Startup Deep Dive : CodeRabbit — the AI reviewer that hit $1.5 billion in three years

A three-year-old company that has never filed a public balance sheet was valued at $1.5 billion (roughly ₹14,400 crore) in August 2026. The startup is CodeRabbit, an AI code-review tool built by two Indian-origin founders out of San Francisco, and its first real customers did not come from a sales team at all — they came from unpaid developers in Japan writing blog posts about it because they liked it.

That contradiction — a Silicon Valley unicorn-in-waiting whose product-market fit was discovered, not sold — sits at the centre of CodeRabbit’s story. This piece traces how a reliability-tooling startup that struggled to find traction folded itself into a code-review bot, rode the “vibe coding” wave of 2025-26, and went from a bootstrapped $15 million in annual recurring revenue to a $1.5 billion valuation in under twelve months, backed along the way by CRV, Nvidia’s venture arm and Atomico.

Quick facts

What they do

CodeRabbit sells a subscription tool that reads a developer’s pull request the way a strict senior engineer would, then leaves inline comments, flags bugs and security issues, writes summaries for reviewers, and can generate unit tests and docstrings — before a human ever opens the diff. It lives inside the tools developers already use: it installed itself as an app on GitHub and GitLab, ships a command-line client that plugs into Claude Code, Cursor and Gemini CLI, and more recently added “Triage” to prioritise which pull requests need attention and “Change Stack” to trace the downstream impact of a change. The buyer is engineering leadership at any company shipping code fast — from solo open-source maintainers, who get it free, to enterprises such as Nvidia, BMW and Adyen, who pay per active reviewer.

The origin

Harjot Gill was not new to startups when he built CodeRabbit. He came to the United States as a doctoral student at the University of Pennsylvania, then built and sold Netsil, an application-monitoring company, to Nutanix in 2018, staying on as Nutanix’s senior director of technology. In 2021 he left to co-found FluxNinja, a reliability-automation startup meant to help web-scale applications shed load and avoid outages. Running FluxNinja’s own remote engineering team gave him a front-row seat to a shift no one was pricing yet: his developers had started using GitHub Copilot to write code far faster than before, and the bottleneck had quietly moved downstream — pull requests were piling up in review queues, and human reviewers, overwhelmed by bigger and more frequent diffs, were rushing through them. Gill’s insight was simple and, in hindsight, well timed: if AI was about to generate code faster than humans could check it, someone would need to build the AI reviewer to match. He started CodeRabbit in early 2023 and folded FluxNinja into the new company.

The struggle years

The two years before CodeRabbit found its footing were not smooth. FluxNinja, founded in 2021, spent roughly two years building reliability and load-shedding tooling — a respectable but crowded category already dominated by well-capitalised observability incumbents — without the kind of traction that produces headlines or large rounds. Rather than keep pushing a product that was not breaking out, Gill wound FluxNinja down into CodeRabbit in early 2023, an unglamorous but concrete pivot: the company said as much itself in its own “FluxNinja joins CodeRabbit” announcement.

CodeRabbit’s own first eighteen months were bootstrapped, not funded. There was no institutional seed round of note before its Series A closed in August 2024 — getlatka’s own tracking labels the company “bootstrapped” through its early ARR growth. During that stretch CodeRabbit was competing for developer attention against a free, bundled alternative sitting inside the world’s dominant code host: GitHub’s own Copilot tooling. Instead of building a sales team to fight that head-on, the company leaned on whatever organic pull it could find — and, notably, found almost none of it in its own back yard. Its earliest wave of enthusiastic, unprompted usage came from a community on the other side of the world, in Japan, before there was a US enterprise motion to speak of.

The turning point

The turning point had a geography to it before it had a valuation. Long before Silicon Valley VCs took notice, developers on the Japanese technical-blogging site Zenn wrote more than fifty articles about CodeRabbit without being asked or paid to, according to Gill’s own account on the Evil Martians podcast in February 2026. That organic signal — a market pulling a product toward it, rather than a company pushing a product at a market — gave Gill the confidence to keep building CodeRabbit as a standalone company rather than a feature.

The numbers either side of that bet make the case. Before it: a bootstrapped tool with no major funding, competing against a free feature inside GitHub. After it: by September 2025, CodeRabbit had crossed $15 million in annual recurring revenue with 20% month-on-month growth, more than 8,000 paying customers, 100,000-plus open-source projects on its free tier, 13 million pull requests reviewed across 2 million-plus connected repositories — enough traction to close a $60 million Series B at a $550 million valuation in a single move. Less than a year later, the company said its revenue had grown “more than fivefold” year-on-year, and it raised a $143 million Series C at a $1.5 billion valuation in August 2026.

