For most of the last century, the corporation was also a school. GE ran Crotonville. The banks, the insurers, the manufacturers, and the phone company all took raw graduates and spent years turning them into workers, on the assumption that training was a cost of doing business, and the investment in an employee was worth the cost. That assumption is gone. The training function and the need didn’t disappear so much as it was handed off — to community colleges, to trades programs, and above all to higher education. What nobody arranged was the handoff of the part that actually made people employable. The classroom got the theory. The apprenticeship was left behind.
Every employer survey in circulation seems to describe the same absence. Julie Peters, who leads US university relations at PwC, told AACSB that her firm’s new joiners arrive looking to “immediately get the right answer,” when the skill that matters is understanding the process that produces one. Deloitte’s 2026 human capital research puts a number on the stakes: organizations taking a technology-first approach to AI are 1.6 times more likely to miss their expected returns than those investing in the human edge, because, as the report puts it, technology is replicable and people are not. The World Economic Forum’s Davos analysis says the same thing from the other direction — creativity, adaptability, and judgment are the hardest capabilities to automate and the most prized by employers, yet they remain “invisible in the job market” for lack of any way to measure them. Wages for AI-technical roles have climbed 27% since 2019 precisely because they are in demand. The human skills that determine whether someone succeeds in a job are not.
And while the findings are common, the gap is not in the student’s acquisition of knowledge, but in the necessary practice required to attain soft skills. You do not learn to read a room, absorb feedback without flinching, or defend a point of view in a meeting by taking a course in it. Times Higher Education’s reporting on Australia’s Employer Satisfaction Survey made the point almost by accident — graduates were rated meaningfully better at collaboration if they had done at least some of their study on campus, because the online-dominant cohort had fewer chances to practice getting along with other people. Soft skills are behaviors, and behaviors are built by repetition in the setting where they’re used. That setting is a workplace. Which means the reality underneath the entire “work-ready” debate is not that students need to know more. They need to have an opportunity to practice more.
Institutions can position themselves better by acknowledging that some work experience beats none, almost regardless of what the work is. A student who spends a summer discovering she hates the job she thought she wanted has learned something a syllabus cannot teach her — and has narrowed her career faster than four years of electives will. Trying and failing at a few jobs is not a detour from career formation. It is career formation. The graduate who has never tested a preference against reality is the one most likely to be, in University World News’s phrase for the global mismatch, employed below their qualification level and unsure why.
A handful of institutions have stopped treating this as a co-curricular nicety and started building it into the structure of the degree. Curry College now offers what it plainly calls a job guarantee: complete the Curry Commitment — a four-year sequence built on NACE’s career-readiness competencies, anchored by a required internship and a 2.8 GPA — and if you don’t land a job or a graduate placement within six months of walking, the college pays up to a year of your federal loans, or places you in a paid internship, or waives tuition for six graduate credits. Tecnológico de Monterrey has gone further into the curriculum itself with Challenge-Based Learning, in which students spend Week i (or Tec Weeks) and Semester i solving real problems posed by real companies and communities rather than reading about them, developing the critical thinking, teamwork, and leadership that employers say they can’t find. The University of Manchester’s vice chancellor, Duncan Ivison, announced this year that the university would aim to give every student industry experience during their degree, stating that in a market being reshaped by AI, the university has “no choice.” Even the faculty are being sent back to work — Inside Higher Ed profiled professors like Utah Valley University’s Dan Hatch, now on his seventh summer externship, returning to campus with a live sense of how students actually break into the field.
These are the institutions doing what the sector at large needs to embrace, evolving into what students and employers now need, rather than defending a model designed for a market that no longer exists. Every one of them that wants to provide a solution has a barrier in the supply side. That is something that higher education cannot solve by itself. Alchemy, a firm which assists placing healthcare students into the clinical internships their degrees require, puts the shortfall in stark terms: over eight million American college students look for internships each year, and only one in three secures a quality one. The gap is widening, not closing, as AI erodes the entry-level tasks that used to justify hiring an intern in the first place. Curry can promise a placement; Manchester can aspire to universal experience; but a promise is not a position, and the positions have to come from somewhere.
Some of the response has been to subsidize the demand side rather than build the supply. Dartmouth’s Center for Career Design funded 336 unpaid or underpaid internships this year — 72.5% of applicants, at an average of $4,350 — precisely so that a student’s ability to take a formative role isn’t gated by whether their family can float them a summer. That is real equity work, and worth imitating. But Dartmouth is careful to say what its own money cannot do: “Funding is not a substitute for compensation.” Nor is it a substitute for the internship itself. You can pay a student to take an opportunity that doesn’t exist only as long as the opportunities exist to take. Alchemy’s model attacks the actual constraint — absorbing payroll, taxes, and HR liability as an employer of record so that hosting an intern stops being an ROI problem for the employer — because the firm understands that the shortage is manufactured by cost and friction, not by a lack of willing students.
So the choice in front of the sector is narrower than the endless skills-gap commentary makes it sound. Higher education inherited corporate America’s training obligation without inheriting its training budget or its factory floor. We can keep pretending the classroom alone can discharge that debt, and keep graduating students who know a great deal and have not had a chance to put it into practice. Or we can be honest that the missing half of the education is work itself — and that if industry is no longer going to run the training programs, it owes higher ed the internships instead. The demand is already here, eight million strong. The institutions willing to build for it are moving. The question is whether employers will supply the one thing no college can manufacture on its own: a place to practice.
