Eric Schmidt once ran Google. He shaped one of the most consequential technology companies of the 21st century. Now he stands before developers and confesses something personal. He is in mourning.
“I’m in mourning,” Schmidt said, “because I started as a programmer when I was basically 13 or 14, and that is essentially over.” The former CEO delivered the remark in remarks that have circulated widely since. He added that his entire identity as a computer scientist had ended in one lifetime. Something that should have passed to children or grandchildren instead vanished in his own time. The Economic Times reported the comments.
His message carries weight. Schmidt does not sell software tools. He has advised the Pentagon on artificial intelligence and backed multiple AI coding ventures. When he tells programmers to stop writing code the old way, the industry listens. Or at least it should.
“If you’re writing code in any traditional way: stop. It’s over,” Schmidt declared. The line lands like a verdict. And the evidence he offers is practical rather than theoretical.
Consider the programmer at one of Schmidt’s startups. Every evening at 7 p.m. this engineer writes a specification for what needs to be built. He creates a test function that will judge the output. Then he hits run and leaves for dinner with his wife. The AI systems work through the night. By 4 a.m. they finish. The engineer reviews the results over breakfast. “This stuff would’ve taken me six months and 10 programmers at Google,” Schmidt recalled. “This poor guy’s sleeping.” The Times of India detailed the account.
That single story captures the shift. The best programmers no longer write most of the code. They define success. They set evaluation criteria. They manage fleets of AI agents. The machines handle implementation, iteration, and much of the debugging. Humans direct. They orchestrate.
Schmidt paints the current state of the art in vivid terms. A programmer arrives at the office. Social creatures that they are, they gather 10 Claude instances or 10 Gemini instances. They assign objective functions. They watch the code appear. Then they go to lunch, confident the tasks are long enough to keep the systems busy. Later they set overnight objectives and head home to family. The work continues without them.
But. The transformation did not arrive gradually. Schmidt notes it accelerated sharply around October 2025. Six months later the gap between old practices and new ones had widened dramatically. Managers who still see engineers typing line by line should ask a direct question. “Why are you still writing code the way you did it six months ago?”
The implications stretch beyond individual coders. Entire software organizations face pressure to adapt or fall behind. Companies that treat AI as an optional productivity tool miss the point. The new advantage lies in specification quality and evaluation precision. Define the problem poorly and the AI will optimize toward the wrong target with impressive speed. Get the criteria right and the systems invent solutions at a pace no human team can match.
Schmidt has made bolder predictions elsewhere. In April 2025 he told an audience at an AI and biotechnology summit that within one year the vast majority of programmers would be replaced by AI programmers. SAN reported those remarks. He spoke of autonomous agents, self-improving systems, and timelines that compress decades of progress into a handful of years. Three to five years for AGI that matches the best humans. Six years for superintelligence that exceeds all of us combined.
Such forecasts invite skepticism. Some developers point out that AI still hallucinates, struggles with novel architectures, and requires human oversight for production systems. Others argue that the real work of software engineering has always involved more than typing syntax. Requirements gathering, system design, stakeholder alignment, and long-term maintenance remain stubbornly human domains. Yet even skeptics acknowledge the pace. Models already write 10 to 20 percent of code at leading labs. That share grows monthly.
Recent coverage reinforces the momentum. As recently as this week, outlets captured Schmidt repeating the core warning while noting fresh examples of overnight AI builds. The personal dimension, his mourning for a craft he began as a boy, adds emotional texture to what could otherwise read as another round of automation hype.
And the market has responded. Schmidt has invested in AI coding startups including Magic, which raised hundreds of millions, and Augment, a GitHub Copilot competitor that emerged from stealth with substantial funding. These companies pursue exactly the workflow he describes. Agents that plan, write, test, and deploy with minimal human intervention. The capital flowing into the sector suggests investors share his conviction.
Yet resistance appears too. At university commencements this spring, Schmidt and other speakers mentioning AI drew boos from graduates. Students expressed anxiety about job prospects, economic concentration, and a future that seems already written. The backlash reveals a tension. Technology leaders celebrate capability and speed. Younger workers see displacement and uncertainty. Both perspectives contain truth.
So what should a working software engineer do in 2026? Schmidt’s advice is blunt. Stop writing code the traditional way. Learn to manage agents. Master the art of specification and evaluation. Treat the AI systems as a team that never sleeps and scales with hardware. Those who excel at directing this new workforce will widen their impact tenfold. Those who cling to the old identity may find their skills commoditized faster than expected.
The change echoes earlier industrial shifts. Agricultural workers moved from farms to factories over decades. This transition compresses into years. The skills that defined a generation of programmers, the joy of elegant algorithms, the satisfaction of a cleanly written module, now share space with prompt engineering, agent orchestration, and criteria design. Some mourn that loss. Others see opportunity in the expanded frontier.
Schmidt himself embodies the pivot. Having built his career on search algorithms and large-scale systems, he now chairs AI advisory efforts and funds the very tools that obsolete parts of his original craft. His mourning feels genuine. It also feels strategic. By voicing the end of an era he helps prepare the industry for what follows.
The coming months will test these claims. If AI agents begin handling complex, multi-step software projects with reliable results, the “vast majority” prediction may prove closer than critics expect. If integration challenges, reliability gaps, or regulatory hurdles slow adoption, the timeline will stretch. Either way the direction is set. Traditional coding is not disappearing tomorrow. Its dominance has already ended.
Programmers who adapt will define the specifications that shape tomorrow’s systems. They will evaluate outputs that once took teams months. And they may, like Schmidt’s startup engineer, enjoy more dinners and fuller nights of sleep while their AI colleagues labor until dawn. The craft evolves. The identity shifts. The work continues. Just not with the same tools or the same daily rhythms that defined it for half a century.
WebProNews is an iEntry Publication