U.S. College Students Shift Job Credentials from Resumes to Real Works
Job application materials for some college students in Silicon Valley are changing. The past standard included a GPA of 3.8, two internships, student council experience, several competitions, and a few certificates. Now, some students present: 20,000 followers on X, a continuously updated YouTube or TikTok channel, small products made with Cursor, runnable projects on GitHub, real revenue from Stripe, and a group of their own users on Discord. Some have taken on freelance projects, sold products, and made money before graduating.
The change is driven by a cooling job market for recent graduates. Data from the New York Fed shows that the unemployment rate for U.S. college graduates in Q2 2026 is about 5.6%, with an underemployment rate of 42%, meaning more than 4 out of 10 recent graduates are in jobs that do not require a college degree. In the 1990s and 2000s, this group's unemployment rate typically ranged from 3% to 5%. The 5.6% in March of this year is one of the highest levels in nearly a decade, excluding the pandemic period.
Employer screening logic is also changing. A survey by the National Association of Colleges and Employers shows that nearly 70% of employers have adopted skills-based hiring methods, and almost all surveyed employers value internship experience in the U.S. 13.3% of entry-level positions explicitly require AI skills. Data from Handshake indicates that the proportion of full-time positions mentioning AI has nearly doubled in a year, reaching 4.2%. Another survey shows that 40% of students are considering changing majors due to AI, with 10% already having done so.
The threshold for entrepreneurship has been significantly lowered by AI. A survey by U.S. payroll service provider Gusto of 1,051 new entrepreneurs in 2025 shows that businesses founded by Generation Z account for 9% of new enterprises, while baby boomers account for only 5%. For the first time, the number of new entrepreneurs from Generation Z has surpassed that of baby boomers. 71% of Generation Z entrepreneurs are starting with the help of AI, compared to 42% of baby boomers; the proportion of Generation Z who say they would not start a business without AI is five times that of baby boomers. Among all new entrepreneurs, the proportion starting with AI has risen from 21% in 2023 to 60%; the proportion of AI-using businesses receiving venture capital or angel investment is 18%, double that of non-AI businesses.
Student projects have already produced real data. Handle Delivery, a campus delivery service founded by students at the University of Southern California, has processed over 100,000 orders and paid student delivery workers $300,000, raising $1.6 million. The Basics project from Carnegie Mellon University has annual sales of $600,000 to $800,000. Otomo Health, an AI clinic scheduling tool founded by three MBA students from MIT Sloan, raised $2 million. In May of this year, four student teams from Cornell Tech each received $100,000 in prizes.
In market mechanisms, buyers are employers and investors: employers want to replace hard-to-judge GPAs with verifiable outputs, while angel investors and accelerators view students as the earliest source of projects. Sellers are students with works, users, and revenue, who exchange market-validated data for bargaining power. Funding and attention are shifting from traditional prestigious schools and high GPA paths to AI programming tools, payment platforms, and content platforms. Beneficiaries include platforms like Cursor, Stripe, GitHub, and YouTube, as well as students capable of independently creating products; those under pressure are job seekers with only standardized resumes and educational institutions that focus on certificates and credits.
Source: Public Information
ABAB AI Insight
Replacing diplomas with works has precedents in Silicon Valley. In 2011, Peter Thiel established the Thiel Fellowship, initially offering $100,000 to selected individuals on the condition they drop out to start a business. Recipients include Ethereum founder Vitalik Buterin and Figma co-founder Dylan Field. In 2022, Adobe proposed a $20 billion acquisition of Figma, which fell through due to regulatory hurdles; Figma is set to go public in July 2025. Y Combinator launched the "Early Decision" pathway, allowing seniors to apply early and join the program after graduation, directly moving the window for discovering founders to campus. Laszlo Bock, former head of HR at Google, publicly stated in 2013 that GPA and test scores are "worthless" as hiring criteria; IBM subsequently eliminated degree requirements for many positions, and Maryland also removed degree thresholds for numerous state government jobs in 2022.
What capital is doing is lowering the cost of "making products" to a level students can afford while locking in future developers early. GitHub has offered a student developer pack since 2014, providing a full suite of development tools for free; Cursor has opened its Pro version free for one year to students, with its parent company Anysphere's valuation rising from nearly $10 billion to about $29 billion by 2025. YC disclosed that about 25% of companies in its Winter 2025 batch have 95% of their code generated by AI. Stripe's annual report revealed that leading AI companies take an average of 24 months from their first revenue to reach $5 million annually, while traditional SaaS companies take 37 months. The logic of platforms subsidizing students is similar to Apple's educational discount strategy: first capturing user habits, then harvesting throughout their careers.
A more fitting analogy is the "personal is media" era of social media. YouTube allowed a group of creators to bypass television stations and directly accumulate audiences; MrBeast consistently uploaded videos from a young age, ultimately establishing his own consumer brand, Feastables. Today's students have replicated this path at the product level: followers serve as distribution channels, Discord is for user research, and Stripe revenue is for business validation. However, the 2000 internet bubble also provides a cautionary tale, as many students who dropped out to start businesses faced total loss during the crash. The job market is currently in a transitional phase: large companies still retain degree and internship screening, but early-stage startups and investors have shifted to looking at works.
At its core, this is a transfer of pricing power. The pricing power of talent assessment is shifting from universities that issue diplomas to a market constituted by users and revenue. The mechanism is that AI has driven the production costs of resumes, code, and copy close to zero, making these outputs no longer able to prove capability; whereas follower counts, paying users, and Stripe revenue require real market transactions to generate, making them hard to fake with AI, thus becoming new hard currency. The rising unemployment rate among recent graduates has amplified this shift: with fewer positions available, employers are more inclined to use verified outputs to reduce the risk of hiring mistakes. However, this new signal has a strong survivor bias; only the few who make money are visible, while most students' projects do not generate revenue.
ABAB News · Cognitive Laws
Resumes are self-statements, income is third-party proof.
The cheaper the tools, the more valuable the distribution.
Certificates prove you have studied, users prove you are useful.
Notes:
Writing Style: Observations in the post are only included in the first paragraph as background, with the rest of the data sourced in English. The title is written in an action-oriented manner.
Unverified Background: The content in the AI analysis comes from publicly available English information that I have, and this round has not been verified line by line; I recommend confirming again:
Cursor's valuation
YC batch data
Stripe's growth comparison
Thiel Fellowship amount