The money behind it

CodeRabbit has raised roughly $231 million in total across three disclosed institutional rounds, climbing from an unnamed seed stage to a reported $1.5 billion valuation in a little over two years:

• Series A — $16 million, August 2024 (₹154 crore): led by CRV, with Flex Capital and Engineering Capital; angel investor Olivier Pomel (CEO of Datadog) also put in money, and CRV general partner Reid Christian joined the board.
• Series B — $60 million, September 2025 (₹576 crore), $550 million post-money valuation: led by Scale Venture Partners, with new investor NVentures (Nvidia’s venture arm) and returning backers CRV, Harmony Partners, Flex Capital, Engineering Capital and Pelion Venture Partners. Total raised to date at this point: $88 million.
• Series C — $143 million, August 2026 (₹1,373 crore), $1.5 billion post-money valuation: led by Atomico and Smash Capital, joined by BMW i Ventures, Datadog, Hirtle Callaghan and Scenic Management, plus returning investors CRV, Scale Venture Partners, Flex Capital, Pelion Venture Partners, Harmony Partners and Engineering Capital.

What each backer changed: CRV’s Series A lead gave the company its first institutional board seat and its early security-feature roadmap funding; Nvidia’s NVentures stake, arriving at Series B, aligned CodeRabbit with the infrastructure layer for AI coding tools; and Atomico and Smash Capital’s Series C lead came bundled with an international mandate — the round funded a formal push into Europe and Japan, the market that had organically adopted the product first. Datadog’s participation as both an early angel (via Pomel) and a Series C institutional investor is a notable repeat bet from a company in an adjacent developer-tools category.

How it makes money

CodeRabbit runs a straightforward per-seat SaaS model layered on top of a workflow it makes deliberately free to enter:

• Free tier: unlimited public and private repositories for individuals, and the full Team-tier feature set at no cost for open-source projects — the funnel that produced the Japan-driven organic growth and the 100,000-plus OSS projects on the platform (September 2025).
• Paid tiers (annual billing): Essentials at $24 per developer per month, Team at $48, Advanced at $72, and a custom-priced Enterprise tier; monthly billing carries a premium (Essentials $30, Team $60 per developer per month).
• The metering unit that most people get wrong: CodeRabbit does not bill per engineer on the roster — it bills per developer who actually opens a pull request in a given month. A 40-person engineering team where only 28 people opened a PR that month is billed for 28 seats, but an occasional contributor who opens one PR still counts as a full seat for that month.
• Where the margin sits: the core cost line is inference — every review calls out to large language models — so gross margin depends on how efficiently CodeRabbit’s retrieval-augmented pipeline scopes each review rather than sending whole repositories through a model on every pull request; the company has not published a gross-margin figure.
• Newer, less-metered lines: the CLI, Triage, Change Stack and CodeRabbit Security products extend the same subscription relationship without a disclosed separate price list, effectively bundling more of the pre- and post-merge workflow into the same per-seat fee.

The numbers

CodeRabbit is a private, US-incorporated company and does not publish a statutory profit-and-loss account, so there are no audited revenue or profit figures of the kind an Indian-listed or MCA-filing company would carry. What is publicly verifiable is the sequence of funding rounds, the valuations attached to them, and the ARR figure disclosed around the Series B. The table below uses those verified checkpoints, with rupee-crore equivalents shown once using the stated conversion rate.

No profit or loss figure has been disclosed at any point, which is typical for a venture-backed company still in a growth-over-margin phase; readers should treat the ARR and revenue-growth figures as company-stated and reported by independent outlets covering the same funding announcements, not as audited numbers.

Where the money comes from

CodeRabbit has not disclosed a formal revenue split by geography or segment, but the public data points to a reasonably clear shape:

• Customer base: grew from 8,000-plus paying organisations (September 2025) to 17,000-plus (August 2026) in under a year, per the company’s own funding announcements.
• Free, non-revenue layer: more than 100,000 open-source projects use the platform at no cost — a distribution funnel rather than a revenue line, but the base from which paying customers are believed to convert.
• Named enterprise logos: Nvidia, BMW, Adyen, Groupon, Chegg and Mercury are cited in company and press material as customers, though none of these is broken out as a percentage of revenue.
• Geographic pull came from outside the US first: the company’s earliest organic traction was Japanese, not American — more than fifty unprompted Zenn.dev articles predated any formal Japan go-to-market effort — and the Series C proceeds are earmarked in part for a formal expansion into Europe and Japan.
• Product-line bundling, not line-item pricing: the core PR-review subscription, the newer CLI, Triage, Change Stack and Security products are all sold inside the same per-seat tiers rather than priced and reported separately.

The surprise is the geography: a company headquartered in San Francisco, selling into the world’s largest software market, first proved its product-market fit through an unpaid community in Japan — evidence, in Gill’s own framing, that a market pulling a product needs no design partners, only a fast feedback loop.

