Apr 25, 2026 · 48m · news
Replit CEO: Why the SaaS Apocalypse is Justified & Why Coding Models are Plateauing | Amjad Masad · 20VC with Harry Stebbings
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this interview, Replit CEO Amjad Masad discusses the fundamental shift from traditional coding to AI-driven 'vibe coding,' the structural collapse of middle-tier SaaS companies, and the complex engineering and economic realities of running modern AI agent platforms.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Harry holds 19.1% of the talking time here. How this is scored →
Harry as informed peer 3.9 Guest teaching 3.7 Guest disagreement 2.1 Harry pushing back 3.4
speaking balance: gold is Harry, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 42:05 Refusing to pick a rival to shoot
When Stebbings forces a choice on which competitor to eliminate, Masad rejects the prompt completely and reframes all competitors as mere followers of Replit's roadmap.
Hardest push from Harry ▶ 39:30 Refusing vague explanation for Apple blockadeStebbings refuses to accept surface-level excuses about Apple blocking Replit's app updates, repeatedly pressing Masad to admit the underlying strategic conflict.
Biggest teaching moment ▶ 29:10 Dismantling Twitter VC echo chambersMasad corrects Stebbings' premise about Cursor's decline by explaining how Twitter distorts reality and tricks VCs who lack broader enterprise market visibility.
Harry holds his own ▶ 11:30 Directly probing 80% gross margin drainStebbings demonstrates deep financial domain knowledge by pressing Masad on whether 80 percent of top-line revenue gets drained directly to underlying model providers.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Harry as informed peer | Guest teaching | Guest disagreement | Harry pushing back | Why |
|---|---|---|---|---|---|---|
| Formal Welcome and the Genesis of Replit | 2 | 2 | 1 | 1 | Stebbings opens with a welcoming question about Replit's history and vision. Masad explains his journey from Jordan to founding Replit and pivoting from learn-to-code to agentic AI. | |
| The Rise of Agentic AI and the Innovation Cycle | 3 | 4 | 1 | 2 | Stebbings asks about value capture between application infrastructure and foundation models. Masad explains the dynamic cycle of building custom guardrails that get absorbed as models evolve. | |
| Building Proprietary Models vs. Leveraging Foundations | 5 | 4 | 3 | 5 | Stebbings presses on whether training custom models is a strategic mistake given rapid frontier model upgrades. Masad explains why model training opportunities ebb and flow over time. | |
| Cost, Margin Optimization, and the Edge of Growth | 5 | 5 | 3 | 6 | Stebbings directly probes gross margins and suggests 80 percent of revenue goes to model providers. Masad corrects the stat and argues performance leadership must precede margin optimization. | |
| The Society of Models as Core IP | 4 | 4 | 2 | 3 | Stebbings asks how Replit routes model queries and raises security concerns regarding Chinese AI models. Masad details Replit's psychological evaluation of models and security boundaries. | |
| Inferences as Marketing and Creative Addiction | 4 | 3 | 1 | 2 | Stebbings references Jason Lemkin's thesis on using inference as a growth driver. Masad validates the premise and highlights non-technical ops teams as high-ROI users. | |
| The SaaS-pocalypse and the Modern Data Warehouse | 5 | 4 | 2 | 4 | Stebbings asks if the SaaS-pocalypse market cap drops are justified. Masad outlines how enterprises build custom software on Databricks warehouses rather than buying middle-tier SaaS. | |
| Traditional Incumbents and Code Maintenance | 4 | 4 | 2 | 3 | Stebbings brings up the operational burden of software maintenance for non-developers. Masad explains Replit's automated testing agents and intentionally blunt AI code review agents. | |
| Price Sensitivity and Monetization Models | 5 | 4 | 2 | 3 | Stebbings asks about buyer price sensitivity around security and predictions for future token prices. Masad contrasts cost-conscious developers with ROI-focused enterprise ops managers. | |
| Is Cursor Dead? Hype vs. Sticky Enterprise Sales | 5 | 5 | 3 | 5 | Stebbings brings up conflicting signals about Cursor being dead versus hitting huge revenue. Masad dismantles the Twitter narrative as VC echo-chamber noise disconnected from enterprise sales. | |
| Are IDEs Dead? The Case for Verification | 4 | 4 | 2 | 3 | Stebbings asks if IDEs will be obsolete within two years. Masad explains that while IntelliSense features are obsolete, safety-critical code still requires inspection interfaces. | |
