data management, and cross-chain operations on behalf of users. ➛ OceanPal/SovereignAI raised $120M to build sovereign AI infrastructure, positioning NEAR at the center of the decentralized AI compute narrative. ➛ Post-quantum security via FIPS-204 cryptography (May 2026) enables quantum-safe key rotation in a single transaction. NEAR joins Algorand as one of the few chains with post-quantum protections live. ➛ Ref Finance / RHEA Finance combines exchange and lending into a chain-abstracted liquidity hub connecting Bitcoin, NEAR, and Ethereum. ➛ Grayscale Near Trust (GSNR) and Bitwise NEAR Staking ETP provide institutional investment vehicles. ⚪ Token Utilities $NEAR powers the Agentic Web infrastructure: ➛ Transaction Fees - Pay for all on-chain operations at sub-$0.01 costs. 30% of fees burned, 70% to smart contract developers. ➛ Staking & Validation - Stake NEAR to secure the PoS network. Validators earn rewards from the 2.5% annual inflation (halved from 5% in Oct 2025). ➛ Storage Staking - Lock NEAR to reserve on-chain storage for smart contracts and data. Refundable when storage is released. ➛ Governance - NEAR holders participate in protocol governance through the NEAR Foundation and community proposals. ➛ Agent Economy - AI agents use NEAR for cross-chain transactions, micropayments, and autonomous operations on behalf of users. ➛ Developer Incentives - 70% of transaction fees flow to smart contract developers, creating a sustainable builder economy. ⚪ Key Features ➛ Co-Founded by Transformer Paper Co-Author - Illia Polosukhin co-wrote "Attention Is All You Need," the paper that created ChatGPT and modern AI. The most credentialed AI founder in all of blockchain. ➛ 1M TPS (Nightshade 2.0) - Publicly verified benchmark. Sub-$0.01 fees. 1.2-second finality. Dynamic resharding (June 2026) scales automatically without human intervention. ➛ Chain Abstraction Across 35 Chains - Near com super-app with email/FaceID login, feeless cross-chain swaps, AI assistance, and agent automation. No seed phrases, no bridging, no chain selection. ➛ User-Owned AI - IronClaw confidential computing, Venice AI private inference, NEAR AI Agent Market. Your AI, your data, your privacy. ➛ Inflation Halved to 2.5% - Cut from 5% in October 2025. 30% of fees burned. Approaching deflationary territory as network usage scales. ➛ Post-Quantum Security (FIPS-204) - Quantum-safe key rotation live on mainnet May 2026. One of the first chains with post-quantum protections in production. ➛ #1 Developer Activity - 950 GitHub commits in H1 2025, highest of any major blockchain. Written in Rust. Apache 2.0 / MIT licensed. ➛ Institutional Investment Vehicles - Grayscale Near Trust (GSNR) and Bitwise NEAR Staking ETP provide regulated exposure. 🔵 Meet the NEAR Team NEAR is led by the most AI-credentialed founding team in all of blockchain. The co-founder literally co-authored the research paper that made ChatGPT possible. ▶️ Core Members: ➛ Illia Polosukhin [ @ilblackdragon ] - Co-Founder & CEO, NEAR Foundation | Former engineering manager at Google Research, where he co-authored the seminal 2017 paper "Attention Is All You Need," which introduced the transformer architecture that powers GPT, BERT, ChatGPT, and the entire modern AI revolution. Core developer of TensorFlow. Master's in Applied Math and Computer Sciences from Kharkiv Polytechnic Institute. The most credentialed AI founder in blockchain. ➛ Alexander Skidanov [ @AlexSkidanov ] - Co-Founder | Former software engineer at Microsoft and Director of Engineering at MemSQL. ICPC gold medalist (2008). Held a position at OpenAI in the period leading up to ChatGPT's release. Co-designed NEAR's Nightshade sharding architecture. ➛ Eric Winer [ @eric_winer ] - CTO & Managing Director | Oversees technical architecture, protocol development, and engineering execution across the NEAR ecosystem. ➛ Alex Shevchenko [ @AlexAuroraDev ] - Product Manager | Drives product development and roadmap execution across NEAR's core platform. ➛ George Xian Zeng - Chief Product Officer | Leads product strategy across NEAR's chain abstraction, AI infrastructure, and super-app initiatives. ➛ Nicci Arsenault - Chief People Officer | Manages organizational development and team growth across NEAR's global operations. ▶️ Advisers: ➛ James Zhang - Adviser | Strategic adviser to the NEAR ecosystem. ➛ Jocy Lin - Adviser | Strategic adviser to the NEAR ecosystem. ➛ NEAR Foundation (Switzerland) - Non-profit organization overseeing ecosystem development, grants, governance, and institutional partnerships. $800M ecosystem fund deployed. 🔵 Ratings ➛ Use Case: ★★★★★ (5/5) - NEAR occupies a genuinely unique position: the only blockchain co-founded by someone who literally co-invented the AI architecture powering the entire modern AI revolution. That's not a marketing claim. "Attention Is All You Need" is the most cited AI paper in history, and Polosukhin is a co-author. The chain abstraction across 35 chains with email/FaceID login solves crypto's biggest UX problem. 1M TPS with dynamic resharding handles enterprise-scale throughput. User-Owned AI with confidential computing addresses the privacy crisis in centralized AI. Post-quantum security future-proofs the infrastructure. The Agentic Web thesis (autonomous AI agents transacting on-chain) is the most forward-looking vision in blockchain. No other chain has this combination of AI pedigree, UX innovation, and infrastructure depth. ➛ Tokenomics: ★★★★ (4/5) - Inflation halved from 5% to 2.5% in October 2025, cutting annual issuance from ~64M to ~32M NEAR. 30% of transaction fees are burned, moving the network toward deflationary territory as usage scales. 96% of supply is already circulating, meaning minimal future dilution. Developer incentives (70% of fees) create a sustainable builder economy. Grayscale Trust and Bitwise ETP provide institutional demand channels. The 1-point deduction is that NEAR is down 89% from ATH ($20 to $2.24), the 2.5% inflation is still net inflationary at current usage levels, and the $800M ecosystem fund creates potential sell pressure if liquidated. ➛ Audits: ★★★★✦ (4.5/5) - NEAR holds a 4.5 star rating on CertiK, one of the highest scores among Layer-1 blockchains. The protocol is one of the most actively developed blockchains in the world, leading all major chains with 950 GitHub commits in H1 2025. The codebase is fully open-source (Rust, Apache 2.0/MIT). Nightshade sharding has been battle-tested since 2020 without a major consensus failure. FIPS-204 post-quantum cryptography adds forward-looking security that few chains can match. The protocol survived the April 2026 Litecoin zero-day exposure ($600K via Intents) with the team pledging to cover losses, demonstrating responsible incident response. The 0.5 deduction is that chain abstraction across 35 chains inherits the failure surface of all connected networks (as the Litecoin incident proved), and the expanding AI infrastructure (IronClaw, Agent Market) introduces novel attack surfaces that require ongoing security hardening. ➛ Community: ★★★★✦ (4.5/5) - NEAR has one of the most technically sophisticated and globally distributed communities in crypto. NEARCON events attract thousands of developers and institutional participants (NEARCON 2023: 3,000-5,000 attendees). The $800M ecosystem fund supports a thriving builder ecosystem. Ref Finance, Mintbase, and dozens of dApps create real on-chain activity. The community spans AI researchers, blockchain developers, and institutional investors. Grayscale Trust and Bitwise ETP bring regulated investor participation. Illia Polosukhin's personal credibility as a transformer co-author brings attention from both the AI and crypto worlds. The 0.5 deduction is that NEAR's community, while technically strong, doesn't generate the viral retail energy of chains like Solana or TON. Mainstream awareness remains below what the technology deserves. The 89% price decline from ATH has tested community conviction. 🔵 Conclusion NEAR Protocol is the most underappreciated infrastructure play in the AI x crypto intersection. The co-founder literally co-authored "Attention Is All You Need," the most important AI paper in history, the paper that made GPT, BERT, ChatGPT, and the entire modern AI revolution possible. And his co-founder worked at OpenAI. They didn't pivot to AI. They started in AI, built a blockchain because no existing one could handle what they needed, and are now bringing the two worlds together in a way nobody else can credibly attempt. The technology speaks for itself: 1M TPS with dynamic resharding, chain abstraction across 35 chains with email login, confidential computing for user-owned AI, post-quantum security, and the highest developer activity in the industry. The Near com super-app makes blockchain invisible to users. IronClaw makes AI private. NEAR Intents make cross-chain execution seamless. Here's the fundamental question: if AI is the most important technology of the 21st century, who should build the infrastructure it runs on? OpenAI and Google, who log your data and censor your outputs? Or a decentralized protocol co-founded by the man who co-invented the transformer, with confidential computing, 1M TPS, and post-quantum security? NEAR isn't betting on the AI narrative. NEAR's co-founder helped create AI. And now he's building the decentralized infrastructure to make sure it stays open, private, and user-owned. That's not a pivot. That's a homecoming.

