Indian startup Rocket.new just closed a $15 million seed round led by Salesforce Ventures to challenge the likes of Cursor and Bolt in the exploding vibe-coding market. The platform has rocketed to 400,000 users and $4.5 million in annual recurring revenue in just three months since launching beta, positioning itself as the production-ready alternative to prototype-focused competitors.
Rocket.new is making waves in India's tech scene from an unlikely location. While most Indian startups cluster around Bangalore or Mumbai, this AI-powered app development platform launched from Surat - a city better known for diamonds and textiles than disruptive technology.
The timing couldn't be better. As vibe-coding tools like Cursor, Bolt, and Lovable capture developer mindshare globally, Rocket.new is carving out its own niche by focusing on production-ready applications rather than quick prototypes. "We are building the first vibe solution platform, which is not solving just a problem of day one, but what we are focusing on is solving the problem of day two," CEO Vishal Virani told TechCrunch.
The strategy is paying off. Since launching in beta just 16 weeks ago, Rocket.new has attracted over 10,000 paid subscribers among its 400,000 user base across 180 countries. The startup's $4.5 million ARR represents impressive unit economics with gross margins hovering between 50-55%, which the team aims to push to 60-70% in coming months.
Salesforce Ventures led the $15 million all-equity seed round, with Accel and Together Fund joining as co-investors. "We saw a clear gap between the magic of AI code generation and the reality of making that code production-ready," Kartik Gupta from Salesforce Ventures explained. "Rocket.new is purpose-built to solve this problem of iteration, maintenance, and deployment at enterprise scale."
What sets Rocket.new apart from competitors isn't just positioning - it's the underlying architecture. The platform combines large language models from Anthropic, OpenAI, and Google's Gemini with proprietary deep learning systems trained on datasets from the founders' previous venture, DhiWise. "Our underlying architecture is completely different from what Lovable, Bolt, and everyone is doing," Virani emphasized.
