How Much Does It Cost to Build an App LikePeloton in 2026?

A lean first version of an app like Peloton comes to about 1,120 to 1,840 hours of senior work in my estimate, which at $150 to $225 an hour is $168,000 to $414,000. That's the software only: no bike, no treadmill, no studio, no music catalogue.

A fuller launch, with an Apple Watch app, TV apps, Bluetooth sensor support, programs, challenges and social features, lands at about 2,040 to 3,400 hours, or $306,000 to $765,000.

Both are my scenario estimates. They are not quotes, and they are not a measurement of what Peloton spent. Peloton is a trademark of Peloton Interactive, and nothing in this article implies any relationship with the company.

Consider a founder who wants "Peloton, but for rowing" or "Peloton, but for physical therapy" or "Peloton, but for boxing." The first page that founder reads probably says the whole thing costs somewhere around $30,000 to $150,000. The second page says something else. Neither shows its work.

So here's the method instead. I list what the real product does, using Peloton's own annual report and its own pages. I split what a v1 needs from what the company built over more than a decade. I price each feature in hours. Then I read the video bill straight off the vendors' price pages and walk through the rules that apply to a fitness app that touches heart rate data and sells subscriptions.

One thing to hold onto as you read: the cheap part of this product is the video. The expensive parts are the hours to build it and the content that goes in it, and the content side has costs nobody publishes.

The published agency ranges for a Peloton clone can't be reproduced from anything on the pages that print them, so I cite them as examples of the claim, not as evidence.

Stormotionsays development costs range from $92,050 to $145,000. That's a curiously precise floor. Precise down to the fifty dollars, in fact. But the page gives no hourly rate and no hours per feature, so there is nothing to multiply and nothing to check. Precision without inputs isn't accuracy, it's decoration.

PerfectionGeeks offers $30,000 to $50,000 for a basic MVP and $80,000 to $150,000 or more for a feature-rich version. Again, no hours and no rate.

SolGuruzgoes lower still: $10,000 to $30,000 for an MVP over three to four months, up to $80,000 to $110,000 plus for an "enterprise platform." Here's a quick sanity check on that bottom number. Three months of one full-time developer is roughly 500 hours. At $30,000, that's $60 an hour for a single person, with nobody left over for design, QA or the backend. It can be done offshore for that rate, sure. It can't also include live classes, a leaderboard and health sync.

The same goes for the numbers these pages lean on around the price: fitness app market sizes, download growth, claims that compliance "adds up to 40%" to a budget, or that maintenance is "about a third" of the build every year. I looked for a primary source behind each one and didn't find one I could fetch, so none of them appear here as fact.

This is the same refusal I made across the series. The Calm cost breakdown runs the same method on the other big wellness subscription, and a lot of the subscription and FTC material overlaps.

Peloton is a public company that spent $242.8 million on research and development in its most recent fiscal year, and that number frames everything else. Its 10-K for the year ended June 30, 2026reports R&D of $242.8 million, up from $234.2 million in fiscal 2025 and down from $304.8 million in fiscal 2024.

Total revenue was $2,446.0 million. Subscriptions brought in $1,675.6 million of that, and Connected Fitness Products (the bikes, treadmills and rowers) brought in $770.4 million.

On people, the filing reports 1,736 employees in the United States, of whom 1,673 were full-time, and 526 internationally. That's 2,262 in total by my addition. The filing spreads them across corporate offices, production studios, warehouses, retail stores and remote roles. It does not say how many are engineers, so I won't guess.

Now the arithmetic, out loud. $242,800,000 divided by $225 an hour is about 1.08 million hours in a single year. My high estimate for a lean v1 is 1,840 hours. That's roughly 0.17% of one year of Peloton's R&D at our top rate.

To be clear, R&D isn't all software hours. The 10-K says it includes personnel costs with stock compensation, facilities, contractors, tooling and prototype materials, and software platform expenses. Tooling and prototypes are hardware costs. And Peloton also capitalizes some internal-use software development, which means part of its software spend never shows up in the R&D line at all. So the ratio is a sense of scale, not a precise comparison.

Think of it like a neighborhood spin studio versus a national gym chain. Same workout. Same playlist energy. Completely different balance sheet. Nobody prices the studio off the chain's annual report, and you shouldn't price your app off Peloton's.

