How much does it cost to build a Chrome extension

Building a Chrome extension in 2026 typically costs $3,000 to $80,000+ in engineer time, depending on scope. A simple content-script plus popup ships for $3k to $8k. A medium extension with backend, auth, and payments runs $10k to $25k. A complex AI side-panel with OAuth integrations, vector storage, and a billing dashboard lands at $30k to $80k or higher.

The biggest cost drivers are: how much logic lives in the browser versus a hosted backend, whether you ship to Chrome Web Store under your own developer account or a managed one, and whether the extension uses an AI model (which adds both inference costs and the UX complexity of a side-panel + streaming pattern).

What actually goes into a Chrome extension

Most founders underestimate the surface area. A Chrome extension is not a single app. It is a small distributed system made of pieces that talk to each other through narrow channels.

A typical extension includes:

• manifest.json (Manifest V3, since V2 is dead): permissions, host access, service worker registration
• Content script(s): code injected into pages to read DOM, modify it, or capture selections
• Popup: the small panel that opens when you click the toolbar icon, usually a tiny React or vanilla HTML app
• Side panel (the 2024+ pattern Chrome rolled out for AI extensions): a persistent sidebar that doesn't close when you click away
• Service worker (background script in MV3): handles network calls, alarms, message passing
• Options page: where users configure the extension
• Storage: chrome.storage.sync for tiny preferences, chrome.storage.local for larger blobs, or a remote DB for anything you need server-side

If you also need accounts, payments, or AI calls, you add:

• A hosted backend (Vercel, Render, Fly, Cloudflare Workers)
• An auth layer (Clerk, Supabase Auth, Auth0)
• A payment integration (Stripe, Paddle, Lemon Squeezy)
• A model provider (OpenAI, Anthropic, or your own vector store with Pinecone / Postgres pgvector)

Most of the cost lives in stitching these together, not in any single piece.

Cost breakdown by extension tier

Here is the honest range we see when builders ship through Cadence. These are engineer-time costs (one-time build). They exclude SaaS subscriptions, model inference, and Chrome Web Store fees ($5 one-time developer registration).

Tier	Example	Engineer cost	Timeline	What you get
Simple	Content-script + popup (price tracker, dark-mode toggle, page-summary button calling a public API)	$3,000 to $8,000	1 to 3 weeks	Manifest V3, popup UI, content script, settings stored in chrome.storage, Web Store listing
Medium	Backend + auth + payments (Notion clipper with accounts, SEO audit tool with daily limits, screenshot tool with team plans)	$10,000 to $25,000	3 to 6 weeks	Everything above + hosted API, user accounts, Stripe checkout, usage limits, admin dashboard
Complex	AI side-panel + OAuth integrations (LLM assistant pulling from Gmail / Drive / Slack, vector search across user history, multi-tenant billing)	$30,000 to $80,000+	8 to 16 weeks	All of the above + side-panel UX with streaming, OAuth to 1 or more SaaS APIs, vector storage, rate limiting per plan, observability

What pushes you up a tier is rarely the visible UI. It is the boring stuff. OAuth flows for Google or Microsoft cost real time. Streaming an LLM response into a side-panel without dropping tokens during a tab switch is a 3-day problem the first time you hit it. Handling Chrome Web Store rejection on a third resubmit is the kind of thing you only price correctly after you have done it before.

Cost breakdown by approach

Approach	Cost	Timeline	Pros	Cons
US full-time hire	$130k to $200k+ per year	6 to 12 weeks to first ship (after a 4 to 8 week hiring loop)	Deep ownership, long-term commitment	Heaviest fixed cost; over-resourced unless you ship extensions continuously
Dev agency (US / EU)	$25k to $120k per project	8 to 16 weeks (with sales cycle)	Project management included, predictable SOW	High markup (40 to 60% over engineer cost), waterfall feel, slow change requests
Freelancer (Upwork / Fiverr)	$1.5k to $20k per project	2 to 10 weeks (highly variable)	Cheap top-line price	Hit or miss; review time can balloon if the manifest is wrong; rework risk
Toptal	$80 to $200/hr (effectively $12k to $40k for a 3-week build)	1 to 2 weeks to match, then 3 to 8 weeks of work	Vetted talent, US-quality	No AI-native baseline, higher hourly rate, long-form contracts default
Cadence	$500 to $2,000/week per engineer	48-hour trial, then ship in 2 to 8 weeks	Every engineer AI-native by default, weekly billing, replace any week, no notice	Less suited to fixed-bid enterprise procurement

The reason the Cadence row reads cheaper is the billing unit. A mid-tier engineer at $1,000 per week shipping a medium extension in 4 to 6 weeks lands the whole thing for $4k to $6k. The agency price for the same scope is usually 4x to 8x, mostly because the agency carries a sales team, project manager, and bench.