looks like the market is finally getting what $NEAR has been building, for those of you who haven't read my article here is what is important to understand about NEAR: agents finally work — opus 4.6 runs 100 tool calls, closed 13 issues across a 50 person org in a single day, autonomous agent traffic grew 7,851% YoY, 2.3% is already hitting checkout pages, the customer agent infra has been built for in advance has finally shown up agents legally cannot use visa/stripe/ach — every KYC framework on earth defines a customer as a natural or legal person, software has no passport, banks cannot onboard it, this is not a policy problem it's a definitional one which is why stripe (the most sophisticated payments co alive) didn't just bolt agent support onto its own rails — instead they co-shipped ACP with openai, shipped USDC on base via x402, AND incubated their own L1 tempo with paradigm, three simultaneous bets telling you the answer is crypto and the answer is plural every serious team is now launching their own chain (hyperliquid, unichain, monad, megaeth, tempo, base, robinhood), the BIS documented in 2022 that fragmentation is endogenous to blockchain economics, this isn't a phase it's permanent and getting worse humans tolerate bridging pain for a hot mint, agents cannot — one user request might touch 6 chains in 8 seconds, no agent is sitting there collecting gas tokens to pay for the next hop near.com already solves this for humans today — one account, one identity, transact natively across any chain without bridging or wrapping — a nice UX for a human, a mandatory UX for an agent the reframe: near is not a destination chain competing for app deployment, it's a coordination layer sitting above every chain, and the metrics that matter are completely different intents has routed $15B across 31 chains, 97% of that volume moves between OTHER chains, near's own L1 is just 3% of throughput, 95% of protocol fees now come from intents this is why TVL and DEX volume are the wrong scoreboard — the market has been pricing near like a dead L1 that lost the solana race, when it's actually the only chain whose volume measures activity between every other chain fee revenue ($30M/yr) is also closing in on post-halving emissions ($40M/yr), token is on the cusp of net deflationary while still early on the adoption curve and the team — illia polosukhin co-authored "attention is all you need," the paper that created modern AI, he started building near in 2018 because he saw the agent economy coming earlier than almost anyone alive, alex skidanov spent a decade at memsql building sharded databases before co-founding it, that combination of AI-native founder vision distributed systems pedigree does not exist anywhere else in crypto everything converged in one twelve month window — anthropic shipped the first deployable agent, stripe ratified crypto rails, every defi team launched their own chain making coordination more valuable not less, near shipped nightshade 3.0 confidential intents ironclaw near.com, tokenomics just inflected post-halving this isn't a roadmap moment it's a now moment, near has been building this tech for YEARS and the thesis is finally playing out -- in this agentic future the coordination layer is structurally the largest take-rate position in the entire stack ($NEAR), since everything flows through the point of lowest friction

