Warp Adds $1M ARR Every 10 Days: How Zach Lloyd Went From a Free Terminal to an Agentic Development Environment

TL;DR: Warp is an agentic development environment — a tool that began as a rebuilt terminal and is now where developers instruct agents, in English, to write features, fix bugs and run deploys. It was founded in 2020 by Zach Lloyd, a former principal engineer at Google who spent roughly eight years on Google Docs and whose first startup failed. Warp raised a $23 million Series A in April 2022 led by GV with Figma's Dylan Field, then a $50 million Series B in June 2023 led by Sequoia — both essentially pre-revenue — for roughly $73 million raised. It spent years with a large free user base and no business model. Then it repositioned as an agentic development environment, and revenue went from 300 days to earn the first $1 million of ARR to $1 million of new ARR every 10 days, as Lloyd shared on The Product Market Fit Show.

What does Warp do?

Developers live in two tools: the code editor and the terminal. Lloyd's original thesis was that the terminal — the black screen with green text — was the worse product, and that this was a product problem, not a technical one. It is enormously powerful if you know the magic incantation and useless if you do not. His diagnosis: developer tools had been built by the most hardcore developers, for themselves, optimizing for the most advanced possible configuration rather than for the person who just wants to get work done.

So Warp v1 was a terminal rebuilt in Rust with a real interface. It got hundreds of thousands of developers and almost no revenue. What Warp sells today is different: you tell it in English to build a feature, fix a bug, find out why servers are crashing or ship a deploy, and it does the work, shows the diffs, and lets you review and edit the code in the same window. Warp does not call itself a terminal anymore.

Key stat: As of its April 2026 open-source announcement, Warp reported nearly one million developers using the product, with customers including Docker, Ramp and Peloton, and usage inside more than half of the Fortune 500 (verified September 2026).

Key stat: In its 2025 year in review, Warp reported agents editing 3.2 billion lines of code, 120,000+ codebases indexed, and a 96%+ acceptance rate on agent-suggested changes (verified September 2026).

On benchmarks, Warp scored #1 on Terminal-Bench and reported 75.6% on SWE-bench Verified in that same 2025 review (verified September 2026). On the show, Lloyd named Cursor and Claude Code as Warp's competitors and called that a good thing: two fast-growing tools prove demand is real. His framing is that this is no longer the developer tools market at all, but the market for automating the production of software — closer to the market for paying software developers, and barely penetrated.

Who founded Warp?

Zach Lloyd, sole founder and CEO. His path was not straight: an undergrad mixing math, computer science and philosophy, a master's in philosophy of science, a year of law school he withdrew from, a stint in a recording studio, and a programming job answered off a Craigslist ad. He then spent about eight years at Google, entirely on Google Docs, building much of the spreadsheet product and leaving as a principal engineer running engineering for the whole Docs team. He was later interim CTO at TIME.

Google, he argues, is close to the worst possible training for founding a company.

"Google is an amazing company that sets you up in a uniquely bad way to be a startup founder." — Zach Lloyd, Warp

Anything launched at Google gets millions of people trying it by default. The hard part of a startup is not building a good product, it is getting anyone to care — a muscle he had never used in eight years there.

He left Google in 2015 to build a photo-sharing app called Camera Club. It got a few thousand users in six to nine months and went nowhere — though it was later folded into his next company for a little equity.

That next company was SelfMade, where he was co-founder and CTO for three years. It raised a seed and a Series A and reached mid-single-digit millions of revenue as a tech-enabled social media service for small businesses. But it was human-powered, operationally heavy, and hard to prove ROI on. Eventually he stopped believing in it, and his co-founder did not.

"It got to a point where I no longer believe in the mission or the viability of the business and my co-founders still did. And I ended up feeling too much cognitive dissonance for me to sell a thing to employees, and future investors, and so forth that I fully didn't believe in." — Zach Lloyd, Warp

He left at the start of 2019 and, before starting anything new, spent a couple of months writing a private post-mortem of the whole SelfMade experience. Out of it came a short list of rules: only work on a problem you actually care about, only build a pure software business, keep the team small and talent-dense. His summary of the filter: you have to care, or you have to know.

