Web3 Games in 2026: The Numbers to Read Before You Put a Token in Your Game

Every few months somebody in a dev Discord asks the same question, usually dressed up as a funding strategy: should we put a token in the game? Seed a community, let early holders bankroll the build, ship with a warchest instead of a Kickstarter and a prayer. It sounds like a cheat code for the hardest problem in indie development, which is paying rent while you finish the thing.

Six years of data now exists on how that went for everyone who tried it before you. It is not ambiguous, but it is also more interesting than the headlines suggest, because most of the scary numbers you have seen are measuring different things and getting quoted as if they were the same thing.

The graveyard, charted

The most cited count comes from CoinGecko's GameFi research, which tracked 2,817 web3 games launched between 2018 and 2023. By the time they ran the numbers, 2,127 of them were dead. That is a 75.5% failure rate across the whole period.

Broken out by launch year, the pattern tells you something the headline number does not:

Web3 game failure rate by launch year (CoinGecko, 2,817 games)

2018  ███████████████████████████████████░░░░░░░░░░░░░  72.7%
2019  █████████████████████████████████████████████░░░  94.3%
2020  █████████████████████████████████████████████░░░  94.2%
2021  ██████████████████████░░░░░░░░░░░░░░░░░░░░░░░░░░  45.9%
2023  ██████████████████████████████████░░░░░░░░░░░░░░  70.7%

Overall: 2,127 of 2,817 dead (75.5%)

2021 is the outlier, and not because 2021 games were better. It is the bull run. A project only counts as failed once its active users fall 99% or more from peak on a 14 day moving average, and money sloshing through the sector kept a lot of otherwise dead projects twitching. 2022 is missing from that chart on purpose: CoinGecko counted 742 games going inactive that year against 693 launched, which is more than 100% and tells you the count is games that died during the year, not games from that year's cohort. Worth knowing before you quote it at someone.

The more recent number is uglier. Research firm Caladan put the figure at 93% dead across 2020 to 2026, with $12bn to $15bn of invested capital effectively written off. Their bar for "dead" is looser (extremely low activity, no fixed threshold), so treat 75% and 93% as a range produced by two different definitions rather than proof the situation got 18 points worse.

The money left before the players did

Player numbers decayed slowly. Funding fell off a cliff.

Web3 gaming venture funding, same quarter three years apart (DappRadar)

Q2 2022  ████████████████████████████████████████  $1.6bn
Q2 2025  ██░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░  $73m   (down 93% YoY)

DappRadar's quarterly gaming reports clocked Q1 2025 investment down 71% year over year, then Q2 2025 down 93% to roughly $73m, the weakest quarter in two years. Caladan puts peak annual funding around $4bn in 2022. Whatever the exact figures, the direction has been one way for three straight years, and it matters to you specifically: the "raise on the token" plan was only ever viable while somebody was buying. That window shut.

Active wallets held up better than funding but still slid. Blockchain gaming logged about 4.8 million daily unique active wallets in Q2 2025, down 17% quarter over quarter, then 4.66 million in Q3. Gaming remains the single largest category of on-chain activity, which is the sector's genuinely strong card and gets buried under the doom.

Axie Infinity is the whole arc in two lines

Axie Infinity daily active users

Nov 2021  ████████████████████████████████████████  2,700,000
Dec 2025  ░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░     99,000

That second bar is not a rendering error. At this scale 99,000 is 3.7% of peak and rounds to nothing. Axie went from the game that made "play to earn" a phrase your relatives had heard to roughly 88,000 daily players, and Sky Mavis is shutting down Axie Infinity Classic in June 2026 to consolidate around newer products.

It is worth being fair here, because Axie gets used as a punchline more than it deserves. It still turned over about $4m in treasury revenue in 2024 and has millions of pre-registrations for its upcoming MMO. Surviving a 96% drawdown and still running a real business is not nothing. It is just not the outcome anyone was pitching in 2021.

The rug pull rate nobody actually has

Here is where most coverage falls apart. There is no clean "rug pull rate for web3 games", and anyone quoting one is almost certainly welding three unrelated statistics together.