| Advice for Students: Should You Study Computer Science? | 4 | 4 | 2 | 3 | Stebbings questions the value of computer science degrees when curriculums lag behind AI. Masad distinguishes intrinsic engineering interest from careerist credential chasing. | |
| Lean Organizations and the Changing Nature of Teams | 5 | 4 | 2 | 4 | Stebbings asks if companies will drastically shrink team size. Masad offers counterexamples where vibe coding efficiency led media companies to hire more builders. | |
| Replit's App Store Battle with Apple | 5 | 3 | 3 | 6 | Stebbings probes deeply into Apple blocking Replit's iOS app updates for three months. Stebbings pushes past diplomatic answers to demand the true commercial motivation behind Apple's delay. | |
| Staying Ten Steps Ahead of Competitors | 3 | 3 | 4 | 5 | Stebbings playfully asks Masad to pick one rival startup to shoot. Masad rejects the premise, asserting that competitors merely copy Replit's product roadmap. | |
| Quick-Fire: High Performance, Health, and Parenting | 2 | 2 | 1 | 1 | Stebbings conducts a quick-fire segment covering peak executive performance, weight loss, and parenting lessons. Masad highlights long-term habit formation and sauna resets. | |
| What is True Product-Market Fit? | 2 | 4 | 1 | 1 | Stebbings asks what Masad wishes he knew at the beginning of Replit. Masad describes true product-market fit as explosive demand where the product is pulled out of your hands. |
Statements from this episode (36)
Masad: Focusing on cost signals a technology has reached an S-curve plateau
“Cost question is secondary to the performance question. When you focus on cost is when you reach a certain asymptotic plateau in the S-curve.”
Masad: Solo founders no longer need to learn traditional programming languages
“They don't need to learn how to code. They need to learn how to create. They need to learn how to build.”
Masad: Google's Gemini AI models offer the best price performance
“Google's Gemini's models have are the best at price performance, for example.”
Masad: Replit once sent more AI tokens to Google than to Anthropic
“Actually, at some point we were sending more tokens to Google than we were sending Anthropic, despite Anthropic being that kind of the core workhorse.”
Masad: AI coding models are approaching a performance plateau
“And we're, you know, we're approaching a certain plateau in how good coding models could get.”
Masad: AI startups prioritizing cost over performance will lose to competitors
“But if you focus on cost at the expense of performance, you're going to lose.”
Masad: Anthropic operates at 40% gross margins due to Nvidia chip costs
“Even, like, Anthropic, every hundred dollars they make, 60 goes to NVIDIA, like, you know, their margins are public, it's 40%, right?”
Masad: Google's Gemini is one of the best AI models for design
“Gemini is one of the best models at design.”
Masad: Adopting Chinese AI models poses security risks, not moral issues
“I don't think there's a moral issue per se. Like, I don't think they're using slave labor or anything like that to do that. I would have a security issue, especially since we have enterprise customers that, that depend on us for sensitive data and things like …”
Masad: Open-source AI is essential to prevent corporate oligopolies from price collusion
“Open source is going to be very important for us to actually have a free market around AI, because if we're going to end up in an oligopoly of AI companies, then the, you know, there's actually an economic theory of how they'll naturally collude on price and p…”
Masad: The US government should fund a national open-source AI consortium
“I would venture to say, like, maybe the US government should start a consortium of companies that are, like, creating the best open source, national open source model so that the market stays competitive.”
Masad: Late 2025 AI coding hype was driven by free token giveaways
“If you think about the hype Period that we got with, like, Claude code and codex towards the end of 25, early 26, a lot of it was driven by how much free tokens they were given out, how much, you know, every other day they're like, oh, it's 50% more tokens or …”
Masad: Agentic AI development is a healthier addiction than social media consumption
“Agentic development is addictive. It's, I think, the better kind of Addiction than like social media or passive consumption. It's like creative addiction, which is kind of good.”
