Andreessen Horowitz, the Silicon Valley venture firm known as a16z, has an investing practice it calls American Dynamism. It backs companies that “support the national interest.” Its sectors are aerospace, defense, public safety, education, housing, supply chain, industrials and manufacturing. Education is on that list. This piece is about one part of education, college.

The case behind American Dynamism is simple and right. The country has to build again. Call it the buildout. To me, the buildout means restoring the physical economy that a generation of pure software let go slack. Building takes capital and it takes talent, and the whole argument is that both should turn back toward the national interest.

I have spent years measuring one place where they are still turning the other way.

Before a single new shipyard or chip line hires anyone, the country makes one of its biggest bets of talent and capital, every year, through its colleges. Eighteen-year-olds and their families pick where to spend four years and, at the costliest schools, well over $300,000. Federal and state governments underwrite a significant share of the bill. And in my experience, few families run the numbers on what comes back.

So I did, across nearly 6,000 schools. I compared what a degree costs and pays back over a lifetime, after taxes, with going straight to work out of high school.

The headline is blunt. At full price, the median four-year degree, the one in the middle of the pack, loses about $82,000 in lifetime value, in today’s dollars, against skipping it.[1] At the net price families actually pay, after grants and scholarships, it still loses about $17,500.

Count every school in higher education, certificate programs included. At more than a quarter of them, most former students, graduates and dropouts alike, earn less than the typical high school graduate six years after they enroll.[1]

About 42 percent of recent graduates are underemployed, working jobs that typically do not require a degree.[2]

Step back to the whole workforce and about 24 million degree-holders are either unemployed or working jobs that typically do not require one.[3]

All of this usually gets told as a consumer story. Buyer beware, read the fine print, do not overpay for a brand. That framing is true and too small.

Run it through the American Dynamism lens and it becomes a story about what the country can build, and a more uncomfortable one.

Start with the capital. Federal student aid alone comes to about $155 billion a year, not counting what families pay out of pocket, and it pays for an investment, the four-year degree, where the median one loses to going straight to work.[4]

That is not money sitting idle. It is money the buildout competes for. Degrees are priced like investments, but at the median they pay back less than going straight to work, more like spending than investing. If you care about where American capital goes, this is one of the biggest single flows, and it gets far less checking than a careful lender would give a loan.

Then the talent, which is worse.

Beyond the price, the system sorts people away from the work the country needs. About one in five of our graduates earns a STEM degree, against more than two in five in China.[5] And the pipeline that sorts them runs on prestige, not on what a graduate can do.

Picture two workers who spent the same years in college. One left with the diploma and one left without it. Over the four decades through 2012, the one with the diploma earned about 36 percent more, on average. In the decade since, the gap is about 48 percent.[1]

More and more of what the market pays is for the diploma itself rather than the years of study behind it. That is exactly backward from what a country trying to rebuild its physical economy needs.

And many of the pathways that staff that economy are the ones the prestige system teaches families to look down on. In the model, a licensed trade apprenticeship beats the median four-year degree by a wide margin for those who finish it, and going straight to work beats it too.[1] Routes like these are how you staff a factory floor and a flight line, not a consolation prize.

The students who choose college can beat the median degree too. Whether a degree pays back what a family puts in turns on three choices: finishing it, where and what to study, and a price the degree can repay.

Even if college were free, about 30 percent of four-year colleges would still leave their median graduate behind against going straight to work.[1]

The system does not sort students on those three choices. It sorts for the brochure. So the country spends the talent and the capital, then buys the average degree when it needs to buy the winners.

If talent is part of the country’s infrastructure, then college spending is an infrastructure problem, and it comes ahead of most of the others, because every other project needs trained people. It is also measurable. A problem you can measure is one you can build against.

Three things follow. The first takes Congress, and people can build the other two.

Finish the outcomes filter Congress started in 2025. Its earnings test asks only whether an undergraduate program’s graduates out-earn a typical high school worker, usually in the school’s own state, and it never asks what the degree cost them.[6] Add the price, so federal money stops funding programs that reliably lose. Any careful lender checks both.

Build the alternative pathways out as real institutions, apprenticeship platforms and skills-based credentials, and make them as easy for employers to read, and as respected, as a four-year degree is today.

And finish moving employers from screening on the diploma to hiring for the skill. Degree requirements shut out many capable people who never finished one. Hiring for skill lets them in, and much of that talent is the kind the country has to find and develop, and cannot afford to waste.

