Nobody Wants to Build for Smart TVs. That’s the Opportunity.
A few years ago I shipped a feature that worked perfectly everywhere I tested it. Chrome, the emulator, two TVs on my desk. Then the reports started coming in from a specific set of users, and they all said the same useless thing: the screen is empty.
No error. No stack trace. No console warning. Just a row of content that rendered on some televisions and silently did not exist on others.
That bug is the entire argument of this post. Not because it was hard to fix — once you know, you know — but because of what it costs to find out. And that cost is why smart TV development in 2026 is one of the few genuinely under-populated corners of front-end work, sitting on top of one of the largest pools of money in advertising.
Let’s start with the platform everyone assumes is dying
Tizen has been written off roughly once a year since 2015. Meanwhile Samsung has been the world’s number one TV brand for twenty consecutive years, with a 29.1% share of the global TV market in 2025 and over half the premium segment above $2,500. Every Samsung TV since 2015 runs Tizen.
And it’s no longer only Samsung sets. The Tizen licensing program has been putting the OS on third-party hardware — RCA, Axdia, EKO, QBELL. If you want proof this is still expanding rather than winding down, look at an unglamorous place: the TV Extension 10.0.0 release notes from February 2026, which quietly add six new licensed ODM vendors (WESTWAY, MTC, GAE, AYONZ, EL ARABY, PANDA) and remove one.
Those same release notes tell you where the platform is actually going. Tizen 10 added new web APIs that have nothing to do with playing video:
- ObjectDetection and PoseEstimation modules — on-device inference, exposed to web apps
- A VoiceInteraction module for building into the TV’s voice layer
ProductInfo.getLicensedBrand()— so your app can find out which licensed brand of TV it’s running on
Sit with that last one for a second. Samsung shipped an API whose entire purpose is to let you branch on hardware you don’t own and can’t buy at your local shop. That is not the API surface of a platform in maintenance mode.
There’s also an MCP server in the Samsung TV tooling now, sitting alongside the VS Code extension. Samsung is building for people who develop with AI agents. I don’t think many people have noticed.
The frozen engine, and the table Samsung publishes about it
Here’s the thing that makes TV different from every other web target you’ve worked on.
The browser engine inside a television is fixed at the factory. Samsung ships firmware updates for years — they’ve committed to seven years of OS updates — but those updates do not bump the Chromium major version. The TV you buy in 2026 will still be running the 2026 engine in 2033.
Samsung publishes the mapping, and it’s worth actually looking at:
| Model year | Tizen version | Web engine |
|---|---|---|
| 2026 | Tizen 10.0 | Chromium M130 |
| 2025 | Tizen 9.0 | Chromium M120 |
| 2024 | Tizen 8.0 | Chromium M108 |
| 2023 | Tizen 7.0 | Chromium M94 |
| 2022 | Tizen 6.5 | Chromium M85 |
| 2021 | Tizen 6.0 | Chromium M76 |
| 2020 | Tizen 5.5 | Chromium M69 |
| 2019 | Tizen 5.0 | Chromium M63 |
| 2018 | Tizen 4.0 | Chromium M56 |
| 2017 | Tizen 3.0 | Chromium M47 |
| 2016 | Tizen 2.4 | WebKit r152340 |
Source: Samsung’s Web Engine Specifications.
If you support the last five model years — which is normal for a streaming app — you are simultaneously targeting Chromium 130 and Chromium 85. That’s a five-year spread of the web platform, live, at the same time, with no way to polyfill your way out of the media layer.
And the spread widens every year, because the seven-year update promise keeps older sets in service rather than retiring them. Most platform commitments reduce your support burden. This one increases it.
The docs exist. They’re still not enough.
I want to be fair here, because I’ve seen this misreported (I’ve misreported it myself). Samsung does publish a feature support matrix — JavaScript, DOM, CSS, media, security, broken out per Tizen version. It’s genuinely useful and you should have it bookmarked.
