LIVE · TUE, JUL 14, 2026 --:--:-- ET
Issue Nº 84 COST TOTAL $14814.42 ARTICLES TODAY 7 TOKENS TOTAL 9.50B
aiexpert
Running the wire
Market Korean leveraged chip ETF crashes 45% in three months, retail investors face steep losses Breaking OpenAI Codex hits 7M weekly active users; 10x growth in 6 months outpaces Claude Code reporting Research Simon Willison shares a cache-friendly uvx recipe for GitHub Actions CI Breaking Cloudflare launches Precursor: session-based bot detection that watches behavioral signals, not just login clicks Breaking Satya Nadella: enterprises pay for AI twice—in cash, then in proprietary knowledge leaked to model makers Funding Prefect acquires Dagster, uniting two leading orchestrators as agentic workflows reshape automation Chips Rochester Electronics and Qorvo expand RF component supply for aerospace, defense Breaking Nscale's £2bn UK AI data centre delayed by grid bottleneck; eyes fuel cells Funding Anthropic recruits Monzo founder Tom Blomfield to lead compute team Funding Prefect acquires Dagster; rivals consolidate into one orchestration platform Chips Tesla AI5 tapes out at Samsung Foundry on 2nm; production starts at Taylor fab in 2026 Policy 69% of US workers back sovereign wealth fund requiring AI firms to transfer 50% stake to public Market AI trio's $4.4T grip frays: funds diversify out of Taiwan/Korea mega-cap tech in emerging markets Market AI chip executives tell CNBC demand remains 'almost unlimited'; enterprises pivot to 'valuemaxxing' from 'tokenmaxxing' Funding Probook raises $40M Series A from a16z & Sequoia; dispatch-first AI for $700B home services market Funding Radical Numerics emerges with $50M seed for multimodal biological AI; previews Omnii genome model Funding Keyfactor secures $1B+ from Summit Partners for machine identity and post-quantum security platform Market Korean leveraged chip ETF crashes 45% in three months, retail investors face steep losses Breaking OpenAI Codex hits 7M weekly active users; 10x growth in 6 months outpaces Claude Code reporting Research Simon Willison shares a cache-friendly uvx recipe for GitHub Actions CI Breaking Cloudflare launches Precursor: session-based bot detection that watches behavioral signals, not just login clicks Breaking Satya Nadella: enterprises pay for AI twice—in cash, then in proprietary knowledge leaked to model makers Funding Prefect acquires Dagster, uniting two leading orchestrators as agentic workflows reshape automation Chips Rochester Electronics and Qorvo expand RF component supply for aerospace, defense Breaking Nscale's £2bn UK AI data centre delayed by grid bottleneck; eyes fuel cells Funding Anthropic recruits Monzo founder Tom Blomfield to lead compute team Funding Prefect acquires Dagster; rivals consolidate into one orchestration platform Chips Tesla AI5 tapes out at Samsung Foundry on 2nm; production starts at Taylor fab in 2026 Policy 69% of US workers back sovereign wealth fund requiring AI firms to transfer 50% stake to public Market AI trio's $4.4T grip frays: funds diversify out of Taiwan/Korea mega-cap tech in emerging markets Market AI chip executives tell CNBC demand remains 'almost unlimited'; enterprises pivot to 'valuemaxxing' from 'tokenmaxxing' Funding Probook raises $40M Series A from a16z & Sequoia; dispatch-first AI for $700B home services market Funding Radical Numerics emerges with $50M seed for multimodal biological AI; previews Omnii genome model Funding Keyfactor secures $1B+ from Summit Partners for machine identity and post-quantum security platform
Research

Simon Willison shares a cache-friendly uvx recipe for GitHub Actions CI

Developer Simon Willison published a concise recipe for running uvx tools inside GitHub Actions without hitting PyPI on every build. The trick centres on UV_EXCLUDE_NEWER, a uv environment variable that pins package resolution to a specific date. That same date string doubles as the GitHub Actions cache key, so developers only need to maintain one value to control both the installed versions and cache invalidation — no separate requirements file required.

The workflow pairs astral-sh/setup-uv with three non-default settings: cache-dependency-glob is blanked so uv stops looking for pyproject.toml, cache-suffix is set to the UV_EXCLUDE_NEWER date, and prune-cache is set to false (overriding Astral's default of deliberately discarding cached wheels). A guard step then sets UV_OFFLINE=1 on cache hits, so any tool missing from the cache causes an explicit failure rather than a silent network fetch. Willison notes this deliberately trades a small speed penalty against eliminating per-run PyPI CDN calls — a tradeoff many ML teams running llm, datasette, or similar tooling in CI will recognise immediately.

Sources