In a candid interview at Bitcoin World's StrictlyVC event in San Francisco, Replit CEO Amjad Masad revealed that the AI coding platform's annual revenue run rate is approaching $1 billion, a massive leap from $2.8 million in total revenue for 2024. The company has experienced unprecedented growth over the past 18 months, driven by its agentic coding experience launched in September 2024, which allows users to create software from simple prompts.
Masad disclosed that Replit's net revenue retention has reached as high as 300%, indicating customers are significantly increasing their investment in the platform over time. Unlike competitor Cursor, which Masad noted operates at negative 23% gross margins, Replit has maintained positive gross margins for over a year. This financial stability allows the company to consider staying independent. "We're going to try to stay independent," Masad stated. "I would love for us to remain an independent company."
The company targets non-technical users who need an end-to-end platform, and has acquired major enterprise clients like Zillow and Meta through organic adoption. When formal bake-offs occur, Replit often wins on product and security. The platform's full-stack approach makes applications inherently more secure, with built-in databases not open to the public.
Replit CEO Challenges Apple's App Store Policies
A significant portion of the interview focused on Replit's dispute with Apple. The company has been blocked from updating its iOS app for months, while a rival, Lovable, recently received approval. Masad believes Apple is threatened by Replit's ability to generate iOS apps, which could reduce reliance on Apple's Xcode development environment. Apple's stated reason for blocking Replit is that the app downloads new code after approval, violating guidelines. Masad called this "a lie" and stated that Replit could prove it in court if necessary.
While the App Store issue is not critical to Replit's revenue, it affects users who genuinely love the app. Masad noted that children in underprivileged communities use Replit on Android devices to learn coding, and executives use it in meetings. The platform has been on the App Store for four years, and the current blockage has frustrated many users.
AI Model Partnerships and Customer Investments
Masad revealed that Replit works closely with Anthropic, Google, and OpenAI. He ranked Anthropic as undefeated on the core agentic loop, with the best tool calling and coherence. He noted that GPT-5 is catching up quickly, while Google's Flash family offers excellent price-performance. Masad also mentioned newer labs like Reflection AI and Chinese models like Kimi, which are competitive with earlier Anthropic models.
Masad reported that churn is very low, and net retention is incredibly high. Enterprises often find that rebuilding apps from Replit into their own stacks makes them worse. Bain & Company, for example, replaced Tableau and Power BI with Replit and Databricks. Customers typically see returns of one to three orders of magnitude on their Replit investment.
Masad revealed that Replit is considering investing in its own customers in exchange for equity, similar to strategies used by Nvidia and OpenAI. He has personally invested in startups that began on Replit, such as Magic School, which generated $20 million in its first year. Other companies started on Replit are now valued at half a billion dollars.
In a separate development, Apple CEO Tim Cook warned that Mac mini and Mac Studio could remain in short supply for several months after AI-driven demand far exceeded the company's forecasts. The open-source AI agent platform OpenClaw, now backed by OpenAI, turned Apple's unified memory architecture into the default hardware for running large local AI models. Apple's M4 Ultra supports up to 192GB of unified memory, letting developers run models that cannot fit on any single consumer Nvidia GPU, which maxes out at 32GB of VRAM.
Mac revenue came in at $8.4 billion for the quarter, up 6% year-over-year. The $599 base Mac mini is sold out in the U.S. with no delivery or in-store pickup available. Upgraded configurations with 64GB of RAM are showing wait times of 16 to 18 weeks. Mac Studio models with 512GB of unified memory disappeared from the store completely.
