NFT games play to earn: what the latest player data reveals

The average token among those casualties has shed about 95% from its all-time high, and projects survived about four months on average before their daily active user count slipped below 100.

The headline reads like an obituary for an industry. Beneath it sits something more nuanced than collapse. The data points less to the death of nft games play to earn mechanics than to a violent re-pricing of what those mechanics were supposed to deliver. The remaining 7% of projects, and the players still logging in, are operating under a fundamentally different social contract from the one that defined the 2021 boom.

The Great Correction: Why 93% of GameFi Projects Collapsed

The Caladan report frames the situation in blunt terms. GameFi, as a category, produced a graveyard. Of the hundreds of projects launched during the 2021–2022 speculative wave, the vast majority never built anything resembling a sustainable economic loop. They minted tokens, attracted mercenary capital, and evaporated once the emissions schedule could no longer reward the next wave of entrants.

That roughly four-month average lifespan before activity collapsed is not an accident. It closely tracks the duration of a typical liquidity-mining incentive before yields normalized and participants rotated elsewhere. The original model had an uncomfortable dependency hidden in plain sight: every participant’s expected return relied on someone else arriving later with fresh capital, fresh attention, or both.

Once that cycle slowed, the game layer could not carry the economy on its own. Many projects had technically functional NFTs, marketplaces, staking systems, and token reward loops. What they did not have was a reason to play after the spreadsheet stopped looking attractive.

Layered on top, 97% of gaming token launches in 2025 underperformed relative to their initial listing prices. This is not a story about a few bad actors or a handful of clumsy launches. It is a structural failure of incentive design combined with a market that, for a brief window, valued token velocity over user retention. When the velocity stopped, the projects stopped.

The distinction matters. Digital ownership itself was never the problem. Players have long spent money on skins, characters, battle passes, cosmetics, virtual land, and collectible items. The failure came from treating a game economy as a mechanism for distributing financial yield first, then attaching gameplay as a retention device.

The collapse was not a referendum on digital ownership. It was a referendum on emissions-funded labor markets that never priced in human attention.

The venture-capital picture tells the same story from another angle. Web3 gaming studios raised approximately $1.6 billion in Q2 2022. By Q2 2025, that figure had fallen below $18 million — a drop of more than 90%.

Capital did not disappear entirely; it concentrated. The studios still raising meaningful rounds in 2025 and 2026 are, almost without exception, those with functioning teams, shippable products, and revenue rather than whitepaper projections. Funding has not dried up so much as it has become unforgiving of ambiguity. A token allocation table is no longer mistaken for a business model. A cinematic trailer is no longer mistaken for a game.

From Farming to Gameplay: The Evolution of Player Behavior

While the project-level numbers look like a mass extinction event, the player-level numbers complicate the narrative in useful ways. DappRadar's 2025 Q4 Gaming Report documented that blockchain gaming daily active users stabilized at roughly 1.2 million across the second half of 2024 and into 2025. That figure is not impressive compared with the speculative peak, but the composition of those sessions has shifted in ways that matter more than the raw count.

In 2021, the average daily session in a blockchain game lasted 18 minutes, and approximately 70% of that time was spent on what the industry called farming: repetitive token-earning loops optimized for extraction rather than engagement. The activity was legible, measurable, and easy to promote. It was also brittle. If the token weakened, the player’s motivation weakened with it.

By 2025, the average session had stretched to 2.3 hours, and 85% of that time was spent on genuine gameplay. That does not mean every surviving Web3 title suddenly became excellent. It means the player base that remained after the speculative exodus was more willing to spend time in systems where progress, competition, social identity, collecting, and mastery mattered independently of a daily token claim.

The remaining players are not primarily grinding yield farms. They are playing games that happen to use NFTs.

That shift is more important than it sounds. The original play-to-earn proposition implied that time-in should translate directly into money-out: attention converted into tokens, then tokens converted into dollars. The newer model is less clean and, for that reason, more believable. Players may earn assets, trade scarce items, unlock utility, or participate in an economy, but the financial component is no longer supposed to be the sole justification for opening the app.