Masad: Future engineers will focus on infrastructure while generalists build applications
“If I were to make a prediction, I would say that we'll still have engineers inside the organizations. Those engineers are responsible for more infrastructure, AI, ML embedded systems, you know, more, more, you know, low level engineering. And then you have pro…”
Masad: Enterprise operations teams achieve higher ROI from Replit than product teams
“And when they use Replit, the return on investment is as good, if not greater than product teams. With product teams, you're cutting down product development lifecycle, which is hugely valuable. With operations teams, you're actually being more efficient. You'…”
Masad: Enterprises are skipping SaaS tools to build directly on data warehouses
“People are skipping the SaaS tools entirely and like building on top of their data warehouse.”
Masad: Replit spends as many AI tokens maintaining code as creating it
“We spent a lot of tokens on maintenance as much as we spent on creating that software.”
Masad: Any problem created by AI can be solved with more AI
“The thing about AI, any problem AI creates, there's more AI that you can build to solve that problem.”
Masad: 75% of Replit's users are non-engineers
“We have some engineers use Replit, but the 75% are non-engineers.”
Masad: AI unit token prices are decreasing slower than the market expected
“Pure, pure unit price is not going down as much as people expected it to go.”
Masad: Frontier AI pricing pressure is low with only two true competitors
“I think part of it is we're living in a world where the true frontier models is actually just between GPT-V and Claude. And so there's no, not a lot of pricing pressure.”
Masad: Frontier AI labs are unprofitable due to high Nvidia chip costs
“I don't think these companies are all that profitable, partly because, you know, the underlying chips are not all that cheap and there isn't a lot of competition.”
Masad: The software generation market will be larger than the internet
“And so it's like an ever expanding TAM and it's going to be perhaps as big as the internet if not more. And that's like, I'm talking about just software generation broadly.”
Masad: Enterprise customers rarely replace AI coding tools unless they fall behind
“Cursor has done well at enterprise sales. And enterprise is a very sticky customer and once they adopt something, it has to really fall tremendously behind, perhaps like GitHub Copilot, in order for people to replace it.”
Masad: Venture capitalists have poor information landscapes because they live on Twitter
“Which why I think VCs don't have the best, I think, information landscape, because a lot of them just live on, on Twitter, but the world is much larger than that.”
Masad: Traditional IDEs have no future and are effectively dead
“I think for all intents and purposes, IDEs are dead. I think the, you know, the limp along because again, some engineers just love that, that control, but there's no future in them and that there's no one's going to be asking for like the latest feature of lik…”
Masad: Mission-critical software like SpaceX will always require traditional IDEs
“If I'm writing, like, mission-critical software for self-driving cars or a NASA or SpaceX mission, I think there's always going to be a need for some kind of IDE.”
Masad: AI vibe coding will mirror JavaScript by capturing low-risk software domains
“JavaScript is the original sort of vibe language, because JavaScript didn't have types. You could like run into errors. But the reason why we adopted JavaScript is because web software was not going to kill anyone. Your Gmail, if someone pushes a bug to Gmail,…”
Masad: Studying computer science just for high Google salaries is over
“Now, if you're not into computer science, if you don't feel like you're drawn to it, like a fly drawn, drawn to a light, then don't go into it because someone told you you're going to make a boatload of money working for Google. That's gone. So that it's prett…”
Masad: Demand for core computer science fundamentals like data structures will remain
“Well, I will say if computer science is, like, the field of computer science where you're, like, learning about you know, data structures and algorithms, that's not going to change. So if you, and there's always going to be need for people to understand the un…”
Masad: Replit aims to eliminate support roles, hiring only builders and salespeople
“For us at Rap Lit, I think what we want to try to eliminate is or reduce is a lot of supporting roles. We want builders, right? We want we want builders and we want salespeople”
Masad: Enterprise software sales reps are evolving into AI educators and consultants
“And also the sales role is changing in that salespeople are becoming more like educators and transformation sort of consultants.”
Masad: Replit has been stuck in Apple App Store review for months
“Replit has been on the App Store since twenty-twenty-two doing exactly the same thing, allowing people to generate or write code. And running it in a browser window. Suddenly they're saying that we are you know, we're not complying with their guidelines. And w…”
Masad: Apple accepts applications created using Replit at a very high rate
“They're accepting apps made on replet. They're accepting them into the app store. I tweeted about one yesterday. There's the acceptance rate. It's very high.”