Applications are now closed. Learn more about our other AI Garage initiatives.
The Mayfield AI Garage is a launchpad for AI founders, meeting them where
they are in their journey and providing them with resources and mentorship to
take their ideas to the next level.
Access to Industry-Leading AI Resources and Partners
Stipend of $25k per founding team member (up to $50k) to fuel your AI startup.
Get access to office space, workshops, mentorship, and all the benefits of SkyDeck’s Pad-13 incubator program.
Gain insights and guidance from the Mayfield team to help scale your startup.
Receive a minimum of $50K in compute credits and other benefits to power your AI development.
Gain expert legal advice to help company formation, financing, and other early-stage legal work.
Current UC Berkeley undergraduate and graduate students
UC Berkeley Alumni up to three years post-graduation
Ideating or working on an AI project
Mayfield AI Garage is a hands-on program designed to help students turn their AI ideas into companies by providing capital, mentorship, compute credits, and other benefits. Over 6 transformative months, students will establish a strong foundation for building their companies through support from Mayfield, NVIDIA, Microsoft for Startups, Orrick and SkyDeck. We’re committed to helping passionate founders develop AI companies that can change the world.
We welcome applications from all current UC Berkeley students and alumni up to 3 years post-graduation (Class of 2025, 2024, 2023).
Students may have an early-stage prototype or minimum viable product, or an AI project with initial customer interest. We will consider projects across all layers of the Mayfield CaaS stack.
The Mayfield AI Garage at Berkeley will run from May to November 2026.
Students who accept a spot are expected to work on their idea full time during the summer, and part-time in the fall (for those returning to school). Furthermore, students may not participate in any other paid or unpaid internship during the program. All members of each team are expected to contribute equally.
Stipend: $25K per team member, for a maximum of $50K.
Compute: $50K in compute credits per winning team (up to $350K) from Microsoft for Startups.
Legal: Preferred relationship terms with Orrick and up to $25K in legal credits.
Become part of NVIDIA’s Inception program for startups (includes all benefits of program); mentorship from NVIDIA leader.
Become part of SkyDeck’s Pad-13 incubator program for startups. Includes office space, workshops, mentorship, and all other benefits of the program.
Geography: Must be based in the Bay Area.
Eligibility: Must be current UC Berkeley student or alumni up to 3 years post-graduation.
This role is structured as an independent contractor (1099) engagement.
There are no restrictions on team size; however, the stipend will be capped at $25K per person, for a maximum of $50K.
As AI scales across specialized silicon, idle compute breaks the economics. Networking is now the critical chokepoint in AI infrastructure.
Why enterprise AI is a trillion-dollar opportunity and how product-led trials, ROI proof, and governance turn experimentation into scaled adoption.
As AI accelerates in 2026, timeless startup fundamentals still win. People first, painkillers not vitamins, and disciplined early-stage investing.
The secret to building billion-dollar companies isn’t just a great idea—it’s the foundation of a long-term founder-investor relationship.
If you’re building in AI, the rules just changed. Here’s a preview of the new math, new skills, and new opportunities shaping the next generation of breakout startups.
Learn why apps are becoming a service layer, how this shift will change product usage and monetization, and why the new question for founders is: “How do I design for a world where intelligence is the interface?”
Navin Chaddha reflects on the firm’s latest LP meeting, reflecting on Mayfield’s commitment to partnering with founders to build the future of AI, together.
Backing founders from day zero who see limitless possibilities where others see constraints.