Here's what the product actually is, split by what a v1 needs. The 10-K describes a library of live and on-demand classes, several app memberships (App+, App One, Strength+ and Breathwrk) and approximately 5.5 million Members as of June 30, 2026. Peloton's own wearables page describes two-way sync with Apple Health and connections to Garmin Connect and Fitbit.

The second list is where most of the money went. The first list is what you're actually pricing.

I leave hardware out of every number in this article because building equipment is a manufacturing business, and Peloton's own filing shows how thin that business can be.

Look at the Connected Fitness Products line. Revenue of $770.4 million in fiscal 2026 against cost of revenue of $680.0 million. That leaves $90.4 million of gross profit, by my subtraction, which is about 11.7% of the hardware revenue. Subscriptions, by comparison, brought in $1,675.6 million against $479.3 million of cost. That's about 71.4% gross margin on the software and content side.

Why is this important to understand? Because it tells you where the value sits. The equipment is what gets people in the door. The subscription is what pays.

The 10-K also says Peloton relies on a limited number of suppliers, contract manufacturers and logistics partners for its equipment, and that some components come from a single source. Its risk factors mention product recalls, including seat post recalls on Original Series bikes. None of that is an app problem. It's a supply chain, safety and inventory problem, with its own engineers, certification testing, warehouses and returns.

So what does a founder do instead? Two options work for a v1. Build for phones and let people bring their own equipment, or support third-party equipment over Bluetooth. Many bikes and heart rate straps broadcast standard Bluetooth profiles, so an app can read cadence, power or heart rate without you making a single piece of metal. I price that Bluetooth support as a fuller launch line below, because it adds a device testing matrix.

The lean v1 adds up to 1,120 to 1,840 hours in my estimate. That assumes one cross-platform phone app for iOS and Android, a web admin for instructors, hosted video for live and on-demand classes, and a senior team. Every row is my scenario estimate, not a quote.

Let me check the sum out loud, because a table that doesn't add up is the same sin as the agency ranges. Low column: 60 + 80 + 100 + 160 + 140 + 100 + 90 + 100 + 30 + 120 + 140 = 1,120. High column: 100 + 130 + 160 + 260 + 240 + 170 + 150 + 160 + 50 + 200 + 220 = 1,840.

Now the money. 1,120 hours at $150 is $168,000. 1,840 hours at $225 is $414,000. So the lean v1 is about $168,000 to $414,000. The middle cases: 1,120 hours at $225 is $252,000, and 1,840 hours at $150 is $276,000.

The table hides the interesting part, so here's where the hours actually go.

Live classes are the biggest line, and the video is the easy part.160 to 260 hours. Pushing an instructor's camera to a streaming service is a solved problem. The hours go into everything around it: a schedule that handles time zones, a countdown lobby, reminders that fire at the right moment, what happens when the instructor's connection drops, and what a member sees when they join eight minutes late. Then the live class has to become an on-demand class when it ends, with the right title, tags and thumbnail.

The leaderboard is the feature people underestimate.140 to 240 hours. A live leaderboard means every phone in the class is sending a small stream of numbers (output, heart rate, time elapsed) to a server, and the server is ranking and sending everyone back the updated board every second or two. That's a real-time system with websockets, reconnection logic and some basic cheating protection. It also has to work for on-demand classes, where you race against the people who took the class before you. That's two versions of one feature.

Here's a thing a phone-only app has to be candid about: without a bike, the phone doesn't know your output. It knows time, and it knows heart rate if a watch or strap is connected. So a phone-only leaderboard ranks on something softer, like minutes or heart rate zones. The honest v1 decides what it ranks before anyone writes code.

Health sync is two integrations, not one. 100 to 170 hours. Apple and Google use different frameworks, different permission screens and different data models, and both have store review rules on top. I cover those in their own section below.

Subscriptions are more than a paywall. 90 to 150 hours. Free trials, store billing on both platforms, receipt validation on your server, grace periods when a card fails, and a cancel path that satisfies the FTC. Our in-app purchase implementation guide walks through the mechanics.

The admin is for instructors, not engineers. 100 to 160 hours. Somebody has to upload classes, set the schedule, tag difficulty and length, and pick the thumbnail. If that tool is painful, your content pipeline slows down, and in a fitness app the content pipeline is the product.

