Google TV Hit 300 Million Devices. A Touch-First Game Catalogue Reaches None of Them.

Google TV crossed 300 million active devices in May 2026, but a touch-first games catalogue reaches none of them — Play's TV rules demand full D-pad navigation. That one requirement — TV-DP in Google's app quality guidelines — disqualifies most casual titles licensed for web or mobile, and it is the first of half a dozen things that break on the way to a television. The 300 million number gets quoted in licensing decks as if it were addressable inventory. It isn't. Here is what actually has to be true before a licensed catalogue plays on a TV, and the one deployment shape where the work pays for itself.

📺 The 300 Million Is Real. The Curve Under It Is Not Flattering.

Google announced the milestone at I/O in May 2026: over 300 million monthly active devices across Google TV and Android TV. It is the largest single television software footprint anyone has published.

The trajectory matters more than the total. 9to5Google's write-up of the announcement lines the milestones up: 150 million in 2023, 270 million by September 2024, 300 million in May 2026. The platform added 120 million devices in one 18-month window and 30 million in the next — roughly 11% growth where it had previously more than doubled.

That is a mature installed base, not an expanding frontier. It changes what a games deal on this platform is worth. You are not buying into a land grab. You are buying access to a stock of devices that mostly already exist, run software they shipped with, and are used for watching things.

Read what Google chose to tell TV developers about in that same May 2026 post: Gemini surfacing app content, pointer-remote support, and the Engage SDK for resuming and recommending content. Three priorities, none of them a game. Developers are told to migrate off the legacy Watch Next API before the second half of 2027. The roadmap is a content-discovery roadmap. Games are tolerated on this platform, not courted.

🎮 Play's TV Rules Assume a Remote, Not a Finger

The gap between "our games are Android APKs" and "our games are on Android TV" is a specification document. Google's TV app quality guidelines set out numbered criteria, and a licensed catalogue built for phones fails several of them by default:

• TV-DP — all app functionality must be navigable with five-way D-pad controls, unless the app requires a game controller.
• TV-ML — the manifest must declare an ACTION_MAIN intent with category CATEGORY_LEANBACK_LAUNCHER. Without it the app does not appear on the TV home screen at all.
• TV-MT — the manifest must set android.hardware.touchscreen to not required. A game that declares a touchscreen as required is filtered out of the TV Play Store.
• TV-LB — a 320x180 full-size banner plus at least a 160x160 xhdpi icon, per title.
• TV-LG — a game must appear in the Games row of the launcher, which means declaring itself a game correctly.
• TV-DB — back button presses must lead back to the TV home screen.
• TV-PS — minimum SDK 31 or lower, to cover the installed base.
• TV-GF (the optimised tier) — the game must fully support both physical controllers and virtual gamepads.

TV-DP is the expensive one. Everything else on that list is manifest work, art assets and a build config — a competent engineer clears it in a day per title once the pipeline exists. D-pad navigation is a design problem.

Why "add D-pad support" is not a port task

A match-3 board, a hidden-object scene, a drag-to-aim slingshot, a drawing game, a tap-anywhere runner: these do not have a focus model. There is no ordered list of controls for a directional pad to walk through. You cannot bolt one on. You have to decide what focus means in that game — a cursor that moves at some speed, a grid selection with a confirm button, a snapping system between valid targets — and then rebalance the difficulty, because input latency changed.

Categories that survive the trip: turn-based puzzle, card games, board games, quiz and trivia, sports and racing built for controllers, anything with discrete menu-driven choices. Categories that don't: most of a hypercasual catalogue, physics-drag games, timing games tuned for a thumb, anything with a virtual joystick overlay.

If you licensed 500 Android titles, the honest question is not "can we ship these to TV". It is "which 40 of these have a focus model, and what does it cost to rebuild the other 460's input layer". Do that audit before anyone signs a TV addendum. Our Android games catalogue page is the right place to start thinking in categories rather than counts.

The pointer remote softens this, slightly

Google's 2026 push on motion-controlled pointer remotes is the one genuine movement toward touch-like input on TV. Apps declare compatibility with android.software.leanback.supports_touch in the manifest, and developers are told to implement hover states, scrollable containers and cursor clicks. A cursor is much closer to a mouse than a D-pad is, and some drag-based games would work under one.

Do not plan around it. Pointer remotes are a subset of a 300-million-device base that is mostly older hardware with a plain directional pad, and TV-DP still applies. Treat pointer support as an upside for titles you were shipping anyway, not as a route to shipping titles you otherwise couldn't.

🌐 There Is No Browser, So There Is No Web Portal

The obvious shortcut for an HTML5 catalogue is to skip the APK problem entirely: point the TV's browser at a portal URL and let the existing games run. This does not work, for a reason that surprises people every time.

Android TV and Google TV devices generally ship without a browser. Chrome is not published as an Android TV app — Google's own Google TV community threads answer this repeatedly, and the workaround people are given is sideloading a phone build, which lands you with a mobile-layout browser being driven by a directional pad. A handful of TV-oriented browsers exist on the Play Store, but you cannot build a distribution plan on the assumption that a household installed one.

The workable HTML5 route on TV is the same route as on phones: wrap the games in a native shell. A WebView container with a leanback manifest, TV banners, and a launcher entry is a legitimate product. But the wrapper does not solve input. Inside that WebView, your games still receive key events rather than touches, and every point in the section above still applies. Wrapping changes the delivery mechanism, not the playability.

⚙️ The Silicon Is a Budget Phone, and That's the Good Device

Performance assumptions carried over from mobile do not hold. Google's own current flagship streaming device, the Google TV Streamer, ships with 4GB of RAM, 32GB of storage and a MediaTek MT8696 with four Cortex-A55 cores at 2.0GHz, per 9to5Google's specification breakdown from August 2024. Its predecessor, the Chromecast with Google TV, had 2GB of RAM and 8GB of storage.

