Mark Cuban: Lovable and Replit have the potential to endure beyond the AI labs.

      At the RAISE Summit in Paris on Wednesday, "Shark Tank" investor Mark Cuban advocated for AI coding tools like Lovable and Replit, as reported by Business Insider. He argued that these tools can compete effectively with major AI labs because they offer bundled services that a raw language model does not provide.

      Cuban is an investor in Lovable, giving him a vested interest in the outcome. However, his argument reflects a significant concern surrounding “vibe-coding” startups. He questioned whether established companies like Anthropic or OpenAI could easily replace these platforms by simply adding features. “When we evaluate front-end tools such as Lovable and Replit, we consider the possibility that the foundational models from the big players might simply take over,” Cuban remarked during a discussion with Lovable CEO Anton Osika. “But your argument is that you possess a localized dataset.”

      Transitioning from coding tool to “AI cofounder,” Osika asserted that Lovable has evolved beyond just being seen as an AI software engineer. Founders can now use the platform to start a company and set up payments, rather than just create an app. “Initially, people viewed us as an AI software engineer, the product itself,” Osika noted. “Increasingly, however, we’re being seen as a partner, essentially as their AI cofounder.”

      This perspective is significant from a commercial standpoint. A tool that generates code competes solely based on model quality. In contrast, a platform that manages the entire business is less vulnerable to being replaced. Lovable has leveraged this comprehensive approach, reportedly achieving $500 million in annual revenue with a small team.

      Recent developments have heightened concerns. After Anthropic released its Opus 4.6 model earlier this year, a noticeable shift occurred. Some founders and developers announced on X that they had canceled their Cursor and Lovable subscriptions in favor of Claude Code.

      This trend is what worries investors. They continuously compare vibe-coding apps to raw models, fearing that an update to Claude Code could swiftly eliminate a startup’s competitive advantage. Some tools have responded by developing their own models to reduce reliance on these larger labs.

      Lovable has openly acknowledged this threat. In a podcast from March, its head of growth, Elena Verna, stated that rival startups weren’t her primary concern. “I’m always more worried about the big players out there,” she remarked, referring to companies like OpenAI, Anthropic, Google, and Apple.

      Her reasoning centers on distribution, as these giants can reach billions of users and integrate coding assistants into widely used products. Cuban’s counterargument is straightforward: by owning the workflow surrounding code, including incorporation, payments, and data, they can create a barrier that is difficult for larger labs to penetrate.

      Whether this barrier will remain effective as models advance remains an open question, with a significant portion of European startup value dependent on the answer.