For the phone app itself, a cross-platform approach is usually what I'd pick here, which is how we'd approach it on our React Native development work. The native pieces (HealthKit, Health Connect, background audio) get small native modules.

What about time? 1,120 to 1,840 hours with three or four senior people working in parallel is usually about five to eight months, design and testing included. That's my estimate from the hours, not a promise.

A fuller launch adds about 920 to 1,560 hours on top of the lean v1, for a total of 2,040 to 3,400 hours. Again, my scenario estimate.

Checking the added rows: 140 + 100 + 120 + 100 + 80 + 80 + 160 + 40 + 100 = 920 low, and 240 + 180 + 200 + 160 + 140 + 140 + 260 + 80 + 160 = 1,560 high. Add the lean v1 and you get 1,120 + 920 = 2,040 and 1,840 + 1,560 = 3,400.

In dollars: 2,040 hours at $150 is $306,000, and 3,400 hours at $225 is $765,000. The fuller launch is about $306,000 to $765,000.

Three of these rows deserve a warning label.

The watch app. 140 to 240 hours. A watch app is a separate app with its own interface, its own battery limits and its own way of talking to the phone. It's also what fixes the phone-only leaderboard problem, because the watch supplies live heart rate. If your concept depends on heart rate, this row moves up into the v1. Our wearable app development guide covers Apple Watch and Wear OS in depth.

Bluetooth sensors. 120 to 200 hours. The code to read a standard heart rate strap is small. The testing is not. Every brand of strap and bike behaves slightly differently, and somebody has to physically pair each one. Budget for buying the devices, too.

TV apps. 160 to 260 hours. Workout video looks better on a big screen, and members ask for it quickly. But a TV app is navigated with a remote, not a finger, so the library and the player need a different interface. Two TV platforms means two sets of review rules.

The music row (40 to 80 hours) is small but odd. It's not the cost of the music. It's the software that shows which song is playing, credits the artist, and hides classes in countries where a licence doesn't apply. That's work you only do once you've signed licences, which brings us to the part nobody can price for you.

Syncing with Apple Health and Google's Health Connect is expected in a fitness app, and both platforms attach rules to it that shape what you can build.

On iPhone, HealthKit is the central store for health and fitness data. Your app gets nothing by default. It asks the user for permission for each data type (workouts, heart rate, active energy and so on), and the user can grant read, write, both or neither.

That per-type model has a design consequence: users can say yes to some types and no to others, and change their minds later in Settings. So every screen that shows health data needs a sensible state for "nothing came back," without nagging.

Apple's App Review Guidelines, section 5.1.3, add the business rules. You can't use HealthKit data for advertising, marketing or use-based data mining, except to improve health management or for health research with permission. And you can't store personal health information in iCloud.

On Android, Health Connectworks in a similar way. You declare each permission in your manifest (reading steps, writing weight, and so on), and there are separate permissions for reading in the background and reading history. Google's docs also say you must complete a health apps declaration in the Play Console before publishing, or users may hit an error because those data types need special approval.

Google Play's Health Connect policy goes further. It wants a clear justification for each data type, a privacy policy that explains collection, use, retention and deletion, and it lists the permitted uses: fitness and wellness, coaching, corporate wellness, medical care, research and a few others. It bars transferring or selling health data to advertising platforms or data brokers, and bars using it to serve ads.

What does that force you to build? Three things. A permissions screen for each platform that explains why you want each type. An analytics setup that keeps health values out of your ad and marketing tools entirely. And a written data map you can hand to the Play review team and later to your own lawyer. Those are in my 100 to 170 hour health line, and the data map partly lands in product management.

Published streaming prices are low enough that video barely registers at v1, and they only start to matter once thousands of members watch every day. I read every figure below from the vendor's own page on September 30, 2026. Check them again before you sign anything.

AWS IVS is Amazon's managed low-latency live streaming. You pay for input (the instructor's stream coming in) and output (members watching). Input on a Standard channel is $2.00 an hour; a Basic channel is $0.20, and Advanced channels are $0.50 for SD and $0.85 for HD. Output from North America is priced per viewer hour and gets cheaper with volume: HD is $0.072 for the first 10,000 hours a month, $0.066 up to 50,000, and down to $0.048 above 500,000. Full HD is exactly double at each tier (Amazon IVS pricing).

