The warning sign came from inside Amazon Web Services. In a post that vanished within hours but was widely screenshotted, one engineer wrote that after AWS automated roughly 90% of its infrastructure, his entire DevOps team was deemed redundant. The details spread quickly: AI agents were being used not only to auto-fix Terraform and scale Kubernetes clusters, but even to negotiate cloud discounts with providers.
For years AWS sold automation tools to its customers. Now it was turning them inward, cutting human operators out of the loop. For DevOps engineers, the story read like a parable: the very software they had used to streamline workloads was beginning to consume their jobs. And if the world’s largest cloud provider believed AI could handle deployments, incident response, and postmortems better than people, the implications for every other engineering team were clear.
The episode crystallized what many in Silicon Valley had already begun whispering. Junior engineers, once the engine room of software teams, are rapidly becoming collateral damage in the shift toward “vibe coding”—a shorthand for letting AI generate, review, and even refactor large portions of code. Companies that once hired armies of entry-level coders are now leaning on a handful of senior engineers to orchestrate AI systems and oversee their output.
Salaries Rising at the Top, Vanishing at the Bottom
In January, the consensus among investors and executives was that the era of the $400,000 software engineer was over. By October, the opposite had happened. Senior engineers were making $600,000 or more in total compensation, and companies complained they could not hire enough of them.
The exact opposite is happening in the market rn btw. Staff salaries shooting to the moon, junior positions more or less gone https://t.co/2SNuOWck86
— Nate Berkopec (@nateberkopec) October 13, 2025
The numbers reflect a fundamental reshaping of the tech labor market. Senior engineers with AI and machine learning expertise now command average salaries of about $245,000 a year, far above the typical entry-level package. In some cases, the premium is staggering: a Staff Engineer at Intuit focused on AI earns nearly $900,000, compared to $515,000 for a peer without those skills. At companies like Snap and Cruise, senior AI engineers routinely earn more than half a million annually.
Meanwhile, the entry-level path is drying up. Data from the Federal Reserve Bank of New York shows that recent graduates in computer science face an unemployment rate of 6.1%, and computer engineering majors 7.5%. That is more than double the rate for biology graduates.
What looked a decade ago like a lottery ticket, the promise that learning to code guaranteed a high-paying job has become one of the riskiest bets for new graduates.
The AI Aristocracy
The consequence is a widening gap: an engineering aristocracy of senior developers who can command extraordinary salaries, and a struggling mass of juniors who cannot find work at all. Staff engineers, once reliant on juniors for routine tasks, now handle more themselves. AI tools eliminate the need for boilerplate coding, but reviewing and refactoring AI-generated output still requires the kind of deep experience only senior developers have.
Even quality assurance, once an entry-level foothold, is increasingly overseen by senior engineers who are the only ones able to understand the architectural nuances of AI-generated code and spot where the bugs hide.
The hiring market has narrowed to a precise target: engineers who can reskill quickly, orchestrate AI systems rather than resist them, catch security flaws that AI tools miss, and design processes for entire teams. But they must also fall just short of having the initiative to start companies of their own. That combination is scarce, which is one reason compensation for those who fit it keeps rising.
Broken Promises
The harshest impact is felt by the generation of students raised on the “learn to code” gospel. Starting in the early 2010s, companies like Google and Microsoft lobbied heavily for schools to prioritize computer science. Code.org, founded in 2013, became a household name through its “Hour of Code” campaign, fronted by tech celebrities from Bill Gates to Mark Zuckerberg. By last year, the number of undergraduates majoring in computer science had tripled since 2012, reaching 170,000.
For a while, the pipeline worked. Graduates from top schools went to work at big tech firms with salaries far higher than their peers in other disciplines. But the pandemic years brought overhiring and subsequent layoffs. Companies ramped up use of H-1B visas to bring in foreign workers, who could be paid less than U.S. graduates. Now, with AI cutting into entry-level demand, thousands of students are finding that the golden ticket has expired.
Natasha Singer, a New York Times reporter who has tracked Silicon Valley’s influence on education for more than a decade, describes the shift as a breakdown in the promises tech leaders made. In her reporting, graduates describe the experience of applying to hundreds of jobs with little response as “frustrating” and “soul-crushing.” One recent computer science graduate, Nathan Spencer, who had excelled academically at Ohio State University, found himself applying to fast-food jobs after months of unanswered applications.
His worry is not just personal. “How are you going to have senior developers if you get rid of all the junior developers?” he asked. Without a training ground, the profession risks becoming a bottleneck.
The Oversight Role
Those senior engineers who have adapted describe their work differently than they did a decade ago. The role is no longer about writing CRUD code but about validating AI outputs, checking edge cases, assessing security risks, and making architectural decisions. One described it as moving from building blocks to oversight—deciding when to trust the AI and when to override it.
Companies increasingly see this oversight as an irreplaceable function. Hiring data shows that demand is strongest for engineers with at least two years of experience, while pure new-grad roles are scarce. Some executives, including AWS’s CEO, have argued publicly that eliminating entry-level roles is short-sighted: “That’s one of the dumbest things I’ve ever heard… how’s that going to work when ten years in the future you have no one that has learned anything?” he said earlier this year.
But the immediate reality is that companies see experienced engineers as the safest bet, especially those who can leverage AI tools effectively. A statement recently predicted that as routine coding is automated, developers will command pay “more like lawyers” and in some corners of the industry, that future is already here.
