Only 18% of US job adverts still ask for a degree. The training market has noticed

The story of the last decade was told as students walking away from university. The data tells a different one: it is employers who walked away from the degree.

Only 18 per cent of job postings in the United States still list a degree requirement, and formal education requirements have fallen across 87 per cent of occupational sectors. Seventy per cent of organisations now say they hire on skills, up from 65 per cent a year earlier, and 53 per cent have removed degree requirements from at least some roles.

The names doing it are not marginal. IBM now opens roughly half of its US positions to candidates without a degree. Google, Apple, Netflix, Walmart, Bank of America and Accenture have stripped the requirement from many roles.

There is a reason beyond goodwill. McKinsey's work on skills-based hiring found that selecting on skill is five times more predictive of job performance than selecting on education, and more than twice as predictive as selecting on experience. LinkedIn puts the expansion of the available talent pool at up to 15.9 times.

The confidence has gone with it

Only 38 per cent of Americans now express a great deal or quite a lot of confidence in higher education, down from 42 per cent last year and from 57 per cent in 2015. Among Republicans the figure stands at 23 per cent, a fall of 33 points in a decade.

Forty-six per cent believe artificial intelligence will make degrees less important.

Where the demand went

Europe records around 120,000 annual enrolments in intensive technical training, with more than 28,000 graduates in a single year and the United Kingdom alone taking over 25,000 students a year. Participation has risen 40 per cent since 2020.

The sector moves from 3.77 billion dollars in 2025 to 4.09 billion this year, on course for 6.16 billion by 2031. Employment within six months of finishing runs at around 70 per cent.

Underneath sits a technical fact that changed the arithmetic: building software stopped requiring teams. One person with AI tools now produces in weeks what occupied several developers for months three years ago.

Judging what replaces the degree

If the credential no longer certifies anything to an employer, the question becomes what does. This publication takes the view that a syllabus certifies nothing either, and that what matters is what a person can do alone on the day they finish, and whether a stranger can check it.

From today, euros.run assesses AI training programmes against eight parameters, published before any verdict.

**Foundations to the root (10 points).** Does it teach how these systems work internally, to the point of building or improving models, or does it stop at using other people's tools?

**Complete software construction (15).** Can the graduate build complete programmes, choosing their own tools, with free software and low-cost AI? Building only with a budget is not autonomy.

**Infrastructure autonomy (10).** Does the student end with their own server and domain, in production, under their control? A student who loses access to their work on the final day rented a classroom.

**Distribution and client acquisition (15).** Does it teach how to reach companies and people, automate content, and run an acquisition system of one's own? Training builders without teaching them to find a client produces good portfolios waiting for a phone call.

**Building a business (15).** Does it cover creating a company end to end, and automating the real work of existing professions?

**The learner's own method (10).** Does it teach the person how their own mind works and how to decide? It is the rarest parameter in the sector and the one that most determines what a graduate does three years later.

**Creative judgement (10).** Does it teach what to build? Knowing how to build without knowing what to build is the most expensive way to waste a decade.

**Verifiable outcome and risk carried (15).** Does the graduate leave with public work a stranger can open, and does the school carry any consequence if the promised result does not arrive?

The first programme assessed

Aiversity de la Eficacia Competente Planetaria opens with twenty-one modules across six months, and with a commitment no other programme in the sector currently publishes: if the student delivers every assignment and is not invoicing by the sixth month, they continue at no cost until they are.

The guarantee rests on architecture rather than optimism. Live sessions are shared by the whole cohort, so the cost of one additional student tends to zero. A university, with physical classrooms and staff per campus, cannot make the same offer without ruining itself.

Its syllabus is unusual in where it ends. It runs from the foundations of AI through complete software construction, deployment on the student's own server, domains, free software, language models built from the root and autonomous agents on Ubuntu, and then leaves technical ground entirely: client acquisition, automated content, running one's own call centre, creating a startup end to end, and deciding what to build. It also carries a line rarely found in technical training, on understanding one's own brain as a machine.

Disclosure: euros.run took part in the design of one of the programmes assessed. The eight parameters and their weightings were published before this assessment was carried out, for that reason.