people remember $NEAR's founder Illia Polosukhin co-authored "attention is all you need." very few know about the second co-founder. Alex Skidanov was an early engineer at MemSQL - now SingleStore, a multi-billion dollar database infrastructure company - and contributed to OpenAI. so NEAR's team is: one person who co-invented the architecture behind modern AI, and one person who built distributed database infrastructure used by enterprises worldwide. both have been heads-down on the same chain since 2018. then look at the technical resumes of the projects in the "AI x crypto" sector that are hyped higher. market's bad at this one.

🎙️ New #TalkingPostgres podcast Ep39 is out! @a_prout, distinguished engineer at Microsoft, talked about his engineering journey from MemSQL to HorizonDB, shared-storage, & why good systems programmers are paranoid 🎧 talkingpostgres.com/episodes… 📺 youtu.be/L_2kyfL9LN0?si=6o5o…

Looking forward to talking to database architect Adam Prout @a_prout today/Wed 6 May at 10am PDT on the #TalkingPostgres podcast! Ep39 topic: From MemSQL to HorizonDB, an engineer's journey Where? Live on the Microsoft Open Source Discord. Join us: aka.ms/talkingpostgres-ep39-…

LSMs typically have a WAL log to make the in-memory layer durable don't they? It's a simpler log vs a B-Tree, but still around. I haven't done a broad study of LSMs, but MemSQL/Singlestore did things this way and so does RocksDB

Back before github, cloud, and containers were a thing, at memsql, we built our own git & CI infrastructure for a 100 person team that ran on a couple bare metal machines. Builds (super complex C codebase) finished in 5 minutes and CI ran in 20 minutes end-to-end. Why? Files cached locally, no startup time, and incremental builds. Today, with the modern kluge of actions, VMs, sandboxes, snapshots, and caches it feels impossible to achieve this type of performance. I wonder if the constant onslaught of infrastructure challenges and increased performance demanded by coding agents will teleport us back to first principles in developing CI infrastructure. I hope it does.