IVS also sells a real-time mode for interactive sessions at $0.072 per participant hour for the first 10,000 hours. That's the one to look at if you want small-group classes where the instructor sees members, not just a broadcast.

Mux covers both live and on-demand. Its rates depend on resolution, so I use 1080p. On-demand input is free, storage is $0.003 per minute per month for the first 50,000 minutes, and delivery is $0.001 per minute for the first 500,000 minutes after 100,000 free delivery minutes a month, stepping down in published volume tiers after that (Mux video pricing docs). Live input is $0.03125 a minute. Plans include a $20 a month Launch plan with $100 of credit and a $500 a month Scale plan with $1,000 of credit, and the page notes that bills around $3,000 a month and up usually move to a sales conversation with volume pricing (Mux pricing).

Cloudflare Stream prices by minutes, not bytes. Storage is $5 per 1,000 minutes stored a month, delivery is $1 per 1,000 minutes delivered, and encoding and ingress are free. Its docs note that preloading and buffering on the viewer's side counts as billable delivery, and that WebRTC billing begins October 15, 2026 (Cloudflare Stream docs).

Why does this barely matter at v1? Because a v1 doesn't have many viewers. Suppose 500 members each watch eight 30 minute classes a month. That's 500 x 8 x 30 = 120,000 minutes. On Mux, the first 100,000 are free, and the other 20,000 cost $20. On Cloudflare it's 120 thousand x $1 = $120.

Twenty dollars. Next to $168,000 of build, that's a rounding error.

So how do you pick? Not on price at this stage. I'd pick on how well the player works on the platforms you ship, whether the vendor handles live to on-demand conversion for you, and how easy it is to protect video behind a login. And whatever you pick, wrap it in a thin layer of your own code so switching later is a contained project instead of a rewrite. If you want to see the same thinking for a pure video service, the Netflix cost breakdown goes deeper on on-demand delivery.

At real traction the streaming bill becomes a line worth watching, and it's dominated by on-demand viewing, not live classes. Here's a labelled usage scenario, not a forecast.

Suppose the app reaches 20,000 paying members. It runs 80 live classes a month, 45 minutes each, and each live class averages 300 viewers. Separately, each member watches 12 on-demand classes a month at 30 minutes. The library holds 2,000 classes averaging 35 minutes.

Work out the quantities first. Live input: 80 x 45 = 3,600 minutes, which is 60 hours. Live viewing: 80 x 0.75 hours x 300 viewers = 18,000 viewer hours. On-demand: 20,000 x 12 x 30 = 7,200,000 minutes. Library: 2,000 x 35 = 70,000 minutes stored.

Put a stack together. IVS for live at HD: $120 + $1,248 = $1,368. Mux for on-demand: $6,066 delivery plus $208 storage = $6,274. Total about $7,642 a month. Swap Mux for Cloudflare and on-demand becomes $7,200 + $350 = $7,550, for a total of about $8,918. Move live to Full HD and add another $1,248, which puts the Cloudflare stack at $10,166.

Two caveats. Mux says custom contract rates with additional discounts are available above $3,000 a month, so the published rate is a ceiling, not the likely number. And these are video costs only. Your servers, database, the leaderboard's real-time traffic, email and push are extra, and they depend on architecture choices I haven't modelled here.

Now put it against revenue. Suppose the subscription is $12.99 a month, which is my placeholder, not anyone's real price. 20,000 x $12.99 = $259,800 a month in gross bookings. Apple's Small Business Program takes 15% for developers with up to $1 million in prior-year proceeds. Google Play's service fee page says that from June 30, 2026, auto-renewing subscriptions in the US, UK and EEA pay a 10% service fee plus a 5% billing fee, which is also 15%.

If every member paid through a store, 15% of $259,800 is $38,970 a month. That's about 5 times the $7,642 video bill. Read that again, because it's the most useful number in this section: at this scale, the store fee is a bigger line than all the streaming combined. That's why so many subscription apps push members toward web sign-up where the rules allow it.

Once you pass $1 million in Apple proceeds in a year, the Small Business rate no longer applies to future sales that year, so check the standard rate in your agreement before you model year two.