But go read it closely and you’ll find entries that don’t behave the way you’d expect. Support isn’t always monotonic across model years — some features are listed as available on older platforms, absent in the middle years, and back again in 2026. Shadow DOM v0 is listed as present on several older versions and gone on Tizen 10. Template literals go from full support to partial.
Whether each of those is a documentation artifact or a real regression is exactly the point: you cannot tell from the table. The docs tell you what to suspect. They don’t tell you what happens. The only thing that tells you what happens is running the code on the television.
Which brings us back to the empty row. These failures are silent. If a missing API threw an exception, the toolchain would have absorbed this knowledge years ago and it would be a solved problem — a lint rule, a compile error, a line in caniuse. Instead the call returns undefined, your map over the result produces nothing, and the component renders an empty div. Your CI is green. Your users are annoyed.
The tooling just got good, and almost nobody re-checked
If you tried Tizen development between roughly 2018 and 2024, you used Tizen Studio, and you formed an opinion. It was an Eclipse fork with a certificate manager that fought you. I know developers who bounced off it once and never came back.
Tizen Studio is now deprecated. The replacement is a VS Code extension, and Samsung’s own writeup describes profilers, static and dynamic analysis, and debugging integrated into the editor, with a streamlined SDK install.
Don’t take Samsung’s word for it. Take the Jellyfin maintainers’, who switched their recommended install path in January 2026 and called the new extension “refreshingly simple” for certificate creation, build and deploy. That is not a phrase anyone has ever applied to Tizen tooling before.
There’s now also an official page of tips for web app development covering React and hot module reload, and Flutter support. The gap between “how I build web apps” and “how I build TV apps” has narrowed dramatically, and the reputation hasn’t caught up. That lag is the window.
One caveat worth knowing: the Tizen developer forum has been shut down. The tools improved while the community surface got thinner. Fewer people to ask, better tools to work with.
Getting an app out, and getting paid
Distribution runs through Samsung Apps TV Seller Office. You need a seller account, a .wgt package, and a certification pass covering UI compliance, API usage and performance.
A detail that trips people up and is barely discussed anywhere: when you sign up you’re a public seller by default, and public sellers can only distribute in the United States. Publishing anywhere else means requesting a partnership. Worth knowing before you plan a launch.
On money — Samsung Checkout handles subscriptions and one-off purchases. But for video, store billing is mostly beside the point. Almost every serious streaming service authenticates against an account created on the web, so no store commission is ever triggered. The real revenue on TV is advertising inside your own player, carriage on a FAST service, and placement deals. Samsung’s own numbers show where the value sits: TV Plus passed 100 million monthly active users with streaming hours up 25% year over year. The television is ad inventory now. That’s why the SDK keeps getting funded.
LG went and shipped Flutter
Tizen isn’t the only platform moving. webOS 26 landed for LG’s 2026 range, with the webOS TV 26 Simulator released in March 2026. The licensing business is enormous — webOS Hub 3.0 passed 600 brands, with LG Channels running 4,000 channels across 33 markets.
The headline for developers is that Flutter for webOS TV is now officially available through the flutter-webOS SDK. LG is rewriting its own core apps in it. When a platform owner bets its own products on a framework, that’s a stronger signal than any roadmap slide.
Amazon threw away Android, and that’s the real story
If you only take one thing from this post, take this one.
Amazon replaced Android-based Fire OS with Vega OS — a Linux platform where you build with React Native or web technologies through Vega WebView. Not React Native bolted on as an afterthought: the native code for React Native and for common dependencies like Reanimated and Gesture Handler is pre-compiled into the operating system itself. By the end of 2025 Vega was live on the Fire TV Stick 4K Select and several Echo devices, with more shipping through 2026.
Existing Android Fire TV apps don’t run on it. Amazon built a cloud-streaming stopgap to keep the big services available while publishers rebuild natively.
