News
Curated articles, my blog posts, and site announcements.
Curated articles, my blog posts, and site announcements.
Curated articles, my blog posts, and site announcements.
Robinhood's chain uses $ETH as its native gas token, meaning every transaction requires holding and spending ETH. This cements ETH as the native onchain asset across the programmable economy and brings 27M users onchain.
powdr has released an early version of yulc, an experimental formally verified Yul-to-EVM compiler written in Lean. If it compiles a program, the EVM bytecode provably computes what the Yul semantics say. The main correctness theorem is already machine-checked for a subset of Yul.
Georgios flags Incur as a gamechanger for building agentic CLIs in TypeScript.
OpenAI's GPT-5.6 family (Sol, Terra, Luna) is starting to roll out in ChatGPT, Codex, and the API.
OpenAI has launched ChatGPT Work, a new agent in ChatGPT powered by Codex and GPT-5.6. It can take action across apps and files, stay on a project for hours, and turn a goal into finished work.
TypeScript 7 is officially released.
Jarred Sumner announces that Bun is being rewritten in Rust.
CL clients are now shipping natively. Barnabé wants to bring the design to the Gnosis-Ethereum bridge and sees @ethlabs_org playing a role in bringing it to the larger Ethereum ecosystem.
Vitalik outlines the Lean Ethereum roadmap after the Berlin researcher meeting: recursive STARKs as enshrined verification, quantum-safe replacements, decoupled availability and finality, multidimensional gas, new state types (keyed nonces, ring buffers, UTXOs), and RISC-V or leanISA as a first-class VM alongside EVM. strawmap.org has the updated diagram.
Austin Griffith recaps a slop.computer podcast episode with dcbuilder: software is easy to make but do not rebuild good open source, and it is time to upgrade from markdown files to ontologies for representing data.

Lighter has executed the burn of 15,638,702 LIT, permanently removing the tokens from circulation.

Lighter has dropped end-to-end p99 latency from 280 ms to 55 ms, tx processing p99 to under 1 ms, and hot path apply time to 100 to 250 µs. On June 5 the exchange hit 811M transactions in a day, averaging 9,388 TPS with 20,740 TPS peaks.

Prime Intellect has raised a $130M Series A led by Radical Ventures, with NVIDIA, Intel Capital, Dell Capital, and existing investors, to build the Open Superintelligence Stack for training, deploying, and continuously improving custom models.
World Chain v2.4.0 ships Full Block Access Lists, an out-of-protocol sidecar to Flashblocks that enables parallel pre-processing of full blocks (shard by shard) before consensus validation. Lower validator hardware requirements and stronger decentralization.
A non-technical primer from the EF Global Policy Strategy team, aimed at policymakers and institutions. Covers how Ethereum works, how it is governed, and how it compares to other L1s on resilience, economic security, client diversity, counterparty risk, and ecosystem.
Venice announces its first outside capital raise, a $65m Series A led by Dragonfly at a $1B equity valuation. The company reached profitability in Q1 and will use the round to vertically integrate compute via its first datacenter build. Venice remains the largest holder of its VVV token.
New Bitcoin Policy Institute research identifies the Party for Socialism and Liberation, tied to Shanghai-based Neville Singham, as a critical mobilizer in 21 anti-data-center campaigns across 14 states. The efforts have blocked or delayed $23.6 billion in US AI investment, including 10 data center moratoria, one permanent ban, and four scrapped AI projects.
In a first-of-its-kind clinical trial procedure at UHN Toronto Western Hospital, Neuralink threaded its electrodes straight through the dura into the cortex without cutting the membrane. The participant was controlling a cursor with their thoughts within an hour of surgery.
Ethereum Institutional launches as an independent non-profit dedicated to accelerating institutional adoption of Ethereum, its L2s, and the wider ecosystem. Anchor funders: BitMine, SharpLink, and Joseph Lubin. Founding team previously built the Ethereum Foundation Enterprise function.
Palantir lays out nine principles for AI sovereignty: data retention, controlling your own weights, rejecting tokenmaxxing, and picking real expertise over politics.
OpenSoftware launches June, a private AI for Mac bundling an agent plus chat, voice dictation into any app, automated meeting notes, and private OSS plus anonymized frontier models. Powered by Venice models, built on the OSS Hermes Agent, MIT-licensed.
Anthropic introduces Claude Science, a new app designed for every stage of research. Artifacts traced to their code, environments managed on demand, and 60+ optional scientific databases. Available in beta at claude.com/science.
Open Standard launches Open USD, a stablecoin designed by 140+ businesses growing the internet economy. Launch partners include Visa, Aave, Plasma, Tempo, ether.fi, Polygon, Anchorage, MoneyGram, OnePay, Spark.
Etched announces a successful A0 tapeout, $1B+ in customer contracts, $800m raised, and SOTA throughput/latency/power efficiency on inference. Introducing Low-Voltage Inference and Cluster-Scale Memory. 400+ engineer team, Taiwan factory, first racks shipping this summer.
Claude Sonnet 5 makes plans, uses tools like browsers and terminals, and runs autonomously at a level that just a few months ago required larger and more expensive models. Performance is close to Opus 4.8 at lower prices, default on Free and Pro today.

Since TGE the Lighter protocol has programmatically bought back ~15.5M LIT (~6.3% of circulating supply) using exchange revenues. Going forward, buybacks will permanently reduce supply through burns.
OpenAI Developers posted a 9-second teaser for a hardware peripheral upgrading the Codex shortcuts experience, dropping July 15th. Replies are calling it an Elgato Stream Deck for AI.
Ethlabs wraps its first week with a quick FAQ thread covering why they launched now, who is on the team, what they will work on, their stance on ETH, and their relationship with the Ethereum Foundation. More on priorities, workstreams, boundaries, and accountability mechanisms is coming.
Coinbase cut AI spend nearly in half while token usage continued to grow, via cheaper open-weight defaults (GLM 5.2, Kimi 2.7), preprocessing-aware routing, cache-aware requests (5% to 60% hit rate in LibreChat), lean context, and full usage visibility.
Vitalik Buterin published a survey of the indistinguishability obfuscation tech tree, mapping STARKs, random mixing assumptions, and the main iO protocol lineage.
News tools
One compact deck for archive access, subscription controls, and a tighter recommendation list.