Will lab-owned coding agents displace independent AI coding tools, or can multiple coding tools coexist?

Mike Cannon-Brookes · Oct 13, 2025

AI coding tools are cheap enough that companies should be liberal and run many of them in parallel rather than standardising on one

The tools are quite cheap

Scope: acknowledges he is talking his own book on Rovo

34:04 20VC: Atlassian CEO on Why Everything is Overvalued & Are We in an AI Bubble | Do Margins Matter & Does Defensibility Exist in an AI World | Is Per Seat Pricing Dead & The Future of Vibe Coding with Mike Cannon-Brookes

Mike Cannon-Brookes · Oct 13, 2025

AI coding tools have durably different strengths and weaknesses (greenfield vs. large codebase ingestion vs. cross-repository changes), so engineers will keep using several rather than consolidating on one

Most are good at greenfield work but poor at long-form edits across hundreds of repos or ingesting large codebases; they will all improve at different things, and preferences and styles persist

34:32 20VC: Atlassian CEO on Why Everything is Overvalued & Are We in an AI Bubble | Do Margins Matter & Does Defensibility Exist in an AI World | Is Per Seat Pricing Dead & The Future of Vibe Coding with Mike Cannon-Brookes

Sebastian Siemiatkowski · Feb 16, 2026 · hedged

Cursor's future is better than Harry's prediction — coding tools like Cursor and Claude Code have distinct personalities and skills and will be used alongside each other, and non-engineers still need an IDE

He personally switches between Cursor and Claude Code depending on the task, and as a non-engineer who never used VS Code, Cursor is his default IDE

Scope: concedes Harry could be right; based on his own usage patterns

52:51 20VC: SaaS is Dead: Why Systems of Record Will Die in an Agentic World | What Revenue Multiple Will Software Companies Trade At? | From 7,000 to 3,000: We Need Less People Than Ever with Sebastian Siemiatkowski

Zach Lloyd · Oct 17, 2025

Competition from the model providers is a genuine issue, but it isn't a reason not to build a company

26:16 20VC: The Startup Adding $1M ARR Every Week | Competing Against OpenAI's Codex and Claude Code: Who Wins | Why Gemini is Failing and GPT-5 Is Winning | Do Margins Matter in a World of AI | The Ugly Truth About AI Coding with Zach Lloyd, Warp

Everett Randle · Nov 10, 2025

People misread Cursor's declining market share because the code generation market itself has exploded — falling share of a market growing from zero to $6–7B ARR in two and a half years is not the same as decline

Cursor went from roughly 80% share to maybe 25–30% of market ARR, but the absolute market grew enormously, so the relevant question is market size not share

Scope: share figures given as approximations

18:50 20VC: Benchmark's Newest General Partner Ev Randle on Why Margins Matter Less in AI | Why Mega Funds Will Not Produce Good Returns | OpenAI vs Anthropic: What Happens and Who Wins Coding | Investing Lessons from Peter Thiel and Mamoon Hamid

Anish Acharya · Feb 9, 2026

The common mistake in AI analysis is assuming efficiency rises while ambition and customer count stay fixed; demand for software vastly exceeds supply, so Cursor, Codex (app and CLI) and Claude Code can all find product-market fit and grow together

Human ambition for more things always grows faster than our means, and there are distinct developer and developer-adjacent archetypes suited to each product

15:03 20VC: Is SaaS Dead in a World of AI | Do Margins Matter Anymore | Is Triple, Triple, Double, Double Dead Today? | Who Wins the Dev Market: Cursor or Claude Code | Why We Are Not in an AI Bubble with Anish Acharya @ a16z

Andrew Ng · Nov 17, 2025

Leadership among AI coding tools is churning so fast that the best tool changes roughly every three months

His own favorite coding tool would have changed every time he was asked over the past year

53:03 20VC: Andrew NG on The Biggest Bottlenecks in AI | How LLMs Can Be Used as a Geopolitical Weapon | Do Margins Matter in a World of AI? | Is Defensibility Dead in a World of AI? | Will AI Deliver Masa Son's Predictions of 5% GDP Growth?

Maor Shlomo · Nov 24, 2025 · hedged

He would rather invest in Cognition at $12B than Cursor at $29.3B

Cognition has more room to grow and can replicate successful features over time, whereas backing Cursor at that price is a bet that the market is winner-take-all

Scope: 'probably'

49:49 20VC: Base44's Maor Shlomo on How Vibe Coding Will Kill SaaS and Salesforce | Why it is BS that Vibe Coding Platforms Do Not Have Defensibility and Bad Margins | Why He Worries About Google, Not Replit and Lovable | Why Long Anthropic, Not OpenAI?

