The Information Machine
The edition

Monday October 12, 2026

New today

01
New

AI firms and the Genesis Mission Consortium

  • At the White House 'Science: A New Golden Age' Summit on October 9, Director Michael Kratsios announced $2.4 billion in commitments from eleven technology companies to the Genesis Mission Consortium, a Department of Energy initiative Trump launched in November 2025.
  • Nvidia leads at $1 billion over five years, AMD at $500 million, OpenAI at $200 million, and Anthropic and Google at $150 million each.
  • The federal government separately pledged $5 billion for science and technology challenges.
The gist

The commitments direct significant private AI compute capacity toward federal scientific research across more than 15 agencies. Jensen Huang said the government is "quickly becoming one of the BIGGEST new buyers of AI compute," reflecting a shift in how federal science programs access frontier computing resources.

Updates

02
Day 6

AI infrastructure capex and the revenue gap

  • Broadcom, SpaceX, and Oracle turned to Wall Street debt on October 12 for a combined $90 billion or more in AI hardware financing, with Broadcom seeking over $50 billion for custom chips for OpenAI, SpaceX arranging $40 billion for Nvidia chips, and Oracle in talks with Apollo and Goldman Sachs to bridge chip payments and cloud revenue.
  • One source described chip costs as too large to absorb on balance sheets.
  • Aravind Srinivas said some Meta engineers spend around $10 million a year on AI coding tools and certain Perplexity Computer users spend upwards of $10,000 a month running agent loops.
The gist

Capital spending commitments have grown far ahead of identified revenue, and debt markets are now being enlisted to bridge the gap between chip purchases and future cloud income. Bain's analysis suggests the revenue shortfall is so large that filling it would require economic growth at a scale beyond existing consumer and enterprise markets.

03
Day 6

AI proofs of Navier-Stokes and Yang-Mills

  • At a Caltech 'Math 2.0' lecture on October 11, Terence Tao argued that blind AI optimization of problem-solving 'is now actively harmful to the long-term health of mathematics,' sharpening his stance on OpenAI's October 6 release of 722 AI-generated manuscripts covering 372 open problems.
  • Northwestern professor Antonio Auffinger reported that some students are now questioning whether to remain in mathematics at all, saying his priority has shifted to retaining talented people in the field.
  • Cryptographer Matthew Green placed a 15% probability on AI causing a functional loss of confidence in existing public-key encryption algorithms.
The gist

The release by OpenAI has prompted institutional pushback, career uncertainty among students, and a formal boycott call from mathematicians, raising concrete questions about the effect of mass AI-generated research on the mathematical profession. Separately, concerns about cryptographic security have grown because none of the 719 abstracts mention cryptography, and at least one cryptographer has stated a non-trivial probability that public-key encryption algorithms could be functionally compromised.

04
Day 11

OpenAI Pro plan pricing overhaul

  • Analysis published October 11 citing a16z data argues that flat-rate and per-seat subscriptions are becoming structurally unsustainable because AI agents consume nearly 5x the tokens humans do, a ratio that grew 14x in six months.
  • Agents working around the clock remove the natural usage cap that human prompting speed provided, pointing vendors toward consumption-based metering.
  • The data frames OpenAI's September 29 decision to halve Pro plan usage and add a $500 tier as part of a broader breakdown in subscription economics.
The gist

The pricing restructuring directly reduces what subscribers get for $200 per month, while agent-driven token consumption is growing at rates that strain flat-rate subscription models. A SemiAnalysis comparison cited by The Neuron estimates Claude Max at $200 per month provides over 5x the API-equivalent usage value of ChatGPT Pro at the same price for heavy agent users.

05
Day 5

OpenAI GPT-6, Intelligent UI, and Astra Ultrafast

  • On October 11, a developer published reverse-engineering findings showing Intelligent UI, ChatGPT's feature for generating interactive components in responses, works through a custom intermediate language called DIL: the AI emits DIL, the server compiles it, and the client renders from roughly 70 native components without additional inference calls.
  • OpenUI followed with a technical breakdown confirming DIL combines Markdown, JSX-like tags, and JavaScript, compiling to a virtual DOM-style representation, and Tech Republic reporting added that GPT-6 can begin answering while still reasoning.
The gist

The rollout gives ChatGPT's roughly 1.2 billion weekly users interactive, task-specific interfaces without leaving the chat window, which changes how some users interact with calculators, comparison tools, and structured information that previously required external sites. The technical architecture, using a compiled intermediate language rather than live code generation, is a distinct engineering approach from competing AI systems.

What is moving now · Every edition · Every story

The daily email

Want this in your inbox?

I send one email each morning with the stories that moved. If you would rather just read here, that works too.

Subscribe free