Thrilled to double down on our partnership with @baseten @tuhinone is one of the clearest thinkers I know on AI infrastructure. He sees a cross section of how AI is actually being used across the leading AI natives, and his job is making them fast, reliable, and cheap at scale. We discuss why inference is becoming one of the biggest markets in software, what happens as open-source models converge with frontier, and a peek into future AI workloads! Enjoy my conversation with Tuhin:

Taught my first class at Stanford Engineering yesterday. "Economics of the AI Supercycle" Condensing everything I've learned investing across the AI stack into a course. Nothing sharpens your thinking like teaching it to brilliant students. Packed room, great questions, blown away by the caliber of students from across the university. This is going to be a fun quarter.

Ads > Subs (!) Ads for AI may be bigger than subs. That's the main finding of Part 3 of "State of Consumer AI" series. Why? Ad Revenue = Time × Ad Volume × Ad Price. All rising. But the most interesting variable is price. "Recommend a laptop under $2k for video editing," on ChatGPT has richer intent than a Google search. Budget, preferences, constraints, all in one prompt. Logged in, so attribution is clean. And bonus, consider AI-native ad formats: conversational ads informed by memory, agentic purchasing, personalized brand moments that feel like features, not interruptions. Some other findings: * ChatGPT owns 68% of all AI time spent. Gemini 16%. Everyone else 16%. * At scale, ads outscale subs. Eg Google's ad biz ($295B) is 6x the largest subscription biz. * Google won't put ads in Gemini. A luxury only a $295B search machine can afford. Full article:

Congrats XBOW on another milestone! If you're a security leader thinking about how to keep pace with AI-powered attacks, the XBOW team (@oegerikus @nicowaisman Niro Albert Jonaki + team) would love to hear from you. When we first invested, the thesis was simple: AI would reshape offense before it reshaped defense. Attackers wouldn't wait for permission to use LLMs. Defenders needed an autonomous system that could match that speed. Since our investment, XBOW has proven the thesis. Top of the HackerOne leaderboard. Over 100 customers. Running safely in live production at some of the most security-conscious companies in the world. What excites me most is what comes next. The shift from point-in-time pentesting to continuous autonomous security testing is just beginning, and XBOW is defining the category. Excited to be on this journey w Ron @Konstantine @laurenmhreeder @wmccarthy11 and welcome our friends Sam Fort Ramin Sayar William Morrison Sanjot Malhi!