Four A55 cores is a little-core-only configuration — the efficiency cluster of a mid-range phone, with no performance cores behind it. And that is Google's reference hardware. Operator set-top boxes and the Android TV silicon inside mid-market televisions are frequently below it, with less RAM and no meaningful GPU headroom.

Two practical consequences. First, output resolution went up while compute did not: you are rendering to a 1080p or 4K panel on a chip that would be unremarkable in a $150 phone. Fill rate, not logic, is what will kill your frame rate. Second, storage is scarce and shared with streaming apps that people actually opened this device to use. A games portal app competing for space against Netflix and YouTube on an 8GB device loses that argument.

💸 Your Monetization Stack Does Not Come With You

This is the part that most often goes unexamined until after the build, and it is the one that decides whether the project has a business model at all.

AdMob does not officially support Android TV. Google's Mobile Ads SDK has never shipped TV support, there is no TV section in its documentation, and Google staff on the AdMob SDK support forum have told developers so consistently for years. It is not framed as a ban so much as an absence — which is worse for planning, because you cannot get a policy answer either way. Connected-TV inventory inside Google's stack is monetised through Ad Manager, using the IMA SDK or the lighter PAL library, both of which serve VAST video.

VAST video is a different product from what a casual portal runs on. There is no rewarded-video opt-in prompt, no MRAID playable, no banner in the corner of a puzzle board. You get pre-roll and mid-roll video, sold as CTV inventory, priced and traded like television advertising. Building for it means restructuring session flow around ad breaks rather than around player-initiated rewards.

In-app purchase is no easier. Play Billing works on TV, but the buying experience runs through a remote control: no keyboard, no autofill, no payment sheet the user has already trusted on their phone. Conversion on impulse purchases collapses when the purchase takes forty seconds of D-pad input.

Which leaves the model that actually fits: someone else already paid. Bundled, subscription-funded, or preinstalled — where the games are part of a package the household is billed for elsewhere. That is the same logic behind a subscription games portal, and it transfers to television more cleanly than any ad-funded design does.

📡 Where TV Does Work: The Operator Set-Top Box

Strip out everything above and one deployment shape survives with its economics intact — the operator-controlled box.

Android TV's Operator Tier lets a pay-TV or telecom operator ship a device with its own launcher, its own app selection, and its own billing relationship with the household. The operator decides what is preinstalled. Nobody has to find your games in a store, and nobody has to enter card details on a remote.

The direction of travel here has been visible for years, though be careful with the numbers. Rethink Research forecast in July 2021, reported by Next TV, that Operator Tier would grow from roughly 5% of the global pay-TV set-top market to about 23% by 2026, with legacy proprietary Linux boxes falling from 85% to 48% and pay-TV households declining from 921 million in 2020 to around 680 million. That is a five-year-old forecast, and forecasts of this kind routinely miss; treat it as evidence of the trend operators were committing to, not as a measured 2026 figure. The observable part — operators consolidating onto Android rather than maintaining bespoke middleware — has held.

What that means for a catalogue owner is specific:

• Curation over volume. An operator wants 30 to 60 titles that work perfectly on a D-pad, not 1,000 that mostly don't. This inverts the usual catalogue-size pitch.
• One hardware target. You are optimising for a known SoC and a known remote, which makes performance work finite instead of open-ended.
• Preinstall, not discovery. The distribution problem is solved by the deal, not by store optimisation.
• Billing already exists. The household has a monthly relationship with the operator. A games tier is a line item, not a new checkout.
• Certification is real work. Operator device programmes have their own QA gates on top of Google's, and they run on the operator's schedule.

If this is the shape you are pursuing, the conversation is a telecom conversation rather than a games-store one, and it looks a lot like the portal work described on our telecom games pages — same buyer, same procurement cycle, different screen.

🧭 The Test to Run Before You Commit

Before you extend a licence to TV, spend a week and answer four questions with evidence:

1. How many titles already have a focus model? Count the ones playable end-to-end with four directions and a confirm button. Not "could be adapted" — playable today. That number is your real TV catalogue.
2. What does one port cost? Pick the hardest of the survivors, do it properly, and measure. Multiply honestly. Anyone quoting a per-title figure before doing one is guessing.
3. What is the frame rate on the worst device in scope? Test on the actual operator box or the cheapest retail Android TV you can buy, at output resolution. Not on a development TV with a good chip.
4. Who pays? If the answer is "ads", stop and cost the CTV video path properly, because the rewarded-video economics you are used to do not exist here. If the answer is "the operator", you have a business.

The honest summary: television is a distribution channel for a small, deliberately chosen subset of a games catalogue, sold to an operator or OEM who controls the device, funded by a bundle. It is not a place to dump 1,000 licensed titles and wait for installs. Anyone selling you TV rights on the strength of the 300 million figure is selling you a number, not a channel.

🌲 Where Forestry Games Fits

We license HTML5 and Android titles across a catalogue of 1,049 games and build HTML5 games in house, which means we can tell you which titles in a shortlist are genuinely D-pad-navigable and which would need their input layer rebuilt — before a TV deal gets signed rather than after. For operator and OEM projects, that curation step is the whole job. Browse the full catalogue if you want to start assembling a shortlist, or bring us the device spec you are targeting and we will tell you what survives on it.

Next action: take the ten titles you would most want on a television, hand someone a remote control, and have them try to play each one on the device your customer actually owns. The list that survives that afternoon is your TV strategy. Everything else is a port budget you have not written down yet.