The biggest costs in a Peloton-style business are not software at all, and the largest of them, music, is not publicly disclosed.

Peloton's 10-K is candid about the dependency. It says the company licenses music from record labels, music publishers, collecting societies such as performing rights organizations, artists and songwriters and other rights holders, and pays them royalties for the territories where it operates. It lists dependence on third-party music licences as a risk factor, and it notes that ownership information for compositions is sometimes unavailable or incomplete.

What it doesn't say is how much. Neither do the labels or publishers. These are private deals. So I won't estimate them, and you should be suspicious of anyone who puts a confident dollar figure on "music licensing" in a quote for your app.

What that means in practice for a v1: pick one of three roads before you design anything. Negotiate licences, which takes lawyers and time and has no published price. Use a royalty-free or production music library under its own terms. Or ship without music in class and let members play their own. Each road changes the product, and the third is the only one you can price today.

Instructors and studios are the same story. The 10-K mentions production studio facilities and instructors, but it doesn't break out what they cost, and instructor pay is set by private contracts. For your own budget, price these as a content business: cameras, lighting, audio, a space, editing time and instructor fees. Get real quotes from people in your city. None of it belongs in a software estimate.

One adjacent example from our own work. We built SnapFit, a corporate fitness platform for Snapchat employees, with Fairfax Training supplying the fitness programming and instructors. The case study describes live streaming classes, an on-demand library, company-wide challenges with leaderboards, direct messaging with instructors, an HR engagement dashboard and Apple Watch integration, on React Native and Node.js with a six-person team. The lesson that carries over: the content partner owned the classes, and our hours went into the platform around them.

Two sets of federal rules shape a Peloton-like app: the FTC's Health Breach Notification Rule for the heart rate and workout data, and the subscription rules for how you sign people up and let them leave. This is not legal advice; it's what the primary pages say as of September 30, 2026.

The Health Breach Notification Rule is in force and likely covers you. The rule, at 16 CFR Part 318, applies to vendors of personal health records and related entities (FTC rule page). The FTC's business guidance says a fitness app that syncs with wearables and draws data from multiple sources is likely a vendor of personal health records, even if some users never turn syncing on.

The part that surprises founders: a breach isn't only a hack. The FTC says that disclosing unsecured, identifiable health information without the consumer's consent is a breach. Sending workout data to an analytics or ad tool without permission can count.

• Notify affected people without unreasonable delay and within 60 calendar days of discovery.
• If 500 or more people are affected, notify the FTC at the same time.
• If fewer than 500, notify the FTC within 60 days after the end of the calendar year.
• Penalties can reach $53,088 per violation under the January 2025 inflation adjustment.

What that forces you to build: a data inventory showing where health values flow, consent screens before any sharing, and a breach response plan with templates you could send within the deadline. The build hours are small (they sit inside my security review and health lines). The discipline is the cost.

Click-to-cancel is not in force, but cancellation rules still are. The FTC finalized "click-to-cancel" amendments to its Negative Option Rule in October 2024. The Eighth Circuit vacated them on July 8, 2025, days before enforcement was due to start, on procedural grounds (Fenwick summary).

The FTC then started over. On March 11, 2026 it announced an advance notice of proposed rulemaking, published in the Federal Register on March 13, with a 30 day comment window (FTC press release). An advance notice has no rule text. Before that, on February 12, 2026, the FTC published a notice revising the Negative Option Rule to conform it to federal court decisions, per the same rule page. As of September 30, 2026, the FTC's Negative Option Rule page lists nothing after it.

Don't read that as a free pass. The Restore Online Shoppers' Confidence Act still requires clear disclosure of terms, express consent before billing, and a simple way to stop recurring charges, and many states have their own auto-renewal laws. The practical build is the same as if click-to-cancel were live: show the price and renewal terms before the trial starts, send a reminder before the trial converts, and put cancel where people can find it. That's already in my subscription line.

What about HIPAA? A consumer fitness app selling directly to the public usually isn't a HIPAA covered entity, which is exactly why the FTC rule exists. That changes if you sell into clinics or health plans. For our part, Frenchy Digital signs a BAA with healthcare clients and doesn't hold SOC 2, ISO 27001 or HITRUST, which you should know before procurement asks.

