Interesting

If you liked this piece, you should subscribe to my premium newsletter. It’s $70 a year, or $7 a month, and in return you get a weekly newsletter that’s usually anywhere from 5,000 to 18,000 words, including vast, detailed analyses of NVIDIA, Anthropic and OpenAI’s finances, and the AI bubble writ large (updated to version 3.0 last week). My Hater's Guides To the SaaSpocalypse, Private Credit and Private Equity are essential to understanding our current financial system, and my guide to how Open

Our progress toward recursive self-improvement, and its implications.

Industrial silicon based quantum computers. Scalable, high-performance, affordable quantum computers built for existing infrastuctures.

I'm a software engineer, completing 10 years of professional experience this year. I started my career as a web frontend engineer (it was easier for me to de...

Focused Energy is the leading Laser Fusion company building on NIF’s foundation & pursuing the most promising path to clean, limitless fusion power

Join over 20,000 brands using podcast advertising to drive revenue.

Transformer-based large language models are increasingly used for long-horizon tasks; however, their attention mechanism scales poorly with context length. To handle this, we study a sleep-like consolidation mechanism in which a model periodically converts recent context into persistent fast weights before clearing its key-value cache. During sleep, the model performs $N$ offline recurrent passes over the accumulated context and updates the fast weights in its state-space model (SSM) blocks through a learned local rule. During inference, this shifts extra computation to sleep while preserving the latency of wake-time prediction. We test our method on controlled synthetic tasks, including cellular automata and multi-hop graph retrieval, as well as a realistic math reasoning task, on which a regular transformer as well as SSM-attention hybrid models fail. We then show that increasing sleep duration $N$ for our models improves performance, with the largest gains on examples that require deeper reasoning.

An OpenAI model solved the 80-year-old unit distance problem, disproving a major conjecture in discrete geometry and marking a milestone in AI-driven mathematics.

Why frontier AI has broken the open CTF format, hollowed out the scoreboard, and made competitive CTF performance a weaker signal than it used to be.

Why senior developers talk in terms of complexity while the rest of the business is worried about uncertainty — and what to do about it now that AI is in the picture.

I recently talked with Joseph Ruscio about AI coding tools for Heavybit’s High Leverage podcast: Ep. #9, The AI Coding Paradigm Shift with Simon Willison. Here are some of my …

What we learned from surveying the top Tiny Teams at the World's Fair

Bitcoin is rallying alongside rising inflation signals, defying the traditional macro playbook.

The tie-up between the tokenization specialist and transfer agent giant lets public companies issue blockchain-based shares without changing market structure.

If productivity gains from AI are “only” as good as computers and the internet, investors will still need to contend with major macro imbalances and could consider assets that provide diversification and scarcity.

The economics of structural change and the post-commodity future of work

A further look into the Autonomous Vehicles section of ARK's Big Ideas 2026

Crypto protocols have been structurally illegible for a decade, and the window to fix that is open right now. The protocols that restructure will attract the capital, and the ones that don't will keep trading at a discount to their own fundamentals.

We gave Codex a foothold. It popped a root shell.

Shopify's data reveals three trendlines pointing to a growing opportunity in the age of AI.

Real-time executive leadership change tracking from SEC filings and press releases.

Paradigm is a frontier technology investment firm that builds and invests in crypto, AI, robotics, and across new frontiers from the earliest stages.

A discussion of the implications of two papers on quantum resource estimates for an ECC-256 break

Interpretability research from Anthropic on emotion concepts

Streamline your fundraising efforts with essential templates and tools designed to help startups manage capital confidently and focus on growth.