Feature-by-feature cost in 2026

If you want to estimate your own build, price the parts and add them up. Engineer hours below assume an AI-native engineer using Cursor or Claude Code; legacy estimates from 2022 era benchmarks are 2 to 3x higher.

Feature	Build cost (engineer time)	Recurring cost
Manifest V3 + popup + content script (basic)	$1,500 to $3,500	$0
Side-panel UI with streaming	$2,000 to $5,000	Inference (per provider)
Auth via Clerk	$1,000 to $2,500	Free up to 10k MAU, then $25/mo + $0.02 per MAU
Auth via Supabase	$800 to $2,000	Free up to 50k MAU, then $25/mo
Stripe checkout + webhook + subscription billing	$2,000 to $5,000	2.9% + $0.30 per transaction
LLM integration (OpenAI or Anthropic, single model)	$1,500 to $4,000	$3 to $15 per million tokens
Vector store with pgvector or Pinecone	$2,500 to $6,000	Postgres hosting ($0 to $99/mo) or Pinecone ($70/mo starter)
OAuth to Google APIs (Gmail / Drive / Calendar)	$2,500 to $6,000	$0 (Google OAuth is free); brand verification adds 4 to 8 weeks for restricted scopes
OAuth to Slack / Notion / Linear	$1,500 to $4,000 each	$0
Admin dashboard (user list, usage, billing status)	$2,500 to $8,000	Vercel / Render ($0 to $50/mo)
Chrome Web Store submission + first-pass review fixes	$300 to $1,000	$5 one-time developer fee

Two line items consistently surprise founders. The first is OAuth verification for sensitive Google scopes (Gmail body content, full Drive access). Google's security review adds 4 to 8 weeks, sometimes longer, and requires a published privacy policy plus a recorded demo. The second is Chrome Web Store review, which typically takes 5 days for a first submission but can stretch to 2 to 3 weeks if your host_permissions are broad or the reviewer asks for justification. Plan for at least one round of changes.

We cover overlapping ground in the breakdown for the Cost to build a Google Workspace add-on since the OAuth dynamics are nearly identical.

The AI side-panel pattern (and why it shifts cost)

A new pattern has dominated AI extension shipping since Chrome 116 introduced the Side Panel API. Instead of a popup that closes the moment you click outside it, the side panel anchors to the right of the browser and persists across tabs. Tools like Perplexity, Monica, and Merlin use it. ChatGPT's own extension uses it.

The pattern looks simple from outside: a sidebar that streams an LLM response while the user keeps browsing. In practice it adds three categories of work:

1. Streaming through the service worker: the side-panel UI and the model call live in different contexts. You stream tokens through a message-passing channel without dropping data when the user switches tabs. Cost: 2 to 4 engineer-days the first time.
2. Context capture from the active tab: reading the visible page, the selection, or the DOM tree into a prompt. Cost: 1 to 3 days.
3. Cross-context auth + rate limiting: the popup, side panel, and backend all need to agree on who the user is and whether they have quota left. Cost: 2 to 5 days.

This is why an AI extension that looks 30% more complex than a vanilla one is often 2 to 4x more expensive. The visible UI is a thin layer over a real distributed-systems problem.

How to reduce cost without cutting corners

A few tactics that consistently work:

• Use SaaS for commodity pieces. Clerk or Supabase for auth, Stripe for billing, Pexels or Unsplash for stock imagery. Building auth yourself for a Chrome extension is a 2-week mistake.
• Pick one model provider for v1. Adding a second adds an abstraction layer you do not need until you have customers asking for it. OpenAI and Anthropic are both fine for 95% of v1 features.
• Skip the dashboard until week 4. Most v1 admin dashboards are 3 SQL queries and a Retool board. Building a Next.js dashboard before you have 10 users is overkill. See the admin-dashboard cost breakdown for the cheapest paths.
• Defer OAuth verification. Use the test-mode flow (up to 100 users) while you validate the product. Submit for Google verification only when you are close to that ceiling.
• Ship as unlisted on the Chrome Web Store first. You can publish unlisted with no review and share the install link with beta users. Listed submission with full review can wait until v1.0.
• Book engineers by the week, not by the project. Fixed-bid agency contracts price in the unknowns. Weekly engineers price the actual work, with the option to stop at any time.

