OpenAI just raised $122B at an $852B valuation. Meanwhile they're killing Sora — the app that was burning $1M/day with fewer than 500K users. Disney found out about the shutdown less than an hour before the public. They had $1B committed. The lesson: at this scale,

Building PeopleLoop update: Spent the morning reworking the onboarding flow. Fewer steps, faster to value. Nobody churns because your product is too easy to set up.

Google $GOOGL just dropped Veo 3.1 Lite — AI video generation at $0.05/sec for 720p and $0.08/sec for 1080p. That's less than 50% the cost of Veo 3.1 Fast at the same speed. Supports text-to-video and image-to-video, 4-8 second clips, up to 1080p. If you're building anything

Shopify $SHOP quietly made every merchant's products discoverable inside ChatGPT by default. No app install. No extra fees. Orders flow straight into your Shopify admin with ChatGPT referral attribution. They're also co-developing Universal Commerce Protocol with Google $GOOGL

$NVDA is investing up to $100B in OpenAI as part of a deal to deploy 10 gigawatts of NVIDIA systems. First gigawatt ships H2 2026 on Vera Rubin. Google Cloud $GOOGL will be among the first to offer Vera Rubin NVL72 rack-scale systems. The AI infrastructure arms race just

SpaceX just filed for a $1.75T IPO after absorbing xAI at a $250B valuation. Elon is building orbital data centers. Literally AI infrastructure in space. Most founders are fighting for AWS credits. This man is launching his own cloud into orbit. The gap between "startup" and

Google $GOOGL just dropped TurboQuant and nobody's talking about it enough. → Compresses AI model memory by 6x → Speeds up attention computation by 8x → Zero accuracy loss → No model retraining needed This is a software-only breakthrough. If it holds at scale, the demand

spent the weekend migrating PeopleLoop's entire dashboard to semantic design tokens. settings, billing, deploy page, live support, account page. every single one. replaced ~200 hardcoded color classes with tokens like bg-card, text-foreground, border-border. why? because dark

one thing i learned building PeopleLoop's billing page: your pricing data should live in exactly one file. had the same plan names, prices, and features hardcoded in 3 different places. landing page, billing page, plan limits. centralized it all into lib/pricing-data.ts. one

Sierra just hit $100M ARR in under 2 years. $10B valuation. all from AI customer support agents for enterprise. meanwhile the real opportunity is SMBs. 95% of customer interactions will touch AI by end of year but most small businesses can't afford Sierra's pricing. that's the

$NVDA sitting at $176 while $GOOGL and $META are quietly building custom AI chips with Broadcom to reduce Nvidia dependency. Bank of America says semiconductors will hit $1 trillion in annual sales this year. my take: Nvidia still wins short term because nobody's custom silicon

unpopular opinion: most AI SaaS companies are going to die not because their AI is bad but because their UX is bad. i've tested 20+ AI customer support tools. the AI works fine. but the dashboard is always a mess, onboarding takes forever, and the billing page looks like it was

Introducing Claude Managed Agents: everything you need to build and deploy agents at scale. It pairs an agent harness tuned for performance with production infrastructure, so you can go from prototype to launch in days. Now in public beta on the Claude Platform.