What Axie Infinity and STEPN Taught Us About Play-to-Earn, and What Skill-to-Earn Does Differently

In November 2021, around 2.5 million people logged into Axie Infinity every single day, and by April 2022 more than half of them were gone, which tells you less about the game itself than it does about where its rewards were coming from. Axie, STEPN and Thetan Arena all rose fast and fell fast for the same reason: they paid players in tokens whose value depended on new players buying in. Skill-to-earn games learn from that by paying winners out of entry stakes that already sit in a prize pool, so the payout never depends on the next player showing up.

Those numbers come from The Block's breakdown of Axie's decline, and they're worth sitting with for a minute, because a drop like that doesn't happen to a game people simply got bored of. Bored players drift away slowly over years. What happened to Axie was quicker and sharper, and the same shape showed up in STEPN and Thetan Arena within months of each other.

This article goes through what those games actually taught the industry, one lesson at a time. It doesn't set out to laugh at them, because the people who built them took a real swing at a real idea. The point is to be clear about which parts broke, so it's easier to judge the new wave of games that call themselves skill-to-earn.

Three games, one shape

Before the lessons, here's a quick look at the three games, because they look nothing alike on the surface.

Axie Infinity was a creature-battling game. You needed a team of three NFT creatures called Axies to play, and you earned a token called Smooth Love Potion (SLP) for winning battles and completing daily tasks. It became famous in the Philippines, where players who couldn't afford a team borrowed one through "scholarship" programs and split their earnings with the owner.

STEPN was a move-to-earn app. You bought an NFT sneaker, walked or ran outside, and earned Green Satoshi Token (GST) for your steps.

Thetan Arena was a mobile battle arena game, closer to a traditional free-to-play title. You could play for free, but earning required owning NFT heroes, and you earned the game's THC token through matches.

Three games, one shape

Axie Infinity

Creature battles. To earn: a team of three NFT Axies. Paid in SLP.

−99%SLP from its February 2022 peak

STEPN

Move-to-earn. To earn: an NFT sneaker. Paid in GST.

−98%GST in about two months

Thetan Arena

Team battle arena. To earn: NFT heroes. Paid in THC.

−$0.61a match for the average earner

One game was about creature battles, one was about jogging and one was a team battle arena. Different genres, different chains and different audiences, yet all three hit the same wall within roughly a year. When three very different games fail the same way, the cause is usually something they share underneath, and in this case that's the earning model.

Lesson 1: Every reward is paid by somebody

This is the lesson everything else comes from.

When Axie paid you SLP, nobody had deposited that SLP anywhere. The game created it as a reward. It only turned into money when you sold it, and the person buying it from you was, most of the time, someone who needed SLP to breed new Axies. Those new Axies were then sold to new players who needed a team to start playing. STEPN worked the same way with GST and sneakers.

So when you trace a play-to-earn payout all the way back, you find that the money came from players who joined after you. The game sat in the middle, minting tokens on one side and selling NFTs on the other, and as long as new players kept arriving the loop looked healthy.

Here's how fast it unwound once arrivals slowed:

• Axie's daily active users fell from about 2.5 million in November 2021 to about 1.2 million in April 2022, and monthly NFT sales dropped 98% from their August 2021 peak, per The Block.
• SLP fell 99% from its February 2022 peak, according to Thomson Reuters Foundation's Context.
• STEPN's GST peaked at $9.36 in April 2022 and lost about 98% of its value in roughly two months, according to CryptoSlate.

A game can be fun, well made and popular and still collapse like this, because the problem sits underneath the gameplay. If the reward is minted and the buyer is the next player, then the reward is only as strong as the game's growth.

Lesson 2: A big entry cost turns players into investors

All three games asked players to buy something before they could earn. Axie needed a team, STEPN needed a sneaker and Thetan Arena needed heroes if you wanted to earn instead of just play.

That changes how people think about the game. Once you've paid for an NFT, you're no longer just a player, you're an investor waiting for your money back. You start counting days to break even, and you only keep playing as long as the math works.

