THE SWELL #24
SAIL Weekly Digest June 27–July 3, 2026 | Issue #24
Washington’s intervention at the frontier has reached a new level. Regulators just abruptly walked back their sudden restrictions on Anthropic’s models, yet they keep the brakes on OpenAI’s rollout. Meanwhile, the White House starts floating ideas about a public wealth fund and taking equity in these labs.
As the feds try to micromanage the top tier, the open ecosystem underneath is steadily widening. Our SAIL authors mapped out exactly where the real momentum is shifting: a complex web of de facto licensing, the massive energy and silicon bottlenecks holding back closed compute, and a quiet wave of open models that are steadily replacing paid, closed subscriptions.
The Week in AI
OpenAI proposed handing the US government a 5% stake — roughly $42.6B at its March valuation — with Altman pitching that every major American lab pay a matching slice into an Alaska-style public fund.
Anthropic’s Claude Fable 5 and Mythos 5 came back online July 1 after nearly a month disabled worldwide under the first US export controls ever applied to an AI model rather than to chips.
OpenAI delayed the full public launch of GPT-5.6 at the government’s request, with Commerce Secretary Lutnick reportedly warning Altman against releasing without prior approval.
This Week from SAIL Authors
The Governed Frontier
The US now has a de facto model licensing system — Timothy B. Lee argues that with GPT-5.6 waiting on government approval, frontier releases now effectively require Washington’s sign-off. — Understanding AI
📈 Data to start your week — Azeem Azhar and Nathan Warren chart China’s young AI founders, public confidence in AI, and the fastest-ever paths to $1 million. — Exponential View
Power and Silicon
🔮 Fifty years of Moore’s Law wasn’t fast enough for AI #580 — Azeem Azhar and Marija Gavrilov on why AI’s appetite has outrun five decades of silicon scaling, plus the agentic frontier and China’s AI job market. — Exponential View
Taiwan’s War on Renewables — Lily Ottinger and Aqib Zakaria unpack how the island at the center of the chip supply chain dismantled its own energy security — 97% imported, 11 days of LNG reserves. — ChinaTalk
Open Models, Local Agents
Latest open artifacts (#22): Zyphra, Cohere, and Poolside are expanding the breadth of the ecosystem — Florian Brand and Nathan Lambert assess the open ecosystem and the motivations behind releasing models. — Interconnects
Using Local Coding Agents — Sebastian Raschka on running open-weight models in local coding harnesses as an alternative to Claude Code and Codex subscriptions. — Ahead of AI
Full Library
Taiwan’s War on Renewables — Lily Ottinger & Aqib Zakaria | ChinaTalk
The US now has a de facto model licensing system — Timothy B. Lee | Understanding AI
📈 Data to start your week — Azeem Azhar | Exponential View
🔮 Fifty years of Moore’s Law wasn’t fast enough for AI #580 — Azeem Azhar | Exponential View
Latest open artifacts (#22): Zyphra, Cohere, and Poolside are expanding the breadth of the ecosystem — Florian Brand & Nathan Lambert | Interconnects
Using Local Coding Agents — Sebastian Raschka | Ahead of AI