Once you have a working v1, the same logic that applies to a Shopify app build applies here: get to revenue first, harden second. The extension marketplace is unforgiving of premature optimization but rewards anything with consistent active users.

The fastest path from idea to shipped extension

Three steps:

1. Cut the spec to the minimum viable extension. One core action, one surface (popup or side panel), no auth if you can get away with it. Most successful extensions started as one feature.
2. Pick the right tier. Cleanup and integrations with good docs go to a junior engineer at $500/week. End-to-end shipping of a medium extension goes to a mid engineer at $1,000/week. AI side-panel + OAuth + billing goes to a senior at $1,500/week. Architectural ownership across multiple extensions, or a complex multi-tenant build, is a lead at $2,000/week.
3. Book the engineer and run a 48-hour trial. If you do not already have a Chrome-extension engineer, book one on Cadence and let them prove they can ship the first feature before you pay. Every engineer on Cadence is AI-native by default, vetted on Cursor, Claude Code, and Copilot fluency before they unlock bookings.

The 48-hour trial is the version of "see if they can do it" that doesn't waste a 4-week hiring loop. Cadence's median time to first commit on Chrome-extension bookings is 27 hours, and 67% of trials convert to active weekly engagements.

If you are mapping a Chrome extension build right now, the fastest path is to skip the agency RFP cycle entirely. Book a mid or senior engineer on Cadence, run the 48-hour trial against your hardest piece (usually the manifest + side-panel scaffold), and only commit to a second week once you have something installable. Weekly billing means you stop the moment the work is done.

FAQ

How long does it take to build a Chrome extension?

A simple extension ships in 1 to 3 weeks of focused engineer time. A medium build with backend, auth, and Stripe takes 3 to 6 weeks. A complex AI side-panel with OAuth integrations runs 8 to 16 weeks. Add 5 days for the first Chrome Web Store review and another 4 to 8 weeks if you need Google OAuth verification for sensitive scopes.

What tech stack should I use for a Chrome extension in 2026?

For the extension itself: Manifest V3, TypeScript, and either vanilla web components or a tiny React app for the popup and side panel. Use Vite or wxt (the most popular extension build tool in 2026) to handle bundling. For the backend, pick Next.js on Vercel, Hono on Cloudflare Workers, or Fastify on Render. For storage, use chrome.storage.local for client-side state and Postgres on Supabase or Neon for anything server-side.

Do I need a backend for my Chrome extension?

Not always. A pure content-script + popup extension that hits a public API (or runs entirely client-side) needs no backend. You need one the moment you have user accounts, paid plans, rate limits, or any logic you do not want shipped to the client. AI extensions almost always need a backend so you can keep API keys server-side and meter usage per user.

Can I build a Chrome extension solo as a non-technical founder?

For a truly simple extension (single content script, no auth, no billing), yes; AI tools like Cursor and Claude Code can scaffold a Manifest V3 extension in a day. The moment you cross into Stripe, OAuth, or side-panel streaming, the failure modes are subtle enough that you want an engineer who has shipped one before. Use the build-vs-buy-vs-hire framework to decide where the cutoff is for your specific feature set.

What are the recurring costs after I ship?

Plan for: Chrome Web Store ($0 ongoing after the $5 developer fee), hosting ($0 to $100/month on Vercel or Render for most extensions), auth ($0 to $50/month at low scale), Stripe (2.9% + $0.30 per charge), and LLM inference (variable; budget $3 to $15 per million tokens, or roughly $0.01 to $0.10 per active user per day depending on usage patterns).

When is it worth hiring full-time vs booking weekly?

If you are shipping one extension and iterating slowly, weekly is cheaper and more flexible. If you are building a portfolio of extensions or a single extension as a core product with daily shipping, full-time wins on long-term cost and ownership. Most founders we work with start weekly through the first 6 to 12 months, then convert their best engineer to full-time once product-market fit is clear.