For developers, this is a harder product problem. A farming loop can be designed around rewards. A game loop has to survive boredom, competition, content fatigue, balance issues, community conflict, and the cruel fact that players can always leave for another game. But that is also the work traditional studios have been doing for decades. The correction pushed Web3 gaming closer to that discipline.

Economic Sustainability: How ARPPU Replaced Speculative Yields

The single most revealing economic indicator in the current dataset is Average Revenue Per Paying User. Under the pure play to earn crypto games model that dominated 2021, ARPPU sat at approximately $12 — a figure shaped by users who paid little or nothing while extracting whatever the emissions schedule dispensed.

Under the play-and-earn models that emerged across 2024 and 2025, ARPPU climbed to $89. The traditional mobile gaming benchmark, the standard against which blockchain games are increasingly measured, sits around $95.

That convergence is not coincidence. The studios that survived the correction recognized something their predecessors did not: sustainable virtual economies require consumers, not just participants. A user who spends on cosmetic NFTs, land utility upgrades, or in-game assets is a fundamentally different economic actor from a user who clicks a button to harvest a token for immediate sale.

The first can underwrite ongoing development, live operations, moderation, and new content. The second drains the system unless there is a permanent source of outside demand willing to absorb the sell pressure. Most early GameFi projects never solved that equation. They simply deferred it.

The difference between the two models is clearest at the point of spending:

1. Speculative yield models reward activity mainly through emissions. The player’s rational behavior is to maximize extraction, minimize attachment, and sell quickly.

2. Gameplay-led economies ask players to spend because the item, progression path, social signal, or competitive edge has value inside the game.

3. Sustainable NFT design makes secondary-market activity an extension of player demand, not the sole source of demand. A skin, weapon, character, or collectible needs to matter before it can be traded meaningfully.

4. Good sinks matter as much as rewards. If every item only produces more value and nothing leaves the system, scarcity becomes a slogan rather than an economic condition.

This shift also reframes what an in-game NFT actually is. In 2021, dominant NFT categories were often yield-bearing instruments disguised as characters or items. Their value derived from an emissions schedule. In 2025 and 2026, the categories retaining meaningful secondary-market volume look closer to traditional digital collectibles: weapon skins with real rarity curves, virtual items with aesthetic scarcity, and character NFTs whose value reflects the quality of the game around them rather than the size of the reward pool behind them.

The market learned, slowly and expensively, that interoperability without utility is just arbitrage. Moving an asset between wallets or ecosystems is technically interesting. It is not automatically meaningful to a player. An interoperable sword still needs a world in which the sword matters.

The Funding Gap and the Shift Toward Quality-First Development

The venture-capital collapse is not simply a story of money leaving. It is a story of money changing its criteria.

The studios raising in 2025 and 2026 are doing so on the basis of playable builds, signed publisher deals, and player-retention metrics rather than token allocation tables. The more than 90% drop in quarterly funding has functioned as an evolutionary filter. Projects that needed continuous capital infusions to conceal weak unit economics did not survive. Projects with working economies, even at small scale, did.

That creates a different development culture. Under the old cycle, teams could lead with a token narrative and promise that the game would arrive later. Under the new one, the game must do the heavy lifting before a token is treated as an asset rather than an obligation.

The macro numbers around the category remain substantial even after the correction. The global play-to-earn NFT games market was valued at approximately $5.40 billion in 2025, with projections pointing toward $6.37 billion in 2026. Asia-Pacific accounts for roughly 38% of that figure, driven primarily by mobile-first users in Southeast Asia and East Asia whose relationship with digital collectibles predates blockchain infrastructure entirely.

Blockchain gaming as a category also accounted for approximately 4.66 million daily active wallets in Q3 2025, representing about 25% of all active wallets across the broader Web3 ecosystem. Gaming remains, by a wide margin, the dominant consumer surface area of the on-chain economy.

But these numbers need to be read with restraint. Daily active wallets are not daily active human players. Botting, multi-wallet activity, and sybil behavior remain endemic in token-incentivized games. Any honest reading of nft game economy data has to acknowledge that a non-trivial portion of wallet activity may be automated, duplicated, or economically motivated without reflecting genuine play.

