Like thousands Earlier this month, influential figures descended on Southern California for the annual Coachella music festival. A Silicon Valley-style show called “AI Coachella” was taking shape a few hundred miles north in Palo Alto. The CS 153 class is one of Stanford’s most popular offerings this semester, and like a music festival, it’s packed with stars – in this case, not pop artists, but Gigantic Tech CEOs.
The course is co-taught by Anjney Midha, former General Partner of Andreessen Horowitz, and Michael Abbott, former Apple Vice President of Cloud Services Engineering. The guest lecturer list reads like a Signal group chat that many VCs would pay to include: OpenAI CEO Sam Altman, Nvidia CEO Jensen Huang, Microsoft CEO Satya Nadella, AMD CEO Lisa Su, anthropic philosopher Amanda Askell and White House senior policy advisor on artificial intelligence Sriram Krishnan and others. This is the fourth year that Midha and Abbott have taught some version of this class. After registration opened this year, the 500 class seats quickly filled up, with dozens of students on the waiting list and thousands more watching lectures posted on the website YouTube.
Andreessen Horowitz co-founder Ben Horowitz spoke on Tuesday. I planned to attend, but at the last minute, Midha’s spokesperson told me the class was too full for journalists to enter.
Part of Stanford’s allure has long been access to Silicon Valley’s elite. Its campus is just a few miles from Sand Hill Road, which is home to high-profile venture capital firms, and it’s not uncommon to see San Francisco startups like Cursor and Vercel recruiting from the school’s computer science clubs. CS 153 combines access to top Silicon Valley executives with education in an extreme way – which is why some people have trouble with it.
After sending a screenshot of the CS 153 guest lecture lineup viral this year on social media, some critics argued that students should be spending time in “real” classes rather than participating in a live podcast recording hosted by a VC. There are rumors on campus that other Stanford professors are irritated by what some see as a celebration of pure power.
“Everyone is taking CS 153. Only 3 people took my Functional Analysis class at Stanford today,” wrote Luke Heeney, an associate professor of economics at Stanford University, in another post. “Remember to eat your vegetables.”
Midha leaned into mockery. He has ordered 500 T-shirts that say “I took CS 153 and all I got was Coachella AI,” which he plans to distribute to students on Thursday. “Critics unintentionally ganged up on my system,” he tells me, describing the failure in the language of an infrastructure engineer. “I thought, what, AI Coachella? Is it a feature or a bug? It’s totally a feature. It’s product-market fit.”
Midha and Abbott recently launched a novel venture capital firm, AMP, which aims to provide AI startups with both capital and computing power. At the beginning of the class, Midha revealed that several guest lecturers run companies he has invested in, including Black Forest Labs, Mistral, Sesame, and Periodic Labs. But that access is part of the appeal of this class.
So what exactly do Stanford students learn at AI Coachella? The coursework mainly covers pioneering artificial intelligence systems that many undergraduate computer science courses only touch upon. Midha spent her first lecture of the year discussing the computing infrastructure that supports artificial intelligence models. He argued that AI chips are not commoditized, meaning their price does not decline over time. To prove his point, he shared internal charts he collected on AMP showing the Nvidia H100 price escalate over the last 90 days.