UpTrajectory Review

Linus Torvalds, the creator and still-dominant gatekeeper of the Linux kernel, has weighed in on one of the loudest debates in software right now: whether AI-assisted 'vibe coding' — writing code by describing what you want to a large language model and accepting what comes back — belongs anywhere near serious development work. His answer, as reported by ZDNet's Steven Vaughan-Nichols, is characteristically blunt and two-sided. Torvalds calls vibe coding a wonderful way to find joy in programming, which is a more generous framing than many senior engineers have offered. But he draws a hard line: it is not how the Linux kernel gets built. That distinction — joy versus infrastructure — is the real story here, and it carries weight far beyond the kernel community.

For small-business operators who run on Linux, whether directly on servers, embedded in appliances, or underneath cloud services, this matters because kernel integrity is the foundation your stack quietly depends on. Torvalds is not anti-AI across the board; he explicitly endorses using AI tools to find bugs, which is a narrower and more defensible use case. The signal to take from this is that the person responsible for one of the most consequential codebases on earth is treating AI as a review aid, not an author. If your business relies on Linux-powered systems — and statistically it almost certainly does — the stability of that ecosystem is being actively protected from a development fad that prioritizes speed over scrutiny.

What is genuinely interesting is the tension inside Torvalds's position. He is not dismissing AI coding tools as useless; he is separating the emotional experience of programming from the engineering discipline of maintaining production-critical code. That is a more nuanced stance than the binary 'AI will replace developers' versus 'AI writes garbage' debate that dominates most coverage. We find his framing persuasive: finding bugs with AI assistance is a legitimate productivity gain, while letting AI generate code wholesale for a project where a single bad commit can affect billions of devices is a risk calculus that does not pencil out. The skepticism worth applying here is whether 'finding bugs' and 'writing code' can stay cleanly separated as these tools improve.

The second-order effect lands on the labor market and on tooling expectations. If the steward of Linux is publicly drawing this line, it gives cover to engineering leaders at smaller companies who want to resist pressure from executives or investors to adopt AI-generated code aggressively. It also raises a practical question for businesses that hire developers or contract out software work: if your vendor is vibe coding, you are getting prototype-quality output at production prices. Conversely, developers who use AI to explore ideas joyfully and then rewrite carefully may end up more productive and more employable than those who either refuse the tools entirely or lean on them uncritically.

Watch whether other foundational open-source projects follow Torvalds's lead in formally articulating where AI assistance is acceptable and where it is not. The Linux kernel's review process is unusually rigorous, but the reasoning applies to any codebase your business cannot afford to have fail silently. If you operate a small business with any custom software, the actionable takeaway is to ask your developers or vendors a direct question: where in your workflow does AI-generated code get reviewed by a human before it ships, and where does it not? Torvalds just gave you the vocabulary for that conversation.

“The Linux creator calls vibe coding ‘a wonderful way to find joy in programming,’ but not for running Linux development.” — ZDNet

Takeaway: Ask your developers where AI-generated code gets human review before shipping — Torvalds draws the line at authorship, not at bug-finding.