My 2012 memsql benchmark post hit the mark well enough, someone hired blackhat SEOs to tank it in search results. Kinda remember it each time I see Neon&friends. dom.as/2012/06/26/memsql-rag…

> be Illia Polosukhin (@ilblackdragon) > born in Ukraine during the 90s, right after the Soviet collapse > grew up with hyperinflation and food shortages > watched The Matrix at 10, got obsessed with AI > started building games that same year > studied math and computer science in Kharkiv > became an ICPC programming champion > moved to California around 2008 > worked at Salford Systems and MemSQL > 2014: joins Google Brain as an AI researcher > works on TensorFlow, machine translation, search > 2017: co-authors "Attention Is All You Need" > the Transformer paper that powers every AI model today > ChatGPT, Gemini, Claude all built on it > leaves Google later that year > co-founds @NEARProtocol to teach machines to code > 2018: hits a problem > needs to pay students in China and Eastern Europe > banking is slow, Ethereum fees are high > realizes blockchain can fix payments > pivots from AI company to blockchain protocol > NEAR Protocol is born > mainnet launches late 2020 > 2024: shares stage with Jensen Huang at NVIDIA GTC > named CoinDesk's Most Influential > faced governance debates over emission cuts > centralization questions around validator power > today, he’s building at the intersection of AI and crypto > pushing decentralized AI agents and user-owned models Full podcast with Illia is live. We covered AI monopolies, privacy risks, stablecoins, and the future of AI crypto. 🎧 Podcast link below, watch now 👇

NEAR - and here's why the team credibility is off the charts: @ilblackdragon literally co-authored "Attention Is All You Need" - the 2017 paper that invented transformers. 173k citations. Everything from ChatGPT to Midjourney traces back to that work. He's not riding the AI wave, he helped create it. @AlexSkidanov ICPC gold medalist, Microsoft → MemSQL founding engineer building core database infrastructure. But credentials aside, what I trust most: they pivoted from pure AI to blockchain because they personally hit payment friction paying global contributors. Real problem → real solution. Now in 2026: 1M TPS tested, 46M monthly users, Grayscale filing for spot ETF, and their AI Cloud actually ships privacy-preserving infra instead of just talking about it. When someone asks "trust?" - I look for builders who could've stayed comfortable at Google/OpenAI but chose to build something new. That's NEAR.

memcachedで思い出したけど、redis/valkeyは元気にやってる一方、当時覇権を争ってたmemsqlとかaerospikeはどうなってるんだろうって思って調べてみたら、どっちもまだ元気にやってた。memsqlはsinglestoreに名前変わってた。

Day 3- History NEAR started in 2017 as NEAR.ai - a pure AI startup The founders were Illia Polosukhin and Alex Skidanov. Illia was a Engineering Manager at Google Research. Alex was a Director of Engineering at MemSQL. Both were super skilled in software and AI!

🚀 NEAR’s @NEARProtocol 5-Year Story: From Vision to Vanguard (2020–2025) 🏁 2018–2019: The Idea & the Build! Before there was a blockchain, there was a dream. Founders: Illia Polosukhin (ex-Google, worked on TensorFlow) and Alexander Skidanov (ex-MemSQL engineer). 👇 #NEAR

This is Ankur Goyal. He built databases at MemSQL (now SingleStore), launched Impira to tackle AI-powered document extraction, and led Figma’s AI team after they acquired his startup. Now he’s the founder & CEO of Braintrust, where he’s making AI reliable in production. His core insight? The hardest part of AI isn’t the demo. It’s making it work every single day. Ankur learned that the moment you improve one use case, another can break. He watched enterprise customers get excited about faster invoice parsing—only to revolt when bank statements regressed. The solution? Evals. Tight feedback loops that turn user complaints into measurable tests, often in just two clicks. For Ankur, shipping AI isn’t about hype. It’s about boring but essential systems: observability, logging, model selection, and ruthless prioritization. He’s seen that most teams drown in evals, dashboards, and vanity metrics. The best? They delete aggressively, focus on the 5–10 examples that matter, and re-evaluate models monthly. His advice to founders: Don’t confuse prototypes with products. Build feedback loops early, price in reliability, and treat LLMs like databases, not CPUs. Choose one, go deep, and adjust as the landscape shifts. The future of AI belongs to builders who can bridge the gap from prototype to production and deliver business value, not just flashy demos. Watch our full interview with Ankur. Link in next post.

Congrats team @SingleStoreDB Thats two @nikitabase founded database companies acquired in the past ~3 months for XXX millions each (Neon -> Databricks, MemSQL/Singlestore -> private equity)? How often has that happened before?

.@NEARProtocol was built by industry veterans. 🔹 Erik Trautman – Wall Street entrepreneur, Viking Education founder 🔸 Illia Polosukhin – Ex-Google AI engineer 🔹 Alexander Skidanov – Ex-Microsoft, memSQL Director of Engineering NEAR’s ecosystem is expanding fast—hiring across engineering, product & operations. The Open Web is being built.

This is ..interesting .. @SingleStoreDB is being acquired. I've covered this database vendor since it was created back in the MemSQL days. It was an early pioneer in the converged OLTP/OLAP (which today is honestly table stakes). businesswire.com/news/home/2…