Morning Edition

The Futurist

Sunday, October 25, 2026
AI & Technology Markets & Crypto Ideas Worth Keeping
S&P 500$7,623.68▼ 0.52%
Dow Jones$52,215.16▼ 1.22%
Nasdaq$25,955.34▼ 0.24%
VIX$19.45▲ 2.15%
BITCOIN$80,429.00▼ 1.39%
RIPPLE$1.27▼ 3.42%
ETHEREUM$2,358.14▼ 1.61%
DOGECOIN$0.08▼ 3.81%
01

AI & Technology

404 MEDIA

An Ancient Virus Explains Why the Human Brain Is Special

Scientists identified a family of retroviruses that invaded primate genomes and silenced key brain development genes — but left human DNA untouched. The finding reframes human cognitive uniqueness not as something we evolved toward, but something we escaped. This may be the most significant neuroscience story of the year.

A16Z

Why Frontier Models Still Can't Parse a PDF

Reducto's CEO makes the case that AI agents are hitting a document wall — PDFs, scans, and spreadsheets remain stubbornly opaque to even the best models. Until the document layer is solved, agentic AI in healthcare, legal, and finance is largely theater.

SIMON WILLISON

Introducing Kimi K3.1 Pro

Moonshot AI's Kimi K3.1 Pro hit the API this week, with a 2-3x faster "Highspeed" tier and a promise to open-weight the full model by November 4. Another capable non-Western frontier model entering the arena puts more pressure on pricing across the board.

ANTHROPIC / SIMON WILLISON

Donating Agent Skills to the Agentic AI Foundation

Anthropic is handing its Agent Skills standard to the Linux Foundation's new Agentic AI Foundation, making it vendor-neutral and community-governed. This is a standards land-grab dressed as altruism — smart move regardless.

ALL-IN PODCAST

CoreWeave Crash, AI Trade Cracks, Anthropic IPO Red Flags

Ben Horowitz joins to dissect CoreWeave's debt selloff and mixed reactions to Anthropic's investor day. The episode also tackles Anthropic's self-replicating model warning — treated as a psyop by some, a genuine signal by others.

404 MEDIA

Texas Portal Used to 'Rubber Stamp' Data Centers Down After 404 Media Reporting

Texas regulators shut down the permitting portal that had been fast-tracking data center applications after 404 Media exposed the rubber-stamping. Accountability journalism moving regulatory machinery in real time.

02

X / Twitter Signal

Token demand on OpenRouter has surged roughly 27x since last November — from about 6 trillion tokens per week to 165 trillion. A16z flags this as the real answer to why GPU rental prices won't come down: it's not speculation, it's genuine consumption, and open-weight models are driving much of it.

Mollick urges people not to dismiss AI cybersecurity warnings from the labs as hype, specifically citing Anthropic's self-replicating model concerns. He's been a reliable calibrator on AI capability claims, so when he says take it seriously, that's worth noting.

Deutscher drops a counterintuitive observation: AI has made him hire more people, not fewer, because efficiency unlocks scope and appetite for work expands to fill the capacity. It's a clean real-world data point that cuts against the simple "AI kills jobs" narrative.

US and Chinese AI billionaires added a combined $44 billion in wealth in the first nine months of the year. Whatever the broader economic story, the AI wealth concentration story is accelerating — and it's happening on both sides of the Pacific simultaneously.

03

The Thread

The document problem is quietly becoming the agentic AI problem. A16z's interview with Reducto cuts to something most AI demos paper over: real enterprise work lives in PDFs, legacy scans, and formats that models still mangle. You can have the most capable reasoning engine ever built and watch it fail on a 2019 insurance claim form. The stack isn't complete yet, and the gap between "impressive demo" and "consequential work" runs straight through the document layer.
Token demand on OpenRouter is up 27x in a year. The infrastructure skeptics are losing the argument to raw consumption data.
Meanwhile, Anthropic donating Agent Skills to a Linux Foundation entity is the kind of move that looks generous and is also strategically sharp. Owning the governance conversation for agentic AI standards — even via a neutral body you helped create — is how you shape an ecosystem before the rules calcify. Google did it with Kubernetes. The pattern is familiar. The ancient virus story deserves more attention than it'll get in the tech press. The finding that human cognitive distinctiveness may be partly explained by what didn't happen to our genomes — viral silencing of brain genes that hit other primates — is the kind of result that reshapes long-standing assumptions. In a week full of AI capability benchmarks and token counts, the most profound intelligence story came from evolutionary biology.