The AI coding tool market is consolidating fast. Cursor, the darling of the developer tools space, is reportedly in talks to be acquired by SpaceX for $60 billion after raising $2 billion at a $50 billion valuation just weeks ago. Every AI coding startup is now facing an existential question: sell, or try to survive on your own?
Replit founder Amjad Masad’s answer is refreshingly blunt: he’d rather not sell. And unlike most AI startup founders making that claim, he actually has the numbers to back it up.
The Margins That Matter
At TechCrunch’s StrictlyVC event in San Francisco on Thursday, Masad laid out the financial case for staying independent — and it starts with a brutal comparison to Cursor.
Part of the reporting on Cursor’s SpaceX deal suggested the company is operating at negative 23% gross margins. That means for every dollar of revenue, Cursor is spending $1.23 just on compute costs before paying a single employee. Masad didn’t gloat, but he didn’t need to: “For us at Replit, partly because we target a different customer set, we’ve been gross margin positive for over a year.”
Replit’s net revenue retention — a measure of how much existing customers expand their spending — is hitting 300%. That’s not a typo. For every dollar a customer spends in month one, they’re spending three dollars within a year. That’s the kind of metric that makes venture capitalists sit up straight.
The contrast is stark. Cursor raised at a $50 billion valuation while burning cash on every user. Replit went from $2.8 million in all-of-2024 revenue to tracking toward a billion-dollar annual run rate in roughly 18 months. One is a venture-backed moonshot. The other is an actual business.
The Billion Software Creators Bet
Masad’s thesis is that the market for AI coding tools is much bigger than professional developers. “Our audience tends to be mostly non-technical users who previously haven’t been able to create any software,” he said. “We provide an end-to-end platform — from the prompt all the way to a deployed application that can scale.”
This is the bet that Cursor isn’t making. Cursor is building for developers who already know how to code and want to code faster. Replit is building for everyone else — the marketing manager who wants an internal tool, the small business owner who needs a booking system, the student who has an idea but doesn’t know Python from a snake.
It’s the difference between selling picks to gold miners (Cursor) and selling picks to everyone who wants to build a house (Replit). The second market is orders of magnitude larger.
The Model Bake-Off
Masad also gave a surprisingly candid assessment of the foundation model landscape from a customer’s perspective:
- Anthropic: “Still undefeated on the core agentic loop. They have the best tool calling; the agent can stay coherent much longer.”
- OpenAI (GPT-5): “Catching up quickly.”
- Google (Flash): “Just amazing on price-performance. If you want something fast and cheap, they’re actually beating open source right now.”
- Newer labs: “Reflection AI is coming out with open-source models we’re hearing great things about.”
- Chinese models: “Kimi is as good as an Anthropic-generation model from January, so it’s only about three months behind.”
Three months. That’s the competitive moat of frontier AI models right now — the time it takes for a Chinese competitor to replicate your capabilities at a fraction of the cost. Which is precisely why Masad thinks the play isn’t to own the model, but to own the platform.
The Apple Fight
Masad also didn’t back down from Replit’s ongoing battle with Apple over App Store restrictions, calling Apple’s claims about Replit “outright lies.” The fight is about whether Replit can offer cloud-based coding on iOS without Apple taking its 30% cut — and it’s a fight that could set precedent for every AI coding tool that wants to reach mobile users.
“If we can’t distribute on the world’s most popular platform because of a 30% tax that Apple doesn’t even apply consistently, that’s not just a Replit problem,” Masad said. “That’s a problem for anyone building the next generation of creative tools.”
Why It Matters for NZ Developers
The AI coding tool shakeout has direct implications for NZ’s developer community:
-
Tool lock-in is real. When Cursor gets acquired by SpaceX, licensing models change. When a tool you depend on gets absorbed, you’re at the mercy of the acquirer’s roadmap. Replit’s independence means its roadmap answers to users, not a corporate parent.
-
Non-technical users are the growth market. NZ businesses that want to build internal tools without hiring developers should be watching Replit’s trajectory. The “prompt to production” pipeline is exactly what small and medium NZ businesses need.
-
The margin math matters. Negative-gross-margin startups eventually have to raise prices or cut costs. If you’re building a business on Cursor’s platform, expect pricing changes post-acquisition. Replit’s positive margins suggest pricing stability.
The AI coding tool space is going to look very different in 12 months. Some of today’s darlings will be absorbed. Others will fold. Replit’s bet on independence — powered by actual positive margins and a broader customer base — might be the contrarian call that pays off.
🔍 The Bottom Line
The AI coding tool shakeout isn’t about who has the best autocomplete. It’s about who has a sustainable business model. Cursor is burning cash at a $50 billion valuation. Replit is profitable and growing at 300% net revenue retention. Masad’s refusal to sell isn’t stubbornness — it’s arithmetic. The market for “AI helps you code” is much smaller than “AI helps you create software.” Replit is betting on the latter, and right now, the numbers are on their side.
