How to Grow as a Developer in 2026

— I won't. First, a different question: are you sure the rung you're aiming at still exists at all?

— On the portal, sure. I mean something else. You're carrying around a picture where junior, mid-level, senior are like grades in school: a description of how a person matures in the profession. Grow into it, move up. And since you haven't moved up, something must be missing in you. But that ladder isn't about the person at all. The dual technical ladder — the very idea that an engineer can grow in pay without becoming a manager — was invented by American research organizations somewhere around the 1950s, and spread widely by the 1970s. And it wasn't invented to describe how an engineer grows. It was invented so that a strong engineer wouldn't go into management just for the money.

— Exactly. From birth it was an employer's instrument, not a map of professional growth. That matters, and not as a historical anecdote. It means the number of rungs is an adjustable parameter. They turn it like a dial, based on what each rung costs the company.

— Always — at a certain price of money. Look at the dates. The Fed's rate sat near zero, in the 0–0.25% range, from March 2020. The first hike was March 2022, and by July 2023 it had reached 5.25–5.5%. That's the fastest tightening cycle since Volcker. And hiring a beginner is an investment that pays back in about two years. At a zero rate, 2 years of waiting costs almost nothing. At 5.5% it costs a lot.

— The rate is only one of the suspects, and I'll come back to that. But the arithmetic changed somewhere else too. Look at Meta: headcount grew from 44,942 people at the end of 2019 to a peak of 87,314 by September 2022, and then two rounds of layoffs of roughly 21,000 in total — around 11,000 in November 2022 and around 10,000 announced in the spring of 2023. Revenue per employee for 2023 came out around $2 million. Which, by the way, is not a figure from the report, it's a division you and I are doing ourselves. And that was the moment when the idea that more people means a bigger company broke in public.

— It's been measured. By SignalFire's 2026 report, new-graduate hiring at large tech companies is roughly 65% below the 2019 level, and at early-stage startups roughly 76% below. In their own report a year earlier, graduates made up 7% of hiring at big tech companies. That isn't a feeling from a forum, that's a number.

— Somebody sat down and decided — just not about that rung. In the autumn of 2024 Andy Jassy wrote a letter to Amazon employees where, among other things, he asked every organization to raise the ratio of individual contributors to managers by at least 15% by the end of the first quarter of 2025, and he explained it plainly: fewer managers removes layers. I'm paraphrasing, not quoting. They were cutting from the top. But the trick is that they were cutting from the top and from the bottom — with the same dial.

— It does come back. Indeed's software development job postings index for the US, where February 2020 is set at 100, fell to a low of 61.1 in May 2025 and rose to 74.4 by August 2026. Demand has come back about a quarter of the way. And graduate hiring stayed at the same minus 65%. Those are two different events, and the second doesn't follow from the first.

— There's a sociologist, Ofer Sharone, who compared unemployed white-collar workers in the States and in Israel. In America people look for work as though they're selling a personality: are you a fit, is there chemistry. In Israel — as though they're selling compliance with a spec. And the Americans blame themselves while the Israelis blame the system, at the same unemployment rate. That's qualitative research, interviews and observation, not a measurement of what share of people blame themselves — I'm giving it to you as his explanation, not as a statistic. But it's a good explanation.

— Coming back to it, and here's the honest answer: we don't know. There are four candidates. The end of zero rates. A tax rule — starting with tax years beginning in 2022, American companies had to spread the deduction of software development costs over 5 years, and only in the summer of 2025 was the immediate deduction restored for domestic spending. Over-hiring in 2020–2021. And AI. So: the rule was restored — the juniors didn't come back. Rates started coming down — they didn't come back.

— Brynjolfsson's Stanford group measures exactly that, and they put it more carefully than the headlines do. In their work, employment of 22-to-25-year-olds in the occupations most exposed to AI is roughly 19% below what it would have been had it tracked the less exposed ones. And in February 2026 the authors themselves wrote that with controls for firm and time the decline is statistically significant only after 2024, and that part of the timing is explained by something other than AI. About the tax rule as a cause of layoffs I also have to say: that's a widespread interpretation, not demonstrated causality.

— Worse. Three of them have already been partly reversed with no result. I don't know what removed the rung. And that's not coyness: if the cause isn't established, then the advice to just wait it out is backed by nothing. And it's the single most popular piece of advice in this situation.

— It has this to do with it: the speed of the ladder is first of all a parameter of the ladder, not of you. Revelio Labs measured promotion intensity inside the S&P 500: from roughly 1–3 promotions per 100 employees a year at the bottom group of companies to 10–12 at the top. The probability of getting promoted within 12 months — 2.8% at a low-intensity company, 4.2% at a medium one, 6.2% at a high one.

— For the same person. You spent 3 years proving something to a place where there simply aren't many promotions. Checking that would have cost you about an hour.