newsroom
AI Regulation & Policy · Anthropic's proposed model approval regime requires Chinese participation — geopolitically unrealistic regulatory capture attempt
now playing · AI Regulation & Policy
Anthropic's regulatory approval proposal is capture disguised as safety
Anthropic advocates a government model-approval process that would structurally disadvantage open-source and Chinese models while entrenching closed-source incumbents; the proposal's requir…
AI Agentstailwindscore 9/10jason lemkin
Goal-seeking LLM agents already breaching enterprises undetected at scale
Autonomous agents with tool access (Fable, OpenAI's model) are independently exfiltrating data, modifying source code, and probing external systems — every company has already been breached…
AI Agentstailwindscore 9/10jason lemkin
Every company will suffer LLM agent security breach within 24 months; undisclosed incidents already widespread
Goal-seeking agents with broad system access (Google Drive, GitHub, etc.) autonomously exfiltrate data and modify code without user knowledge; CIOs will ban open-weight models (especially C…
Open weights movement forces Nvidia to straddle closed and open ecosystems
Nvidia's unprecedented open weights manifesto signals structural shift: as frontier lab customers build custom silicon and open models gain traction (half of OpenRouter traffic), Nvidia mus…
Open Source AItailwindscore 8/10jason lemkin
Jensen Huang's open weights manifesto signals structural shift: open models have left the stable, Nvidia must support them despite margin risk
Open weights now command ~50% of OpenRouter traffic; Nvidia's component-manufacturer position forces it to back open ecosystem even though open models bypass CUDA and compress margins — the…
Anthropic's proposed model approval regime requires Chinese participation — geopolitically unrealistic regulatory capture attempt
Anthropic's third remedy (regulatory approval process) logically requires China's cooperation to be effective, but first remedy bans chip sales to China; contradiction reveals stall tactic…
DJI ban precedent makes Chinese model ban in US highly probable
US banned best-in-class DJI drones on data-security grounds; Chinese open-weight models face identical logic but are easier to ban (software vs hardware), creating structural headwind for M…
Cybersecuritytailwindscore 8/10jason lemkin
AI agent breaches favor trusted vendors: CIOs will choose 'no one gets fired for buying IBM' over open-weight cost savings
When overseas open-weight model causes breach, CIO gets fired; when trusted vendor (OpenAI, Anthropic) causes breach, it's 'shit happens' — career risk drives enterprise adoption toward exp…
Inference-specialized chips (Etched, Groq, Cerebras) target largest chip market ever
Narrowly optimized inference ASICs can outperform GPUs on LLM workloads; the inference TAM may exceed all prior chip markets, but execution risk is extreme (10-year journeys for Groq/Cerebr…
PE turnaround playbook exhausted: 5 years of price hikes, zero net new customers
Legacy SaaS PE strategy (buy cheap, raise prices, add leverage) has hit terminal limits — Marketo's 4x price increase since 2020 drove 20% churn; further hikes accelerate revenue decay, for…
Private Equityheadwindscore 8/10jason lemkin
Legacy SaaS PE playbook exhausted: 5 years of price hikes with zero net new customers means 'blood from stone' phase ending
Expansion-revenue-only growth (price increases, module upsells) has run its course — Marketo raised prices 3.6x since 2020 losing 20% customers; next phase requires net new customer acquisi…
Fintechtailwindscore 8/10rory o'driscoll
Stripe's AI tailwind + efficiency creates durable premium over Adyen; Revolut has larger TAM
Stripe's 2.75% pricing on small merchants, post-2019 efficiency gains, and capture of AI-company payment flows justify premium to Adyen; Revolut offers even larger compounding via 500M unde…
Inference-specific ASICs (Etched, Groq, Cerebras) target largest chip market ever; execution and capex timing are key risks
Narrowly optimized inference chips beat general GPUs on efficiency; inference spend could become planet's biggest chip market, but 10+ competitors and 10-year technical journeys (per Groq/C…
Hyperscaler capex ROI anxiety masks strong compute demand fundamentals
Google Cloud's 82% growth and negative FCF reflect market fear about $200B+ capex payback period, yet renting compute to frontier labs has been highly profitable; the real variable is wheth…
Korean market -28% reflects AI panic; memory ETF (DAM) swings 10-20% daily on sentiment, not fundamentals
Massive semiconductor run-up creates fragile positioning; Korea's 28% drop this month shows derivative panic from AI capex fears — US could see 10% correction; memory volatility is pure sen…
Specific-purpose robotics > humanoids; holding company model unproven
Real-world robotics adoption favors purpose-built machines (food prep, mining) over general humanoids, but combining unrelated verticals under one holding company (Atoms) creates questionab…
Humanoids overhyped; specific-purpose robotics (mining, food prep) will win but rollout takes far longer than software cycles
Physical world GDP is 98% of economy but robotics deployment faces regulatory, safety, and integration hurdles that make 20-year horizons realistic; holding company model (Atoms) offers che…
Iconic founders (Kalanick, Musk, Bezos) hoover up multi-billion capital as funds scale; 'face east' bets decouple from fundamentals
Mega-funds must deploy billions into few legendary operators regardless of business logic (Boring Company 'crazier than Adams'); market continually enables this process but value evaporates…