How much has Warp raised?

• Seed (2020): roughly $6 million, led by GV (Google Ventures), raised off a deck before any product existed.
• Series A (April 2022): $23 million, led by GV with Dylan Field, Figma's founder and CEO, as the lead check. On the show Lloyd described the A itself as "15 to 17" million.
• Series B (June 2023): $50 million, led by Sequoia Capital.
• Total: roughly $73 million, per funding trackers as of September 2026. Other listed backers include BoxGroup, Neo, Sam Altman, Marc Benioff, Jeff Weiner and Elad Gil.

The striking part is what the money was raised against. Asked on the show whether the $50 million Series B was pre-revenue, Lloyd said it was.

"We raised like 15 to 17 at the A, and then the B was big. It was like a $50 million Series B." — Zach Lloyd, Warp

"Again, we were growing the free user base well, but we didn't have a business model really that was working. We didn't have revenue growth." — Zach Lloyd, Warp

His read on why investors did it anyway: Warp had hard-won affection from a developer audience, and they were betting that hundreds of thousands of engaged developers become a business one way or another. It is the same pattern as Chainguard's $50M Sequoia round with no revenue and the infrastructure bets in Pinecone's story after 40 VC rejections.

How did Warp get its first users? A Hacker News post, not friends

Warp's first distribution lesson was a failure. The team spent six to nine months building something good enough to use internally, then handed it to people they knew. None of them cared. Instead of treating that as a verdict on the product, Lloyd treated it as a verdict on the sample.

"It's because your friends might not be your target first user. There's real danger in getting such a biased signal." — Zach Lloyd, Warp

So they posted on Hacker News — a new terminal, written in Rust, with a waitlist — and let strangers self-select.

Key stat: 10,000 developers signed up on the first day from that Hacker News post, as Lloyd shared on the show.

From there the loop was deliberately sequential: release cohorts off the waitlist, check whether people engage, check whether retention flattens rather than decays, and only then grow the free base. Warp reached hundreds of thousands of developers on word of mouth and a swag referral program, with no paid marketing of any kind until it had revenue.

Which left the actual problem: an excellent free product competing against a terminal that already ships free with every computer.

Why Warp stopped calling itself a terminal

Warp put AI into the product before ChatGPT existed, using Codex to turn English into terminal commands. When ChatGPT launched, the first instinct was the same one everyone had — a chat panel bolted onto the side of the app. People used it. Lloyd's team decided it was still the wrong shape.

In spring 2024 they shipped Agent Mode: instead of a panel you copy answers out of, the whole application became something you instruct in English. (Lloyd claims Warp was the first product to use the term, now industry-standard.)

That version monetized, slowly. Fixing a Git or Docker problem is genuinely valuable, and genuinely small.

Key stat: It took Warp roughly 300 days to add its first $1 million of ARR, as Lloyd shared on the show.

The unlock was going further than "AI in the terminal" and rebuilding the product for how software is now actually written — telling a computer in English what you want, then reviewing what it produced. That meant adding the parts of an IDE agentic work requires: editing code, browsing files, reviewing every diff the agent creates.

"Neither a regular terminal nor a regular IDE is like the right approach for agentic development. It's like you really want a tool that's built for this workflow." — Zach Lloyd, Warp

Warp launched the agentic development environment in June 2025, then Warp Code in September 2025 and Agents 3.0 in November 2025. In April 2026 it open-sourced the ADE and its Oz cloud-agent orchestration platform, with OpenAI as flagship sponsor (all verified September 2026).

Key stat: Warp was at roughly $1.5–2 million ARR at the ADE launch and, by the time of the interview about three months later, was adding $1 million of ARR every 10 days and closing in on $1 million a week, as Lloyd shared on the show.

