Meta’s pursuit of leadership in artificial intelligence has encountered an unexpected obstacle: its own internal systems. Despite significant investments in AI, the company’s internal developer tools have proven to be too slow, prompting CEO Mark Zuckerberg and his team to seek external solutions. A report from Business Insider indicates that Meta is encouraging employees involved in AI initiatives to transition from the company’s proprietary platforms to popular external services such as Vercel and GitHub. This change comes as frustration grows within Meta’s Product and Applied Research (PAR) division, led by former GitHub CEO Nat Friedman, due to prolonged deployment times.
In a candid memo issued in late September, Aparna Ramani, head of infrastructure at Meta Superintelligence Labs (MSL), acknowledged that the internal systems require “too long” to implement changes and are “not conducive to vibe coding.” Vibe coding describes a modern practice where engineers heavily depend on AI to generate and enhance code in real-time. While smaller tech teams outside of Meta can execute updates within minutes, internal delays can extend to nearly two hours for minor modifications. To address this, Ramani proposed a two-part plan: immediately adopt Vercel to alleviate bottlenecks for AI teams while concurrently developing a quicker in-house alternative, code-named “Nest,” specifically tailored for TypeScript applications.
Her objective is to reduce deployment times from 99 minutes to under two. The shift is already yielding positive results. By mid-September, at least 10 AI projects within Meta’s PAR group had transitioned to Vercel, allowing updates to be deployed in minutes rather than hours. The combination of Vercel and GitHub — Microsoft’s widely used code-hosting and collaboration platform — has provided teams with the necessary speed to efficiently build and test AI applications. Vercel, a rapidly growing player in cloud and AI app development, specializes in creating and deploying scalable web applications and has attracted major clients such as Netflix, Adobe, and Stripe. Notably, both Friedman and Alexandr Wang, Meta’s newly appointed Chief AI Officer, are investors in Vercel.
The platform recently secured $300 million at a valuation of $9.3 billion, highlighting its increasing significance. For Meta, the choice to utilize external services is part of a broader strategy to secure a leading position in the AI field. Zuckerberg has invested millions in Meta Superintelligence Labs, recruited top AI talent with lucrative compensation packages, and reorganized teams to advance Artificial General Intelligence (AGI) — a vision of AI with capabilities akin to human cognition. However, as the report indicates, ambition alone may not suffice. Without the agility that faster development platforms offer, even a company as large as Meta risks being hindered in the global competition for AI innovations.
Currently, Vercel and GitHub seem to be the essential resources Meta requires to maintain momentum in its AI projects.