I am biased on that last part, and I will say so. I am a first-generation immigrant who got one of the bets that paid, a full scholarship at a school that taught a skill and charged a fair price. The system found me. It is finding fewer people like me than it should, and a market that pays a growing premium for the stamp itself is part of why.

The buildout the country wants needs the opposite. Find the talent, teach it something real, and stop burning the capital on the way.

I built the model that runs this school by school. Anyone can check their own school and major at collegeroi.org. The full argument is in the book.

Reference Sources

1. Author’s model, built on U.S. Department of Education College Scorecard data, and Shivamber, Leon. We Need To Talk About Higher Education. 2026. The median four-year degree’s lifetime gap against the high school path, after costs, taxes, forgone earnings, and the time value of money: about $82,000 at full cost and about $17,500 at the net price families actually pay. A majority of former students, finishers and non-finishers together, earn below the median high school graduate six years after enrolling at about 1,379 of 4,963 institutions across all of higher education, more than a quarter of schools. The premium for the bachelor’s credential itself rising from about 36 to 48 percent, the author’s replication of Caplan’s sheepskin estimate on the General Social Survey, comparing 1972–2012 with 2014–2024. A licensed electrician apprenticeship, for those who finish it (the book’s Table 27-A), and going straight to work, each beating the median four-year degree. Roughly 30 to 50 percent of registered apprentices do not finish. About 30 percent of four-year colleges, 500 of the 1,679 bachelor’s institutions the model scores for lifetime return, would still leave their median graduate behind the high school path at zero cost, because their graduates’ median earnings four years after graduating fall below $53,701, the zero-cost break-even. You can run any school and major at collegeroi.org.
2. Federal Reserve Bank of New York. “The Labor Market for Recent College Graduates.” Federal Reserve Bank of New York, 2026. Accessed August 17, 2026. Underemployment rate for recent college graduates of about 42 percent.
3. Shivamber, Leon. “25 Million in Crisis, and We’re Blaming the Wrong People.” shivamber.com, 2025. Accessed August 17, 2026. Derived from U.S. Bureau of Labor Statistics, Current Population Survey, September 2025, Table A-4 (1.92 million unemployed plus about 22.1 million underemployed degree-holders, 33.7 percent of the 65.59 million employed, about 24.0 million), using the Federal Reserve Bank of New York underemployment rate for all college graduates, the June 2026 reading.
4. College Board. “Trends in College Pricing and Student Aid 2025.” College Board Research, 2025. Accessed September 29, 2026. Federal student aid of $155.1 billion in 2024-25 (Table SA-1), 56 percent of the $275.1 billion in total student aid from all sources. The total also counts $85.1 billion in colleges’ own grants. Federal Student Aid. FY2024 Annual Report. U.S. Department of Education, 2024. Accessed September 29, 2026. Federal Title IV aid of $120.8 billion in fiscal 2024.
5. Oliss, Brendan, Cole McFaul, and Jaret C. Riddick. “The Global Distribution of STEM Graduates: Which Countries Lead the Way?” Center for Security and Emerging Technology (CSET), Georgetown University, November 27, 2023. Accessed September 25, 2026. In 2020, about 20 percent of U.S. graduates earned a STEM degree, against over 40 percent in China. CSET counts STEM as information and communication technology, engineering, construction, manufacturing, natural sciences, mathematics, and statistics. U.S. National Science Foundation. “Higher Education in Science and Engineering.” Science and Engineering Indicators, NSB-2023-32, November 2023. Accessed September 29, 2026. By count, China awarded about 2.0 million science and engineering first university degrees in 2020 and the United States about 900,000. NSF’s science and engineering fields also include psychology and the social sciences, which CSET’s STEM count leaves out.
6. Congressional Research Service. “Amendments to the Higher Education Act Made by P.L. 119-21, the FY2025 Budget Reconciliation Law.” Report R48727, January 8, 2026. Accessed September 25, 2026. The law, enacted July 4, 2025, sets an earnings test for Direct Loan eligibility beginning July 1, 2026. An undergraduate program can lose eligibility if, in two of three years, its completers’ median earnings fall below those of working adults aged 25 to 34 with only a high school diploma, usually in the school’s own state (the whole country if fewer than half the school’s students live in that state). The statute is 20 U.S.C. 1087d(c).