The trend lines still matter: deeper engagement, longer sessions, and higher revenue per paying user are all more useful signals than a wallet count alone. But the absolute wallet total should be treated as an upper bound, not a precise census of the people actually enjoying the game.

Overcoming the Onboarding Barrier in the Post-Speculative Era

If the play-and-earn thesis is correct, the binding constraint on growth is no longer token design. It is onboarding.

A 2025 Blockchain Game Alliance survey identified onboarding friction as the largest obstacle to adoption, cited by 51% of respondents. A further 37.2% pointed to poor gameplay as the primary issue. The latter figure is a reminder that better tokenomics cannot rescue a game that feels unfinished, repetitive, or structurally less interesting than its non-blockchain competitors.

The onboarding problem is not primarily technical anymore. Wallet creation, gas-fee subsidization, and fiat on-ramps have improved substantially since 2022. Account abstraction has also made the user-facing experience noticeably smoother. A player no longer has to understand every layer of blockchain infrastructure before trying a game.

The deeper problem is conceptual. A player arriving from a conventional title still needs to understand what an NFT does in context, why ownership should matter, and what rights actually attach to the asset they acquire. Does the item confer access? Cosmetic status? Tournament eligibility? Governance influence? Resale rights? A claim on future rewards? Too many projects have answered these questions with vague language because vague language was useful during a speculative cycle.

It is less useful now.

The studios navigating this successfully in 2026 have generally adopted one of two approaches:

• Radical simplification: hide the blockchain layer, let players interact with familiar inventory systems, and resolve ownership on-chain only when items move to a marketplace or beyond the game’s closed environment.
• Radical transparency: make the NFT layer an explicit feature, build digital identity around on-chain accounts, and market ownership or governance rights as a genuine differentiator.

Both approaches can work in isolation. Neither has conclusively proven itself at scale. Simplification lowers the barrier to entry but risks making the blockchain component invisible enough to feel unnecessary. Transparency can attract Web3-native users but may overwhelm players who simply want to install a game and start playing.

The studios that solve onboarding will not be the ones with the best smart contracts. They will be the ones who make the smart contract irrelevant to the player’s first hundred hours.

The answer is unlikely to be a universal wallet standard or a cleverer marketplace interface. It will be product design. The NFT layer has to arrive at the moment it improves a player’s experience, rather than demanding the player learn a new financial vocabulary before the game has earned their attention.

The Open Question

The data tells a coherent story about the past three years. Speculative GameFi collapsed because it conflated emissions with engagement, and engagement with economics. The survivors learned to separate those concepts.

Players are spending more time in the games that remain. Paying users are spending at levels much closer to traditional mobile benchmarks. In-game NFTs are being treated less as thinly disguised investment products and more as digital goods whose scarcity and utility depend on the quality of the game around them.

That does not mean the original promise of earning tokens in Web3 games has vanished. It means earning is no longer credible as the entire product thesis. A player may still earn, trade, collect, govern, or own. But those actions need to sit inside an experience that would retain attention even during a weak token market.

The original play-to-earn model is effectively dead. The play-and-earn model that replaced it has produced measurable improvements in engagement and revenue, while the funding environment has stabilized around a much smaller set of serious builders.

The unresolved question is whether the surviving 7% of GameFi projects can scale their p2e gaming model sustainability beyond a niche audience of Web3-native players without losing the on-chain properties that made them interesting in the first place. Traditional mobile gaming reaches billions of users. Blockchain gaming, even at its corrected high, reaches millions.

The gap is not merely a marketing problem. It is a question of whether interoperable digital identity, ownership, and programmable scarcity can survive contact with an audience that has no interest in those abstractions. The next wave of player data will matter less for what it says about token prices than for what it reveals about retention, spending, and the quiet decision every player makes: whether to come back tomorrow.

The hedging instincts that defined the 2022 speculative boom have not entirely vanished. They have migrated, and a look at how digital-asset volatility correlates with conventional hedging strategies offers a useful parallel for anyone tracking the space. What has changed is who bears the risk. In 2021, much of it lived on the balance sheets of late entrants. In 2026, it increasingly sits with development teams and the funds backing them.

Whether that distribution proves more durable is the experiment now running in real time.