Should My Teen Still Study Computer Science?

This site argues that the trades are undersold. That is not the same as arguing that a computing degree is a mistake, and a parent asking this question deserves the fair version rather than the convenient one.

So here is the fair version, including the parts where the honest answer is that we do not know.

What can anyone actually say about this job market?

Less than the headlines imply. Nobody, including BLS, can tell you what the entry level market for computing graduates looks like in 2032. Projections are models with stated assumptions, and models are at their weakest exactly where a technology is changing fastest.

What can be said is narrower and more useful. There are federal projections for specific occupations, published with a stated decade. There is a small amount of primary research on where generative AI overlaps with occupational work activities. And there is programme-level data on what graduates of a particular course at a particular school actually earned. That third one is the closest thing to a real answer for your family, and almost nobody uses it.

Which numbers do we have, and what do they cover?

Two IT occupations sit close enough to a computing degree to be worth quoting, and both are support and administration rather than software engineering. That distinction is not a technicality: it is the difference between running systems and building them.

BLS projects computer support specialists to decline by 3 percent from 2024 to 2034, from employment of 882,300 in 2024, while still producing about 50,500 openings a year. The median wage was $61,550 in May 2024.

BLS projects network and computer systems administrators to decline by 4 percent over the same decade, from 331,500 in 2024, with about 14,300 openings a year and a median wage of $96,800 in May 2024. Typical entry education for that one is a bachelor’s degree.

Read those two together and you get something more interesting than a verdict: modest projected decline, substantial ongoing openings driven by people leaving, and two medians from the same May 2024 survey that are nowhere near each other. Career outcomes inside computing vary enormously, which is exactly what you would expect and exactly what a single headline about the field cannot capture.

What this page is deliberately not telling you

There is no software developer figure on this page. That is on purpose. This site prints a number only when it has been checked against the primary source with its reference period recorded, and a software developer projection is not in our verified set right now.

BLS publishes one. It is on the same site as everything linked above, under computer and information technology occupations. Go and read it directly, note the reference period printed at the top of the page, and note whether the figure quoted at you elsewhere matches it. Doing that once with your teenager is worth more than any summary, including this one, because the skill survives the next round of headlines.

It is also worth saying what the AI research does and does not cover here. In the 2025 Microsoft Research study Working with AI, computer user support specialists score 0.334 on a 0 to 1 applicability scale, well above any of the trades this site covers. The authors are explicit that this measures overlap between an occupation’s work activities and what a chatbot was observed doing in 2024 logs, not job loss, and they caution that downstream effects of new technology are hard to predict. A higher score is a reason to ask better questions, not a forecast. The full picture is on which jobs are actually safe from AI.

Was “learn to code” oversold?

Yes, in one specific way. It was sold as a guarantee, and the guarantee was the false part, not the subject. A field can be genuinely valuable, intellectually serious and well paid, and still fail to promise every entrant a particular outcome.

That is worth saying carefully, because it is the same error the trades are now at risk of committing. “Learn to code and you will always have work” and “go into the trades and you will always have work” are the same sentence with different nouns, and the second is being repeated as confidently in 2026 as the first was ten years ago. Anything sold as a guarantee should raise the same eyebrow regardless of which direction it points.

It is also the same shape as “get a degree and you will be fine,” a claim many parents reading this were given and found to be true for some people and expensive for others. The lesson is not that degrees fail. It is that a general claim about a route tells you very little about a specific person on a specific programme.

The four questions that work on any path

This is the framework worth keeping, and it applies equally to a computer science degree, an electrical apprenticeship and a fifteen week certificate.

How many openings a year? Not growth percent, openings. BLS publishes both, and openings include replacing people who leave, which is where most hiring actually happens. If that distinction is new, start with how to read a career growth statistic.

What does entry cost, in money and in years? For a degree, use College Scorecard at programme level rather than the university’s own brochure, and work out what is left after aid. The maximum Federal Pell Grant for the 2026-27 award year is $7,395, and Pell is a grant, not a loan. For an apprenticeship, the entry cost is usually tools and time rather than tuition, and what tools actually cost in year one puts real numbers on the first side of that.

What do the first two years look like day to day? This is the question that predicts whether someone finishes. A teenager who loves building things and hates lectures will have a hard first two years in a degree, and a teenager who loves the theory will find a first year apprenticeship frustrating in a different way.

What are the exit options? If the plan does not work out at 22, what does this person have? A degree, a licence, a set of hours logged toward a licence, a portfolio. Any of those is an exit. None of them is guaranteed.

What this looks like in practice

Rashad’s daughter is a strong maths student who has been writing code since she was thirteen and has been told, repeatedly and by adults who mean well, that AI has closed that door. She still wants to do it. Rashad does not want to fund a mistake, and he does not want to talk her out of the thing she is best at either.

He runs the four questions with her instead of arguing. On cost, they pull up the specific programmes on College Scorecard and work out the real net price rather than the sticker. On the first two years, she visits and sits in on a class. On exits, they list what she would hold after two years if she stopped, and after four if she finished. On openings, they read the BLS pages themselves, at the source, including the declining categories above.

They end up choosing the degree, at a cheaper school than the one she originally wanted, with a plan to look at electrical engineering as a nearby option if the software side disappoints her. The figures and the outcome here are illustrative. What is not illustrative is the method: nothing in it required predicting the future, and all of it was available free.

The short version

Do not let anyone, including this site, turn your child’s aptitude into a bet on a trend. If she wants to study computing, the questions are the same ones you would ask about any other five figure purchase: what does it cost, what do people who did it actually end up doing, and what is left if it does not work.

If your teenager is on the other side of this conversation entirely, start with my teen does not want to go to college.