A degree used to be the price of admission to Silicon Valley. A growing number of the most consequential AI founders of the 2020s decided not to pay it.
Mercor: The youngest self-made billionaires in the world
Brendan Foody, Adarsh Hiremath, and Surya Midha met at Bellarmine College Preparatory in San Jose, where they were debate teammates and captains of one of the top-ranked debate programs in the country. After graduating in 2021, Foody and Midha enrolled at Georgetown to study economics and foreign service respectively, while Hiremath enrolled at Harvard to study computer science and machine learning. In the spring of 2023, during their sophomore year, all three dropped out. Foody told Fortune he had already decided to leave before finals and simply did not sit the exams.
The idea emerged from a hackathon in São Paulo, where the three discovered they could match companies with skilled engineers abroad, handle logistics, and take a cut. Their first client paid $500 a week for a developer. Mercor kept roughly 30 percent. By the time they left campus, they had a concept, a co-founder team, and no outside funding.
What followed was one of the fastest revenue ramps in startup history. Mercor raised $3.6 million in seed funding from General Catalyst in 2023, $32 million in Series A from Benchmark at a $250 million valuation in September 2024, $100 million from Felicis at a $2 billion valuation in February 2025, and $350 million in Series C at a $10 billion valuation in October 2025. The company pivoted from software recruiting to matching expert contractors, doctors, lawyers, engineers, and bankers, with AI labs needing human feedback to train their models. Clients include OpenAI and Anthropic. The platform now employs more than 30,000 contractors earning an average of $85 per hour and pays out more than $1.5 million daily. All three founders received Thiel Fellowships in 2024. Forbes named them the youngest self-made billionaires in the world following the $10 billion valuation, displacing Alexandr Wang, who had held the title.
Etched: The Harvard dropouts betting the company on transformers
In 2022, Gavin Uberti and Chris Zhu were Harvard students when Uberti completed a summer internship working on compilers and came back convinced there was a hardware opportunity nobody was pursuing. The pair dropped out and brought in Uberti’s college roommate Robert Wachen. Together they founded Etched in Cupertino with a focused and deliberately risky thesis: build a chip that does one thing only, which is run transformer-based AI models.
The chip is called Sohu. It is an application-specific integrated circuit fabricated on TSMC’s 4-nanometer process, designed to hard-wire the matrix multiplication patterns specific to transformer inference. Uberti claims one eight-chip Sohu server replaces 160 Nvidia H100 GPUs and delivers more than 500,000 Llama 70B tokens per second. The thesis in his own words: “If transformers go away, we’ll die. But if they stick around, we’re the biggest company of all time.”
All three founders received Thiel Fellowships. In June 2024, Etched raised a $120 million Series A co-led by Primary Venture Partners and Positive Sum Ventures, with participation from Peter Thiel, GitHub CEO Thomas Dohmke, and Cruise founder Kyle Vogt. In January 2026, the company raised approximately $500 million in a round led by Stripes at a valuation of approximately $5 billion, bringing total funding to around $620 million.
Pika: The Stanford PhD dropouts who built a video generator for everyone
Demi Guo and Chenlin Meng met as PhD students in Stanford’s Artificial Intelligence Lab, where both had enrolled in 2021. Guo had graduated from Harvard before enrolling at Stanford and had previously worked as a research engineer at Meta AI Research. In the winter of 2022 to 2023, the pair participated in an AI Film Festival organized by Runway and came away frustrated by the quality of existing video generation tools. In April 2023 they dropped out of their PhD programs and founded Pika, with the goal of building an AI video generation platform accessible to people without technical backgrounds.
Pika launched first on Discord and gained traction rapidly. In November 2023 it raised $55 million in a Series A led by Lightspeed Venture Partners, following earlier rounds led by Nat Friedman and Daniel Gross. The company subsequently raised additional funding, reaching $135 million total at a $700 million valuation. By October 2025 the platform had reached 14.5 million users. The company is headquartered in Palo Alto.
Guo has described the product’s core insight simply: most people will never try to produce a film using AI, but many people want to make short videos for self-expression. Pika 2.5, the latest version of its engine, shipped in early 2026.
Turbo AI: Five million users, profitable, and barely drinking age
Rudy Arora and Sarthak Dhawan have been friends since middle school. Dhawan had previously built UMax, a self-improvement app that reached number one on the App Store with 20 million users and $6 million in annual revenue. In early 2024, both were enrolled in college, Dhawan at Duke, Arora at Northwestern, when they built Turbo AI, then called Turbolearn, as a side project to solve a classroom problem Dhawan described simply: he could not listen to a lecture and take notes at the same time.
The tool records lectures and automatically generates notes, flashcards, and quizzes. It spread first through Duke and Northwestern, then reached Harvard and MIT through dorm chats and group messages, and eventually expanded across campuses nationwide. Both founders dropped out of college in 2024.
By October 2025, Turbo AI had reached five million users, was adding approximately 20,000 new users per day, and had reached eight-figure annual recurring revenue, while remaining profitable throughout its existence. The company had raised only $750,000 in total outside funding, having grown entirely through product-led and social media-driven acquisition. Subscriptions are priced at approximately $20 per month. The 15-person team is based in Los Angeles.
Professionals including consultants, lawyers, doctors, and analysts at firms including Goldman Sachs and McKinsey have adopted the platform, uploading reports and converting them into summaries or podcast-style audio for commutes. The company rebranded from Turbolearn to Turbo AI to reflect this expanding user base.
Discover more founders in various industries:
San Francisco Founders Under 30 Who Are Already Reshaping AI
Restaurant Tech Founders Driving Innovation In San Francisco’s Dining Economy
AgTech in the San Francisco Bay Area: The Founders Rewiring Food, Farming, and Biological Production
