Duke Math and CMU CS Grads Top $250K, But Data Lags AI Shift

Federal earnings data tracking college graduates four years after commencement reveals a stark divide in the U.S. labor market: while median pay across all institutions sits at $64,500, alumni from specific technical programs at elite universities are pulling in more than quadruple that figure. Yet those numbers reflect hiring conditions from before artificial intelligence began restructuring entry-level tech recruitment.

Mathematics graduates from Duke University lead the highest-earning cohorts, reporting median annual earnings of $297,029. The figure stems from a cohort of just 17 students who completed their degrees during the 2017–18 and 2018–19 academic years. Because the dataset relies on federal tax records processed by the higher education research firm HEA Group, it captures compensation before the recent wave of layoffs and hiring freezes swept through major technology firms and financial institutions.

Carnegie Mellon University dominates the broader upper tier of the earnings distribution. Six of its programs produced graduates earning above $150,000 annually, more than any other institution in the analysis. Computer science alumni reported median earnings of $268,121, followed closely by electrical, electronics, and communications engineering graduates at $250,168. Across roughly 20 programs nationwide where median pay exceeded $200,000, the total student population covered amounts to approximately 2,000 individuals—a fraction of the millions holding bachelor’s degrees.

Burcu Akinci, dean of Carnegie Mellon’s college of engineering, attributed the compensation levels to employer demand for systems thinkers capable of solving complex problems tied to emerging technologies. The university’s admissions model differs significantly from peers with similar earning outcomes; its acceptance rate for the Class of 2029 stood at 11.1%, compared to sub-5% rates at Duke and several Ivy League schools.

The premium attached to these degrees arrives alongside steep upfront costs. Total annual attendance for on-campus students at Carnegie Mellon approaches $94,000, while many comparable private institutions have crossed the $100,000 threshold. Students financing these programs through federal loans face a repayment landscape that has shifted considerably since the tracked cohorts entered the workforce.

Recent labor market indicators complicate the historical earnings narrative. According to the Federal Reserve Bank of New York, computer engineering graduates faced a 7.8% unemployment rate in 2024, the second-highest among all college majors. Computer science graduates experienced a 7.0% rate. Both figures sit well above the 4.2% overall unemployment rate for recent college graduates, signaling that the entry-level pipeline for historically lucrative technical roles has narrowed.

This divergence between past earnings peaks and current hiring friction is accelerating federal scrutiny of higher education return on investment. Under rules finalized by the U.S. Department of Education, post-secondary programs must demonstrate that their graduates earn more than comparable adults holding only a high school diploma. Programs failing this benchmark risk losing access to federal student loan funding.

Approximately 95% of enrolled students attend programs projected to clear the earnings test. Those most vulnerable to losing federal aid eligibility include short-term certificate programs in cosmetology and vocational trades, alongside degree programs in music, fine and studio arts, and certain health-related fields.

Steve Taylor, policy director at the economic mobility nonprofit Stand Together Trust and a member of Virginia’s State Council of Higher Education, noted earlier this year that while higher education carries civic and cultural value beyond wages, requiring students to take on federal debt demands a reasonable path to repayment.

The Department of Education continues to expand its College Scorecard database, integrating subsequent graduation cohorts to update program-level earnings metrics as older data ages out of relevance.