Harry Stebbings · Feb 9, 2026 · hedged

There is a chance Cursor loses half its revenue this year as Claude Code cannibalizes it

He knows almost no one who has not moved to Claude Code, and is surprised when he hears someone is still on Cursor

Scope: Framed as 'a chance', not a prediction; Based on his own anecdotal network

14:47 20VC: Is SaaS Dead in a World of AI | Do Margins Matter Anymore | Is Triple, Triple, Double, Double Dead Today? | Who Wins the Dev Market: Cursor or Claude Code | Why We Are Not in an AI Bubble with Anish Acharya @ a16z

Harry Stebbings · Feb 16, 2026

Cursor will lose half of its revenue in 2026

Claude Code has eaten their lunch — he doesn't see any engineering team still using Cursor

52:37 20VC: SaaS is Dead: Why Systems of Record Will Die in an Agentic World | What Revenue Multiple Will Software Companies Trade At? | From 7,000 to 3,000: We Need Less People Than Ever with Sebastian Siemiatkowski

Brendan Foody · Sep 15, 2025

Switching costs between coding tools are surprisingly low, and the resulting competition is what drives the negative gross margins seen in the coding space

Sources of defensibility — data flywheels, custom models for your codebase — are being built but take more time to develop

Scope: products are moving toward adding switching costs over time

39:31 20VC: Mercor: From $1M to $500M in 17 Months: The Fastest Growing Company in the World | How to Think About Margins and Revenue Sustainability in AI | Why Evaluation Benchmarks in AI are BS Today with Brendan Foody

Harry Stebbings · Oct 17, 2025

The main risk in underwriting standalone AI coding companies is that switching costs are low, making users easily switchable to OpenAI or Anthropic, which have immense budgets

Codex and Claude Code will be incredibly difficult competition with immense budgets, and the ability for users to switch is high

Scope: applies to Cursor as well as Warp; not a concern about customer quality — Warp users are likely high-quality devs

25:09 20VC: The Startup Adding $1M ARR Every Week | Competing Against OpenAI's Codex and Claude Code: Who Wins | Why Gemini is Failing and GPT-5 Is Winning | Do Margins Matter in a World of AI | The Ugly Truth About AI Coding with Zach Lloyd, Warp

Harry Stebbings · May 12, 2025 · hedged

Cursor is by far the biggest AI coding tool in use, with little adoption of Windsurf among the teams he hears about.

Based on what he hears from companies about internal usage.

Scope: anecdotal, based on who he has heard from

26:44 20VC Exclusive: Mercury Founder Launches First $26M Fund | Why Founders Should Take the Highest Price | Why Serial Entrepreneurs are Better | Why AI Is So Overhyped | The Future of Venture Capital with Immad Akhund

Cem Kansu · Jun 20, 2025 · hedged

Cursor will win over Windsurf in the AI coding tool market

Internally at Duolingo he sees far more usage and creation coming out of Cursor

Scope: based on internal Duolingo usage only; admits they may not have experimented enough with Windsurf

26:17 20Product: How Duolingo Build Product 10x Faster with AI | Duolingo's Biggest Lessons on Paywalls, Push Notifications and In-App Purchases | Why Small Teams are the Future of Product | Why PMs Will Become Extinct with Cem Kansu, CPO @ Duolingo

Anish Acharya · Feb 9, 2026

The developer tooling market will resemble the cloud oligopoly (AWS, Google Cloud) with reasonable margins rather than the Uber/Lyft dynamic where price is competed away

Uber and Lyft are extreme pure substitutes; cloud providers are roughly substitutes with specializations and have all done well, and in the apps layer some users want a rich IDE while others want to be closer to the metal

16:04 Dev tooling becomes a cloud like oligopoly with healthy margins

Miles Clements · Mar 9, 2026

The belief that Cursor is tied to the IDE is a misunderstanding; the company has already moved decisively to agents.

Cursor's IDE innovation was so disruptive that people over-assume; but 2x more people use agents than the tab feature, 90% of users are daily active on the agent product, it grew 15x last year, and the three-month-old Cloud Agent product already drives 35% of merged PRs

6:35 Independent tools already moved to agents so the ide dependency thesis is wrong

Harry Stebbings · Oct 17, 2025

The model providers have no inherent lock-in advantage, but they have enormous cash and will win by hoovering up talent and outspending everyone

Their advantage is capital, not structural lock-in

26:07 Capital and talent not lock in let labs win coding tools

Harry Stebbings · Oct 17, 2025

The usual lazy VC objection that a big incumbent 'could just do that' is more valid in AI coding because the incumbents actually are doing it

Historically 'Amazon could just do that' was always technically true but idle; here the model providers are actively shipping competing products

26:23 Incumbent could just do that objection is valid here

Zach Lloyd · Oct 17, 2025

The model providers are not winning in AI coding tools; their launches produce spikes that settle rather than durable share

Warp can see usage of Claude Code, Codex and others from where it sits in the stack, and the spikes settle down; those products are basically equivalent to each other while Warp is not

27:48 Lab launches spike then settle without durable share gain

Jonathan Ross · Sep 29, 2025

Coding tools have low durability with cutting-edge engineers who switch to the best tool instantly, but retain enduring value with enterprises locked into long-term deals

Groq's engineers rotate monthly between Sourcegraph, Anthropic and Codex; enterprises stick with whatever deal they signed a year ago

59:36 Durability differs by segment elite engineers churn tools enterprises stay locked in

Alexander Embiricos · Feb 21, 2026

Codex isn't competing directly with Lovable and Replit today, but overlap is coming as less technical users build simple things inside Codex instead of going to a specialised tool.

Less technical people are already building hello-world-type things in the Codex app, and Codex is now available on free and Go ChatGPT plans, which massively widens availability.

37:44 Lab coding tools expand downmarket into vibe coding territory