Three different failure modes, three wildly different rates. A studio that ran out of runway and shut the servers down did not rug anyone. A token that lost 99% of its value because nobody wanted it is a market outcome, not a crime. An actual rug pull, where the team drains liquidity and vanishes, is a specific act of fraud and is the rarest of the three by a wide margin.

That distinction matters if you are the one building. The overwhelming likelihood is not that you become a scammer. It is that you become a statistic in row one: a real game, made by people who meant it, that could not hold players once the token stopped going up. The 3.59% number is the fraud rate. The 75% number is the you-tried rate.

The catch is that your players cannot tell the difference in advance, and after five years of watching row three happen they have stopped trying to. That is the actual tax on shipping a token in 2026, and it is a marketing problem, not a technical one.

What is genuinely working

The exception that gets cited most is Off The Grid, Gunzilla's battle royale, and it is instructive for the opposite reason people usually cite it. It cleared 12 million sign ups and reportedly peaked north of 500,000 daily players, and it launched on Steam in 2025 as the first blockchain-adjacent title on the platform since Valve's 2021 ban.

It managed that by making the game work completely without the chain. The Steam build ships no web3 features at all, because Valve's rules forbid it. Everything on-chain is optional and lives outside the storefront. The lesson is not "web3 games can be big". It is that the one that got big is the one where the token was a side door, not the front entrance, and where the game had to stand up as a game first. Simple titles and web2/web3 hybrids retained users; the ambitious metaverse plays did not.

The distribution problem you cannot engineer around

Even a perfectly designed token economy runs into two walls that have not moved in five years.

Steam still bans blockchain and NFT titles outright. That is not a minor storefront. For most indies it is the entire commercial plan, and the Off The Grid workaround (ship a clean build to Steam, keep the chain elsewhere) means maintaining two versions of your game forever.

And your peers are not on board. GDC's industry survey has consistently found roughly 70% of developers reporting their studios are not interested in crypto or NFTs at all, with 61% actively opposed to blockchain in games against 17% in favour. Player sentiment has drifted from openly hostile toward roughly neutral, which is an improvement on the Ubisoft Quartz era, but neutral is not demand. Nobody is waiting for your token.

So is it worth building one?

Straight answer: for the overwhelming majority of indie studios, no, and specifically not for the reason most people are considering it.

If the token is your funding plan, the plan is dead. That market closed in 2022 and the funding chart above is what closing looks like. If the token is your retention plan, six years of data says extrinsic financial rewards buy you a spike and then a cliff, because players who came for yield leave when the yield leaves, and they were never your audience. Play to earn optimised for the one thing a game cannot survive: an audience whose enjoyment is denominated in dollars.

There is a narrower case that still holds up. If you are building something where genuine asset ownership is a design pillar rather than a monetisation layer, if the game is fully playable and worth playing with the chain switched off, and if you can absorb the distribution hit from Steam and the trust hit from a burned audience, then it is a legitimate creative choice. That is a very different sentence from "we should do a token".

If you are shipping one anyway

Assume your players have been rugged before and are reading you with that in mind. Concretely:

• Make the game standalone. If the chain going offline breaks your game, you built a wallet with graphics.
• Publish the token identity properly. Network and contract address, never just a ticker. Symbols are trivially spoofable, addresses are not.
• Show the liquidity and the holder distribution. A top-ten holding 60% of supply is the single loudest rug signal, and hiding it reads worse than the number itself.
• Do not gate the fun. The moment the best way to play is to spend, you have selected for speculators.
• Say what happens when you shut down. Every studio has an ending. Having an answer ready is what separates you from the 2,127.

We built PokeIndie's Web3 games section on those assumptions. Tokens are identified by network and contract address rather than symbol, every integration goes through moderation before it renders publicly, and price data is displayed with its own history so a chart cannot quietly start on a convenient day. If you want to attach one to a listing, the integrations documentation covers the setup. And if you would rather just find games that are good without any of this attached, the main browse page is still where most of them are.

The token was never the hard part. Making something people want to play on a Tuesday night when there is nothing to earn is the hard part, and it always was.