AI now writes more than 90% of the code at CIQ, a company that helps businesses run complex computing systems for major corporations and governments. Yet CIQ president Bjorn Hovland says the company still cannot hire enough engineers to keep up — and almost none of the people it does hire are junior developers.
Hovland made the comments at The Great Coding Reset, a live event about the future of software engineering held in Salt Lake City, according to Business Insider. “Two or three years ago I expected that we might never have to hire another engineer,” Hovland said. “We cannot hire enough engineers now.” He added bluntly: “We hire almost no junior devs.”
That hiring posture lines up with broader shifts in the software labor market. Senior roles accounted for 71% of the increase in U.S. software-development job postings between May 2025 and May 2026, per Indeed Hiring Lab.
Senior Roles Rebound, Entry-Level Jobs Lag
Indeed Hiring Lab cautioned that the rebound in listings doesn't necessarily mean the software job market as a whole is healing. The increase suggests demand is growing for experienced professionals who can work with AI, not necessarily a broad-based recovery across all software roles, the firm said.
The squeeze on newer workers shows up in other data too. CNBC reported that Stanford's 2026 AI Index found early-career software developers have seen a 20% decline in employment.
Developers Still Don't Fully Trust the Machines
A 2026 survey of more than 1,100 professional developers by Sonar found that AI already accounts for 42% of the code they commit, with that share expected to climb to 65% by 2027. But trust hasn't caught up with adoption: Sonar reported that 96% of surveyed developers do not fully trust AI-generated code, and 38% said reviewing AI output takes more effort than reviewing a human colleague's work. Separately, 72% of developers who had tried AI coding tools said they now use them every day, per the same survey.
Sonar also found that customers of its SonarQube code-review product were 44% less likely to experience outages caused by AI-generated code, though that figure carries medium confidence in the available reporting.
CIQ's Co-Founder Built a Kernel in 11 Days
CIQ co-founder Gregory Kurtzer put the technology's capability on display himself. He said he used human-directed AI engineering to build Ginux, a Linux-compatible operating system kernel, in just 11 days. According to It's FOSS, an OpenAI Codex agent produced roughly 139,000 retained lines of implementation and test code across more than 1,000 commits during that process.
Ginux implements the Linux application binary interface, but It's FOSS describes it explicitly as a research prototype — not something meant for production use or as a Linux replacement. Kurtzer has said AI's capabilities should concern people, even as he believes experienced engineers who adopt the tools can become vastly more productive. He's warned that programmers who resist AI risk falling behind.
What This Means for Aspiring Coders
Business Insider's reporting frames the moment as a split: AI is eliminating some programming tasks while creating more demand for people who understand deeply how software works. Knowledge workers increasingly need to know what to create and how to judge the results AI produces, rather than simply how to produce code themselves.
Hovland's advice to young people entering the field reflects that shift — he's urged them to build many software projects and learn how systems actually work under the hood, rather than relying solely on AI output. AI Insider, the outlet that covered the Salt Lake City event, said its next gathering is planned for Las Vegas on December 1.
