A quiet week for new posts, but a heavy one for money and infrastructure. Anthropic edged toward the biggest IPO in history, Google’s leadership shakeup kept rattling markets two weeks on, and researchers confirmed the first fully autonomous AI cyberattack on a government.
SAIL authors spent the week on the ledger side of all three: what a bubble actually looks like, what a $6 agent costs, and who's actually footing the bill for open models.
The Week in AI
Anthropic—which confidentially filed S-1 paperwork on June 1—could publicly file for an IPO as early as late August following a $65 billion annualized revenue run rate in July, targeting a raise that matches or beats SpaceX’s record $75 billion at a potential $2 trillion valuation.
Suspected China-linked hackers executed what security firm Dream calls a near-autonomous AI cyberattack using open-source frameworks, compromising 85 Taiwanese government accounts and extracting over 2,500 personnel records before probing the country’s nuclear safety agency.
Alphabet shares fell following the August 5 departure of Chief Scientist Jeff Dean, who left after 27 years to start the AI-for-science venture Discovery Loop alongside fellow senior Google researchers Sanjay Ghemawat, Quoc Le, and Oriol Vinyals.
This Week from SAIL Authors
Reading the Bubble
🫧 Is AI a bubble yet? Our five gauges say no — Azeem Azhar and Nathan Warren on why demand still justifies the spending, even as unwinding it gets harder. — Exponential View
🔮 The market misread Google’s AI exodus — Azeem Azhar and Hannah Petrovic argue Jeff Dean’s departure says the AI investment cycle is earlier than investors think. — Exponential View
Pricing the Machines
🔮 The curious economics of a $6 AI agent — Azeem Azhar and Marija Gavrilov on Amazon’s $1.8 million Claude project that ran unnoticed for five months, and why nobody can price AI work yet. — Exponential View
📈 Data to start your week — Azeem Azhar and Hannah Petrovic chart an AI revenue update, the frontier usage gap, hidden solar capacity, and China’s robot dominance. — Exponential View
Open Models, Real Money
Teaching Everyone to Fish for Tokens — Nathan Lambert on why Nvidia is spending $26 billion to keep open-source model building alive, and why it wants you training your own. — Interconnects
Building an AI Text Detector From Scratch — Sebastian Raschka walks through dataset construction, model training, local deployment, and RLVR end to end. — Ahead of AI
Full Library
🫧 Is AI a bubble yet? Our five gauges say no — Azeem Azhar & Nathan Warren | Exponential View
Teaching Everyone to Fish for Tokens — Nathan Lambert | Interconnects
📈 Data to start your week — Azeem Azhar & Hannah Petrovic | Exponential View
🔮 The curious economics of a $6 AI agent — Azeem Azhar & Marija Gavrilov | Exponential View
🔮 The market misread Google’s AI exodus — Azeem Azhar & Hannah Petrovic | Exponential View
Building an AI Text Detector From Scratch — Sebastian Raschka | Ahead of AI
