Boris Cherny, the architect behind Anthropic's Claude Code, believes the tech industry has settled on the wrong vocabulary for what AI is doing to software development. At a recent San Francisco conference, he expressed frustration with the term "vibe coding," a phrase that has become shorthand for AI-assisted programming since OpenAI researcher Andrej Karpathy popularized it in early 2025. For Cherny, the casual language fails to capture the economic and technical significance of what's actually happening in the field.

The push to rebrand isn't merely semantic nitpicking. When a technology generates billions in revenue and fundamentally reshapes how developers work, the language used to describe it matters. "Vibe coding" was even named Collins Online Dictionary's word of the year, reflecting its cultural penetration. Yet Cherny argues that a term borrowed from casual internet slang doesn't adequately convey the sophistication and business impact of AI-driven code generation.

What's Wrong With "Vibe Coding" as a Technical Term?

The core issue, according to Cherny, is that "vibe coding" sounds too informal for a technology that's becoming a cornerstone of professional software development. It's technically accurate but lacks precision, much like calling an aircraft a "flying thing." The term doesn't signal the gravity of what Claude Code and similar tools represent in terms of productivity gains and economic value.

Anthropic currently markets Claude as an "AI-powered coding assistant," while OpenAI describes its Codex as a "coding agent." Neither label fully captures the transformative potential. Cherny has suggested "agentic engineering" as a possible replacement, though he acknowledges it falls short of the mark. The challenge is finding terminology that reflects both the technical sophistication and the real-world impact of these tools.

How to Participate in Redefining AI Coding Terminology

• Share suggestions on social media: Cherny has encouraged audiences to propose replacement terms for "vibe coding" online, opening the terminology evolution to public input and crowdsourced creativity.
• Consider the economic context: When evaluating new terms, think about whether the language reflects the billions in revenue these tools generate and their role as revenue machines in the software development ecosystem.
• Evaluate technical precision: Assess whether proposed terminology captures the sophistication of AI-driven code generation and distinguishes it from earlier, less capable automation tools.

The search for better terminology reflects a broader pattern in how emerging technologies are named and understood. Early terms often stick, even when they're imprecise. Consider how "cloud computing" persists despite being vague about what's actually happening with distributed infrastructure. The terminology we choose shapes industry perception and adoption rates.

Why Infrastructure Matters as Much as the Model Itself

During the conference, Cherny also discussed Anthropic's new partnership with SpaceX to expand compute capacity. This development underscores a critical shift in how AI companies think about their bottlenecks. The limiting factor is no longer the quality of the model itself; it's the raw computing power needed to run it at scale. More GPUs and faster processing aren't just incremental improvements; they're prerequisites for the next generation of AI-driven coding tools.

This infrastructure expansion signals Anthropic's confidence in Claude's market position and its readiness to scale operations. As demand for AI coding assistants grows, the companies that can reliably deliver compute capacity will have a significant competitive advantage. The SpaceX deal represents a strategic bet on future growth and the ability to meet that demand without infrastructure constraints.

The terminology debate may seem like a minor issue compared to technical breakthroughs or business deals, but it carries real weight. Language shapes how industries perceive and adopt technologies. If "vibe coding" remains the dominant term, it risks trivializing a technology that's generating billions in revenue and reshaping how software gets built. Cherny's push for a more precise vocabulary is ultimately about ensuring that the industry's language matches the magnitude of the transformation underway.