Line the three of them up. Samsung: Linux, web runtime. LG: Linux, web runtime. Amazon: looked at Android, and walked away to build a Linux platform with a JavaScript runtime. Three of the largest living-room platforms on earth independently arrived at the same architecture, and it is the architecture you already know.
The React and JavaScript skills you have are not transferable to TV. They are the native stack.
The money got here before the developers did
Streaming passed cable and broadcast combined in the US during 2025, and has been setting viewing-share records since. US connected TV ad spend is running around $38 billion for 2026 on eMarketer’s numbers, growing at double digits while linear declines. FAST — free ad-supported streaming — is now mainstream rather than marginal: eMarketer forecasts 131.4 million US FAST users in 2026, about 54% of all CTV users, and the global FAST channel count keeps climbing at close to 20% a year.
Roughly 1.1 billion households worldwide will own a smart TV this year. In India, the smart TV and OTT market is compounding at around 18%.
Now count the developers. Front-end web has millions of practitioners. People who can tell you what breaks on a 2022 mid-range Samsung set are a rounding error.
So why is nobody here?
Because the barriers are real, and they’re the boring kind that don’t make for good conference talks.
You need the hardware. There’s no npm run dev shortcut that tells you the truth. Emulators don’t reproduce the media pipeline, the memory ceiling, or the focus behaviour. You either own televisions or you rent time in a remote test lab. That single requirement eliminates most people before they start.
There’s no Stack Overflow long tail. The questions are too specific and the population is too small. When you hit something strange, you are often the first person to write it down.
Certification is a real schedule risk. Review cycles are measured in weeks, and rejections come back for focus handling, DRM and low-memory behaviour — things you can’t sanity-check in a browser. Plenty of side projects die at this step.
Cross-platform frameworks create false confidence. React Native reaches TV platforms with large amounts of code reuse, which is great — right up until the build fails on a four-year-old mid-range set and the team concludes “TV is hard” rather than “we didn’t know the constraints.”
And here’s the part that changed recently. AI collapsed the cost of writing the app. Scaffolding a Tizen project, wiring D-pad navigation, packaging a .wgt — that’s a conversation with a coding agent now. What AI cannot do is tell you what will silently fail on a model year it has never observed, because that information isn’t in any training set. It exists in the memory of people who ran the code on the television and watched.
The scarce input flipped. It used to be typing. Now it’s knowing what to type and what will quietly not work. That inversion is what makes this a viable niche for one person in 2026, and didn’t in 2022.
The honest counterarguments
I’d rather you hear these from me than find them later.
Tizen alone isn’t a strategy. Globally it’s a leader, but in the US market — where the ad money concentrates — Roku and Fire TV are ahead of it, with Android TV alongside. Anyone selling “Samsung TV apps” and nothing else is standing on one leg.
The volume is shifting to Chinese OEMs. TCL and Hisense combined have overtaken Samsung and LG combined on unit share, and they carry Google TV, Vidaa and whaleOS. If you index entirely on the Korean platforms, you’re indexing on the shrinking half.
Platforms can orphan your work. Hisense is rebranding Vidaa to Home OS, first reported by Lowpass and confirmed at CES 2026. Transitions like that are opportunities if you’re early and a write-off if you’re late.
Fragmentation is a genuine recurring cost. I’m not pretending it’s secretly fun. It’s a tax. The argument isn’t that the tax doesn’t exist — it’s that being one of the few people who knows how to pay it efficiently is a durable position.
Where I’d start
If any of this landed, the practical on-ramp is smaller than it looks. Pick one web app you already have. Read the web engine table and decide which model years you’ll support. Install the VS Code extension. Beg, borrow or buy one television that isn’t this year’s model — the old one is more informative than the new one.
Then ship something small and watch what breaks. The gap between what the documentation says and what the television does is where the whole opportunity lives.
Everyone else left the living room a decade ago and never checked whether the lights came back on. They did.