a16z invests $35 million in a two-year non-accredited university alternative for students aged 16 to 22, applications for the founding class open this week
Andreessen Horowitz has invested $35 million in the Horowitz Andreessen Academy in San Francisco, creating a non-accredited two-year university alternative for students aged 16 to 22. The academy is independently operated by Udemy co-founder Gagan Biyani, with Marc Andreessen and Erik Torenberg joining the board. Total funding amounts to $42 million, with approximately $7 million coming from founding partners.
Founding partners include Anduril, Anthropic, Coinbase, Google, Meta, Nvidia, OpenAI, Palantir, Replit, and Stripe, providing software, computing power, office space, and internships. Each student receives over $50,000 in computing credits and a $5,000 travel research budget, and can access the network of fund investors. Guest speakers include OpenAI's Sam Altman and former Windows chief Steven Sinofsky. Additionally, over 40 companies, including Applied Intuition, Databricks, and Shopify, are recruiting partners.
Applications for the founding class open this week: approximately 50 students, one year, tuition-free, with enrollment in fall 2027, allowing for a gap year or deferred college admission, prioritizing recent high school graduates, and accepting current undergraduates and international applicants. The full two-year program is planned to launch in fall 2028, pending regulatory approval, with tuition comparable to elite private universities, where Stanford's undergraduate annual fee is about $68,000. The academy explicitly does not award degrees or credits.
About 80% of the curriculum consists of projects rather than lectures, with no traditional grades, exams, or homework, and instructors do not need teaching qualifications. Biyani states the goal is to cultivate the "most ambitious young builders" who can enter top tech companies or start their own, treating the ability to build in the AI era as the number one skill.
Those buying are looking to lock in builders and interns from model companies and defense tech firms, while those selling are compressing four years of university into a San Francisco-based program for venture capital. Funding is shifting from a donation-based education narrative to a talent pipeline that can be recruited. Beneficiaries include partner companies that can provide computing power and workspaces, as well as the fund's project flow, while traditional four-year colleges that sell general education credits and families who can afford comparable tuition but do not receive accreditation are under pressure.
The two-year program cannot enroll students until approved. The founding class will be selected based on work and online results, without considering standard transcripts.
Source: Public information
ABAB AI Insight
Andreessen Horowitz has long characterized universities as institutions that delay entrepreneurship; this time, instead of discouraging dropping out, they are establishing an unaccredited school to harness dropout energy within their own board. Biyani has transitioned from selling courses at Udemy and small classes at Maven to a full-time on-campus model, transforming the online education customer acquisition model into a San Francisco housing plus project assessment approach. The fund's $35 million investment and model companies providing computing power equate to replacing state university credit production with GP funds and GPU quotas.
The capital path is talent options. The $50,000 in computing power ties students to the cloud of partners, while recruiting partners move interviews to before the second year, and guest speakers turn Altman and Sinofsky into recruitment advertisements. The academy does not issue degrees; instead, it provides a ticket to enter the a16z network. Tuition is benchmarked against Stanford, indicating that pricing is based on elite selection premiums rather than teaching costs; the initial 50-student tuition-free class will serve as a portfolio before seeking a two-year license from regulators.
This is analogous to Peter Thiel's dropout scholarships, Minerva and Lambda School's project-based alternatives, and Y Combinator's batch education. The difference is that this school is directly opened under the fund's brand, with partners from Anthropic to Anduril covering models and defense. The industry phase is tech companies transitioning from campus recruitment to building their own preparatory programs: the certification system cannot keep up with model iterations, and companies are now using project work as resumes. A failure comparison is unaccredited coding boot camps struggling with employment rates and loans.
The structural change represents a transfer of pricing power. The pricing power of a four-year degree lies with accrediting bodies and federal student loans, while the pricing power of a two-year program lies with computing quotas and recruiting lists. The mechanism is: AI jobs require work that can adjust models, not general education credits; whoever can simultaneously provide computing power, workspaces, and the first internship will rewrite the talent funnel during the year of turning 18, leaving universities on a track that requires licensing.
ABAB News · Cognitive Law
After not discouraging dropping out, venture capital begins to establish schools
Computing quotas become admission tickets earlier than credits
An unaccredited academy sells recruitment lists, not diplomas