Year two costs usage plus upkeep plus content, and only the first two can be priced from public pages. Usage you can read off the monthly bill above: about $7,642 to $10,166 a month for video in my 20,000 member scenario at published prices, before volume discounts and before servers.

Upkeep is steady and unglamorous. Apple and Google ship new OS versions every year. HealthKit and Health Connect change their permission rules. The streaming SDK ships a major version. Watch and TV platforms move too. Each of these is a small project.

My scenario for that: 40 to 60 hours a month of senior maintenance. At $150 an hour, 40 hours is $6,000. At $225, 60 hours is $13,500. So about $6,000 to $13,500 a month, or $72,000 to $162,000 a year, before new features. That's my estimate, not a benchmark.

Add them up for a rough year two in the scenario: video at $7,642 to $10,166 a month is about $92,000 to $122,000 a year, plus $72,000 to $162,000 of upkeep. That's roughly $164,000 to $284,000 of software and streaming cost, with store fees, servers, instructors, studio and music on top.

Compare that with 20,000 members at a $12.99 placeholder price, which is $3.1 million a year gross. The software is not what sinks a fitness subscription. Churn and content costs are. We go through the full ownership math in app total cost of ownership.

A real v1 of a Peloton-style app is bigger than our starter packages, and the arithmetic says so before I have to.

Our MVP development page publishes three bands: $15,000 to $25,000, $30,000 to $50,000, and $55,000 to $75,000 plus. At $150 an hour, $75,000 buys 500 hours. My lean v1 starts at 1,120 hours. So the top package covers about 45% of the smallest version of this product.

What fits inside a package is a narrower first step. For instance: on-demand classes only, one platform, a subscription with a free trial, and Apple Health sync. No live classes, no leaderboard. That's a real product, and it tests the only question that matters first: will people pay for your classes?

The terms that stay constant: a senior build rate of $150 to $225 an hour, full source code and IP transfer to you on full payment, and a 30-day post-launch warranty. We've been building since 2016, first in France and as a US company since 2019, from Los Angeles.

What we won't do is quote "Peloton" at $40,000 and discover the other 1,000 hours in change orders.

The biggest red flag in a quote for a Peloton-like app is a total with no hours behind it. Here are the others I'd look for.

• A single price that quietly includes hardware integration with your own bike. Ask which device, which protocol and who supplies it.
• A dollar figure for music licensing. Those rates are not published, so the number is a guess.
• A live leaderboard with no mention of what the phone measures without a bike or watch.
• No line for the HealthKit and Health Connect permission flows and the Play health declaration.
• Analytics or ad SDKs wired to workout data. That is a store policy problem and possibly an FTC breach.
• A subscription flow with no trial reminder and no easy cancel, justified by saying click-to-cancel was struck down. ROSCA still applies.
• No device testing budget for watches, straps and TVs.

Any one of these is fixable in a conversation. Three or more usually means the quote was built from a template, not from your product.

Several things in this article rest on less than a full primary check, and you should know which ones.

• Every hour figure is my scenario estimate. None is a measurement of a real project or of what Peloton spent.
• Music licensing, instructor pay and studio costs are not publicly disclosed, so I did not estimate them.
• Peloton's 10-K does not say how many of its 2,262 employees are engineers.
• The date of the Eighth Circuit decision comes from a law firm summary; the FTC rule page confirms a February 12, 2026 notice conforming the rule to federal court decisions and the later ANPRM, but not the decision date.
• Mux contract pricing above $3,000 a month is negotiated and not published, so my bill uses its published pay-as-you-go tiers.
• Server, database, real-time and email costs were not modelled.
• The $12.99 subscription is a placeholder for arithmetic, not a market price.
• Vendor prices were read on September 30, 2026 and change often.

Three steps, and none of them require hiring anyone.

• Decide what your leaderboard ranks on without a bike. If the answer needs heart rate, the watch app moves into v1.
• Pick your music road: licences, a royalty-free library or no music. Write it down before design starts.
• Estimate your minutes: members x classes a month x minutes per class. Multiply by $0.001 to bracket your on-demand bill.

Then take the feature list to any developer, us included, and ask for hours per feature. If they won't give you hours, you've learned something.

Time to get to work.