Replit CEO Masad on Independence, Cursor's Margins, and AI Model Rankings

Masad claims Replit grew from $2.8M in 2024 to ~$1B ARR — roughly 350x in run rate terms over ~18 months

The inflection point, per Masad, was September 2024 when Replit launched its agentic coding experience. He also cited net revenue retention of up to 300%, meaning the existing customer base would triple revenue over a given period even with zero new customer acquisition — a strong signal of product stickiness.

Masad prefers independence but stopped short of ruling out a sale, citing fiduciary duty

He said partners 'sometimes bring up these topics' but that Replit will 'try to stay independent.' He framed independence as fulfillment of a 10-year vision — 'a billion software creators' — which he pitched at YC in 2018 when people 'sometimes actually laughed at that dream.'

Cursor reportedly in $60B SpaceX acquisition talks; Masad claims Cursor operates at -23% gross margins

Masad used these figures — attributed to industry reports and his own claims — as a foil to argue Replit has built a more durable independent business. He said Cursor's negative margins combined with model-training ambitions make independence structurally untenable. The -23% figure has not been independently verified; it reflects Masad's characterization of external reporting.

Replit's September 2024 agentic coding experience credited as the primary revenue growth catalyst

The platform takes users from natural-language prompt to fully deployed, scalable application, handling security, databases, and migrations. Masad positions this end-to-end approach as differentiated from developer-focused tools like Cursor, primarily serving non-technical users who previously couldn't create software.

Masad willing to take Apple to court over App Store dispute he calls 'outright lies'

Details were limited in the available content, but Masad's statement signals a combative posture toward Apple's platform policies. This is Masad's own characterization of the dispute, not independently verified.

Masad's model rankings from operational experience: Anthropic leads agentic loop, GPT-5 catching up fast, Google Flash best on price-performance

These rankings represent Masad's working assessment from Replit's real infrastructure decisions at scale. Anthropic described as 'still undefeated on the core agentic loop' with best-in-class tool calling and long-session coherence. GPT-5 described as catching up quickly. Google Flash cited as the price-performance leader for speed-sensitive workloads. These are Masad's opinions, not independent benchmarks.