That is when Lloyd says he knew.

"I think it was after this ADE launch. It was verifying that that launch wasn't a spike, because sometimes you get launch spikes and it just sort of diminishes and goes back to baseline." — Zach Lloyd, Warp

The qualitative feedback changed shape too. Before the launch, people said Warp was a much nicer terminal than iTerm. After it, they said they had built things they could not otherwise have built.

For other founders who repositioned into the AI-native version of their category, see Tessl's $125M bet on spec-driven development and Suger reaching $2M ARR with five people.

Key lessons from Zach Lloyd's playbook

1. Care or know — pick a problem you have one of. Lloyd spent three months exploring healthcare ideas with potential co-founders and got nowhere: he neither knew the industry nor cared about it. Warp came from a tool he had used every day for most of his career. 2. Your friends are the worst possible first users. They give false negatives and false positives with equal confidence. Ship where your actual users already hang out and let strangers self-select. 3. A large free user base is not a business. Warp had hundreds of thousands of developers and two large rounds before it had a working business model. Lloyd was explicit that his fallback plan was an acquisition by someone who wanted developer distribution. 4. Chat panels are the lazy version of an AI product. Warp's chat panel worked. The step change came from rebuilding the entire interaction model around agents, not bolting AI onto the old one. 5. Go where the competition is. Competing with Cursor and Claude Code signals that demand is real, not that you should find a quieter niche. 6. Write your operating principles down and publish them. Warp's "How We Work" is a public document covering hiring, product process, values, even Slack etiquette — so the arguments happen once, and candidates who hate it opt out. 7. Reason it out yourself. His one piece of advice for early-stage founders:

"Ultimately, you should be able to convince yourself that what you're doing makes sense and explain to yourself why. You'll find it's logically impossible to follow the advice of everyone you get, because the advice will be conflicting." — Zach Lloyd, Warp

FAQ: Warp

Q: What is Warp? A: Warp is an agentic development environment that grew out of a rebuilt terminal. Developers instruct it in English to write features, fix bugs, debug production and run deploys, and review the resulting diffs in the same tool. Warp open-sourced the ADE and its Oz cloud-agent orchestration platform in April 2026.

Q: Who is the CEO of Warp? A: Zach Lloyd, the sole founder. He spent about eight years at Google as a principal engineer leading engineering for Google Docs, was co-founder and CTO of SelfMade, and served as interim CTO at TIME.

Q: How much funding has Warp raised? A: Roughly $73 million — a seed of about $6 million led by GV, a $23 million Series A in April 2022 led by GV with Dylan Field, and a $50 million Series B in June 2023 led by Sequoia Capital.

Q: Who are Warp's investors and customers? A: Investors include Sequoia Capital, GV, Dylan Field, BoxGroup, Neo, Sam Altman, Marc Benioff, Jeff Weiner and Elad Gil. Warp's April 2026 announcement named Docker, Ramp and Peloton as customers, and usage inside more than half of the Fortune 500.

Q: How fast is Warp growing? A: It took about 300 days to reach the first $1 million of ARR. After the agentic development environment launched, Warp was adding roughly $1 million of ARR every 10 days, per Lloyd on the show in October 2025.

Q: What is Agent Mode in Warp? A: Agent Mode, launched in spring 2024, replaced the copy-paste chat panel: instead of typing terminal commands, you instruct Warp in plain English and it executes. Lloyd says Warp coined the term.

Sources: Listen to the Full Founder Story

• Zach Lloyd, Founder and CEO of Warp — left a principal engineer role at Google, failed at a photo app, walked away from a startup he stopped believing in, then raised $50M pre-revenue and turned a free terminal into an agentic development environment adding $1M ARR every 10 days.

Listen to the full interview: He quit Google and his 1st startup failed — but his 2nd grows at $1M ARR every 10 days.

Listen to the full episode at pmf.show for the complete story.

Last updated: September 2026