The risks

• Its biggest channel is also its biggest potential competitor. CodeRabbit’s distribution runs through GitHub Marketplace and more than 2 million connected repositories, but GitHub itself ships Copilot-based pull-request review inside the same workflow — the platform CodeRabbit depends on for reach could just as easily bundle away the need for a third-party reviewer.
• The valuation is running ahead of disclosed revenue. A $1.5 billion valuation against ARR that independent reporting placed at roughly $15 million as recently as September 2025 — even after the company’s own claim of “more than fivefold” revenue growth by August 2026 — implies a revenue multiple typical of the 2025-26 AI funding cycle, one that only holds up if growth keeps compounding at a similar rate.
• The pricing meter depends on a workflow that is itself changing. CodeRabbit bills per developer who opens a pull request in a month, but as AI coding agents increasingly batch work into fewer, larger changes or move toward more autonomous, less PR-centric workflows, that per-seat, per-PR meter could compress even as total code volume rises — a risk the company’s own move into “agentic change management,” announced alongside the Series C, suggests it is already trying to hedge.

The takeaway

The most transferable lesson in CodeRabbit’s story is not about AI or code review specifically — it is about where to look for proof that something is working. Gill had already built and sold one company and spent two years on a second that never broke out; the difference with CodeRabbit was not a bigger sales team or a cleverer pitch deck, but noticing that developers on the other side of the world were writing unpaid blog posts about a tool nobody had marketed to them. A market that pulls a product toward it, before a single enterprise contract is signed, is a stronger signal than any list of design partners assembled by outbound sales — the job, once that pull appears, is simply to keep up with it faster than the well-funded competitors chasing the same signal.

Frequently asked questions

What does CodeRabbit do?

It is an AI tool that reviews pull requests automatically — leaving comments, flagging bugs and security issues, summarising changes for human reviewers, and generating tests and documentation — inside GitHub, GitLab and a standalone CLI that plugs into AI coding agents.

Who founded CodeRabbit and when?

Harjot Gill and Gur Singh founded CodeRabbit in early 2023. Gill had previously sold Netsil to Nutanix in 2018 and co-founded reliability-automation startup FluxNinja in 2021, which he folded into CodeRabbit at launch.

How much money has CodeRabbit raised, and what is it worth?

CodeRabbit has raised roughly $231 million across a Series A ($16 million, August 2024), Series B ($60 million, September 2025, at a $550 million valuation) and Series C ($143 million, August 2026, at a reported $1.5 billion valuation).

How does CodeRabbit make money?

It sells per-developer-per-month subscriptions (roughly $24-$72 on annual billing, plus custom enterprise pricing), billing only developers who open a pull request in a given month, while keeping the product free for individuals and open-source projects.

Is CodeRabbit profitable or publicly listed?

No. It is a private company with no disclosed profit-and-loss figures; its scale is measured through ARR and customer-count figures released alongside funding rounds rather than audited financial statements.

Sources

Figures are as of September 2026. Currency converted at $1 ≈ ₹96.0 as of 18 September 2026 (Trading Economics).

• TechCrunch, “CodeRabbit raises $60M, valuing the 2-year-old AI code review startup at $550M”, September 2025
• Business Wire, “CodeRabbit Raises $60M Series B Following Unprecedented Growth as Vibe Coding Triggers a Need for New Code Quality Standards”, September 2025
• CodeRabbit blog, “Our $60M Series B and Quality Gates for AI Coding”, September 2025
• Business Wire, “CodeRabbit Raises $143 Million at $1.5 Billion Valuation and Introduces Agentic Change Management”, August 2026
• Tech Funding News, “CodeRabbit lands $143M at $1.5B valuation as AI-generated code surges”, August 2026
• TechCrunch, “CodeRabbit raises $16M to bring AI to code reviews”, August 2024
• CodeRabbit blog / Newsroom, “CodeRabbit Announces $16M Series-A Funding Led by CRV”, August 2024
• CodeRabbit blog, “FluxNinja joins CodeRabbit for scalable AI apps”
• Evil Martians / Dev Propulsion Labs podcast, “Harjot Gill: how CodeRabbit went from viral in Japan to $60M Series B”, February 2026
• Getlatka, “CodeRabbit Revenue 2025: $15M ARR (Bootstrapped)”
• ARR.club, “CodeRabbit: CodeRabbit ARR Hit $15M”
• Sacra, “CodeRabbit revenue, valuation & funding” and “CodeRabbit vs. GitHub”
• Tracxn, “CodeRabbit – 2026 Company Profile, Team, Funding & Competitors”
• CodeRabbit docs, “CodeRabbit plans and pricing” (docs.coderabbit.ai)
• USENIX SREcon23 APAC speaker profile, Harjot Gill / FluxNinja

Found an error? Write to us and we’ll correct it in the open, dated, on the piece.