Engineering as a career option in the age of AI: Still secure, but no longer simple

The software engineer is not being declared dead. Not yet. But the job is being rewritten with unusual confidence these days, and the most striking thing is who is doing the rewriting.Boris Cherny, the creator of Claude Code at Anthropic, has said, “Today coding is practically solved for me, and I think it'll be the case for everyone regardless of domain.” By itself, that could sound like the optimism of a man standing very close to his own product. Fair enough. Builders often believe their tools will change the world. Sometimes they are right. Often, they are merely early. But Cherny went further.

He predicted that “the title ‘software engineer’” will start to “go away”. That is no longer a statement about speed, productivity, or cleaner code. It is about professional identity. When the person who helped build the tool says the tool may eat into the name of the job itself, the claim cannot be brushed aside as demo-day bravado.And this is not just one builder getting carried away.The pattern is visible at the level of institutional leadership.

Anthropic’s CFO has said that “90 plus percent” of the company’s code is now written by Claude Code, and that as a result “everyone kind of becomes a manager.” That casual phrase carries a fairly large bomb inside it. If true, it means the engineer’s work is moving, at least inside some AI-first companies, from writing code to directing, checking, correcting, and owning what the machine produces.

OpenAI president Greg Brockman has made a similar claim in public remarks, saying that AI is now writing roughly 80% of OpenAI’s code.

These are not sleepy firms retrofitting AI into old workflows. These are among the most consequential AI laboratories in the world. And by their own accounts, their engineering teams are already spending a large part of their time reviewing and steering AI output rather than producing every line from scratch.

That does not make engineers redundant. But it does make the old, comforting picture of the engineer — alone with code, turning logic into product line by line — look increasingly dated.The people building the hardware under this revolution are saying much the same thing, though in a different register. Nvidia’s Jensen Huang has said: “Nothing would give me more joy than if none of our engineers were coding at all, and they were just purely solving undiscovered problems.” That sentence complicates the panic. Huang is not imagining a world without engineers. He is imagining a world where engineers no longer spend their best hours on what machines can now do quickly, and in a more cost effective way.

In that version of the future, the engineer does not disappear, he is pushed towards problem framing, systems thinking, architecture, taste, accountability, and the human ability to know what is worth building in the first place.The anxiety these comments have generated, has numbers attached to it. A Tech Insider layoff tracker — which compiles company-announced workforce reductions and should be read as a curated estimate, not official labour-market data — recorded 45,363 confirmed tech layoffs worldwide through early March 2026.

Of these, approximately 9,238, or 20.4%, were explicitly attributed to AI and automation by the companies themselves.

The same tracker notes that this is a sharp jump from 2025, when AI was cited in fewer than 8% of layoff announcements. The numbers are not the whole story, they rarely are. But they reveal something more unsettling.AI has moved to the brutal side of corporate arithmetic, where companies are beginning to admit, in varying degrees of directness, that some work can now be thinned out because machines can do enough of it to change the hiring equation. This is where the question facing engineering students becomes sharper, and less comforting: If machines can now do enough of the old engineering work to change the hiring equation, what kind of engineer will still be worth hiring tomorrow?

Engineering survives, the old engineer may not

The first answer is that engineering is not being erased in one clean stroke.

That would be too simple. Too convenient. The more difficult truth is that the centre of the job is moving. If routine coding can be done faster, cheaper and at scale, value shifts elsewhere — to the engineer who can frame the problem, direct the tool, test the answer, understand the domain and finally own the decision. Because someone still has to own it.The World Economic Forum’s Future of Jobs Report 2025, based on inputs from more than 1,000 employers representing over 14 million workers across 55 economies, projects that broader labour-market shifts could displace 92 million jobs globally by 2030, while creating 170 million new ones, leaving a net gain of 78 million roles.

