Morning Scan

The AI revenue story keeps splitting in two directions: numbers so large they barely sound real, and the cash-flow mechanics underneath them. Apple's regulatory headaches are compounding, and OpenAI cannot make an announcement without triggering a backlash cycle.

$65B — Anthropic's revenue run rate as of late July, up from $47B in May and $9B in late 2025. (Bloomberg)

Machines & Minds

Sources: Anthropic's revenue run rate reached $65B by the end of July (paywall)

Up from $9B nine months ago — a growth curve that makes IPO bankers salivate and everyone else question its durability. (Bloomberg)

Situational Awareness sold part of its $5B Anthropic stake at a 20% discount (paywall)

Wall Street noticed the fund was "being hunted" before it confirmed it needed cash — a reminder that even AI darlings' backers can get squeezed. (WSJ)

Crusoe in IPO talks with four Wall Street banks at a $35B valuation

The AI-infrastructure buildout keeps producing IPO candidates faster than anyone can evaluate them. (Axios)

SEC staff says data center securitization sits outside Dodd-Frank risk retention rules

Comes as Nvidia and a consortium including Apollo, BlackRock and KKR line up $500B in AI infrastructure financing — the plumbing matters as much as the headline number. (The Next Web)

Nvidia backs OpenAI data center, Anthropic news, Google buys Spirit Airlines data

Ben Thompson's read: data might finally be the new oil, and everyone's drilling. (Stratechery)

Machines & Minds

OpenAI's big slowdown

New safeguards will deliberately slow future development — a rare instance of an AI lab hitting its own brakes. (Alex Heath)

OpenAI built a safer ChatGPT, but not for you

The teen version gets guardrails the general product still lacks, which is its own kind of statement. (Machinesociety)

BREAKING: OpenAI's unraveling has begun

Gary Marcus is never subtle, but the reaction to Altman's Tuesday announcement suggests patience with OpenAI's messaging is wearing thin. (Marcus on AI)

David Sacks: "Amodei believes frontier AI is too powerful to distribute; we believe it is too powerful to centralize"

The Washington AI policy fight in one line — decentralization vs. safety-first control, with real regulatory stakes behind the soundbite. (X)

Connected World

Apple acknowledges App Store regulatory changes are hurting its $100B+ services business (paywall)

US App Store spend fell 6% YoY in Q2 — the first hard evidence that antitrust pressure is denting Apple's most profitable segment. (Financial Times)

Meta faces mammoth fine in landmark social media trial

The addictive-design allegations finally get their day in court, with real money on the line. (SiliconANGLE)

Discord suspends livestreams in Brazil after child-safety order

Another data-protection authority moving faster on platform accountability than US regulators have managed in years. (AP)

Baidu posts fifth straight quarterly sales decline, lagging AI rivals (paywall)

Revenue down 2% YoY to ~$4.62B — a slow bleed that shows what happens when you're not setting the AI pace. (Bloomberg)