GitHub CEO Thomas Dohmke Warns Startups: “Vibe Coding” Alone Won’t Scale Your Business 2025

Paris, June 15, 2025 — At the forefront of the AI revolution in software development, GitHub CEO Thomas Dohmke is sounding a cautionary note for startup founders: relying solely on AI-powered “vibe coding” tools isn’t enough to build a sustainable, investor-ready company. Speaking at VivaTech in Paris, Dohmke emphasized that while AI coding assistants can help non-technical founders quickly launch products, true value and scalability still require skilled developers and deep technical expertise.

AI Lowers Barriers, But Human Developers Remain Essential

Dohmke highlighted a dramatic shift in the startup landscape over the past two years. Where once founders needed convincing to adopt AI, today’s market is driven by a “FOMO” (fear of missing out) mentality, with more startups leveraging AI tools to bootstrap their ventures. These tools, often described as “vibe coding,” allow founders to generate code using natural language prompts, enabling small teams to build products with minimal funding.

However, Dohmke warned that this democratization comes with a downside: a crowded market of lookalike startups, making it harder for investors to distinguish and back truly innovative ventures. “The investors would ask, ‘Why would I invest in you instead of the 10 other people?'” he said, noting that startups built predominantly with coding assistants often lack the complexity and depth needed to justify further investment rounds.

The Limits of Vibe Coding: Why Technical Depth Matters

Dohmke’s message is clear: non-technical founders may be able to launch a product, but scaling a startup to attract serious investment requires more than AI-generated code. “A non-technical founder will find it difficult to build a startup at scale without developers because they can’t build a complex system to justify the next round,” he explained.

He added, “The value of your startup isn’t defined by what you can develop using cheap measures.” Startups must demonstrate a deep understanding of their systems and the ability to build robust, secure, and scalable products—attributes that AI tools alone cannot provide.

Coding Skills / Developer Skills Still Crucial in the AI Era

Despite the excitement around AI-driven engineering, Dohmke insists that foundational coding skills remain essential. He has previously advocated that “every kid, every child, should learn coding,” stressing the importance of knowing when to rely on personal abilities versus AI assistance1. Software engineers, he says, must develop the “willingness to learn” about AI prompts, but also the discernment to know when not to use them: “If I figure out how to write a prompt for something I can do myself, it’s a waste of time”.

GitHub’s Vision: AI as a Developer’s Companion / Developer Skills

Under Dohmke’s leadership, GitHub is pushing the boundaries of AI-driven development with tools like Copilot, which now writes a significant portion of code in many projects. Still, Dohmke envisions AI as a companion, not a replacement, for human developers. He predicts that software development will become “AI native” in the next five years, with automated workflows and coding practices becoming standard—but always alongside human oversight and expertise.

Security, Open Source, and the Future

Dohmke also addressed the growing importance of security and open-source collaboration. GitHub, now with over 150 million users, is investing in advanced security measures and partnerships to protect developers from emerging cyber threats. He sees AI as a tool to reduce vulnerability response times and empower developers to address issues before they escalate.

Looking ahead, Dohmke is optimistic about AI’s role in democratizing software development, especially in rapidly growing markets like India. However, his core message to founders remains unchanged: AI can help you start, but only real developers can help you scale and stand out in a competitive, investor-driven world