Among the fastest-growing categories are AI and machine learning specialists, fintech engineers and big data specialists.That sounds reassuring, but not entirely.For engineering students, the question is not whether some new technology jobs will be created. They will. The harder question is whether the old route into engineering still works when the work that once trained juniors is increasingly being handed to machines.

GitHub’s research on Copilot, now used by 90% of Fortune 100 companies and 20 million developers globally, found that developers using AI assistance completed tasks 55% faster.

By 2025, approximately 41% of all code written globally was AI-generated or AI-assisted. Useful? Certainly. Harmless? Not quite.This does not end the need for engineers. But it does change the starting line for sure. The premium now moves towards those who can direct the machine, question its output, catch the elegant error, understand the domain, and know when a technically correct answer is still a bad answer. The money is beginning to follow that distinction. PwC’s 2025 Global AI Jobs Barometer, based on nearly a billion job postings across six continents, found that workers with demonstrable AI skills earn, on average, 25% more than peers in the same roles without those skills.

Industries with deeper AI adoption are also growing revenue at four times the rate of less AI-intensive sectors.For AI engineers, the signal is sharper. The U.S. Bureau of Labour Statistics projects 23% growth in computer and information research scientist roles between 2023 and 2033. Average AI engineer salaries in the US have reportedly reached $206,000, a $50,000 jump from the previous year. The real disruption may not be the disappearance of engineering, but the thinning out of the rough, repetitive, badly paid early work through which engineers once learnt to become useful.

If that rung weakens, the profession may still have a top, but far fewer people will know how to climb towards it.

The new ladder begins with judgement

If the lower rung is weakening, the next question is not some academic afterthought, it is the question engineering colleges and employers have not yet answered properly: what, exactly, replaces the early work through which young engineers once learnt their craft? It cannot simply be more syntax, or faster typing.

Jensen Huang’s comment becomes useful here because it points to the direction of travel. When he says he would prefer engineers not to code at all, he is not asking them to become spectators in their own profession.

He is saying that the valuable work is moving towards problem discovery, system design, trade-offs, risk, context and responsibility.This is where upskilling becomes the central challenge. Gartner said in October 2024 that generative AI would require 80% of the engineering workforce to upskill through 2027, as new roles emerge in software engineering and operations.

It also argued that the future would need AI engineers who combine software engineering, data science and AI or machine-learning skills.That does not mean every engineering student must become a frontier-model researcher, which is both unrealistic and unnecessary. However, it does mean that the useful engineer of the next decade will need enough AI literacy to work with models, enough software depth to test their output, enough domain understanding to build systems that matter, and enough skepticism to remember that fluent code is not the same thing as correct code.

India’s engineers need more than the degree now

For India’s budding engineers, the point is not that engineering has suddenly become a bad bet. It has not. AICTE data shows B.Tech enrolment touching an eight-year high of 12.53 lakh in 2024-25, up 67% from 2017-18, while the vacancy rate in engineering colleges fell to 16.36%. That is not a picture of students fleeing the discipline. It is a picture of families still investing in it, and reasonably so. But the degree now needs stronger company.The World Economic Forum’s Future of Jobs Report 2025 found that 63% of employers globally see skills gaps as the main barrier to business transformation, while 85% plan to prioritise upskilling. For Indian students, that turns the college-industry gap from an old complaint into the central employment question. A B.Tech may still open the first door, especially in a country where engineering remains tied to mobility, respectability and family ambition, but it will not automatically protect a graduate from a market where routine work is being compressed by AI.Stanford research adds the sharper warning: recent graduates in AI-exposed roles have seen a 16% employment decline, while experienced workers have remained more stable. The clue is in that contrast. AI is not only changing the work; it is changing the first few years through which young engineers once learnt the work. So the safer bet is no longer engineering in the generic sense. It is engineering with fundamentals, AI fluency, domain depth, and the ability to test what machines produce.

The old promise was simple: Learn to code, get placed. The new one is narrower, and less forgiving: Learn to think with machines, or be measured against them.