STEPN shows this clearly. CryptoSlate reported that a basic sneaker's daily earnings went from as much as $37 in April 2022 to about $0.72 by June, while the cheapest sneaker still cost around $104. That moved the break-even point from roughly 16 days to about 144 days. A new user looking at a 144-day payback on a token that was falling every week had every reason to stay away, and when they stayed away, the token fell further.

Days for a new STEPN user to break even

April 2022, at the peak16

June 2022144

Thetan Arena ran into a version of the same problem even earlier. The game was a real hit, reaching 10 million downloads in its first two months after launching in November 2021, with about 60% of them coming from Brazil, the Philippines and Indonesia, according to Deconstructor of Fun. But a simulation cited in the same analysis found that an average player was earning about $0.48 per match while spending about $1.09 per match, which means they lost roughly $0.61 every time they played to earn. High NFT prices, a falling token and earning caps made the "earn" part a losing trade for most people.

The lesson here is that when getting in costs a lot, the game's audience narrows down to people who can afford to speculate, and their patience lasts exactly as long as the returns do.

Lesson 3: Fixing a leaking economy usually means paying people less

Once these economies started to wobble, the teams behind them did what made sense on paper. They tightened things up.

Axie raised its marketplace fee from 4.25% to 5.25% and reduced how much SLP the game handed out, as Vulcan Post noted in April 2022. Scholars on the ground felt that directly. One manager interviewed by Context said the daily SLP target his scholars could hit dropped from 150 to about 50, and that the gameplay had become harder.

These were reasonable decisions, and they were probably necessary. The trouble is what they mean for a player. In a minted-token economy, the only way to slow the leak is to mint less, which means paying each player less for the same hours. Lower pay pushes more players out, fewer players means fewer buyers, and the team ends up tightening again.

That's the trap the model creates. Once the economy is leaking, every fix that protects the token also hurts the player.

The reward-cut trap

1. The economy leaksReward tokens are minted faster than anyone wants to buy them
2. The team mints lessLower daily targets, higher fees
3. Each player is paid less for the same hoursAxie's daily SLP target fell from 150 to about 50
4. Players leaveAnd with them, the buyers the token needed

↺ So it leaks again, and the team tightens again

Lesson 4: Timing beat skill

In a growing play-to-earn game, the players who earned the most weren't always the best players. They were usually the earliest ones.

Someone who joined Axie in early 2021 bought cheap Axies, earned SLP while it was climbing and could cash out at high prices. Someone who joined in early 2022, playing just as hard and maybe better, paid more to get in and earned a token that was sliding toward zero. Vulcan Post pointed at exactly this, describing how early players captured outsized rewards while later players were left with less and eventually walked away.

When the main thing deciding your earnings is when you arrived, it stops feeling like a game and starts feeling like a queue. That's also why late players carried most of the losses. They weren't less skilled or less committed. They just joined when fewer people were left behind them.

Lesson 5: Players who came only for money left with the money

This one is less about economics and more about people.

A lot of play-to-earn players weren't there because they loved the game. They were there because it paid, and for many of them, especially in places like the Philippines, the earnings genuinely mattered. When the payouts collapsed, most of those players left, because there was nothing else holding them.

Yield Guild Games, one of the biggest scholarship groups, said as much. Its Philippines manager told Context that the players who were only motivated by financial rewards had moved on. Context also reported that Axie's daily users had fallen as low as about 250,000 from a peak of 2.7 million, citing data from Cryptogambling.tv.

The lesson for any game that pays people is that the game itself has to be worth playing when you aren't winning. If the only reason to open the app is the payout, the audience disappears the moment the payout does.

Lesson 6: Where the money sits matters

The last lesson came from the worst week of Axie's life. In March 2022, the Ronin network that Axie ran on was hacked for about $625 million, which TechCrunch described as the largest DeFi hack at that point.

The hack didn't cause Axie's economic decline, since that was already months underway, but it showed how much trust players had to place in infrastructure they couldn't see. Any game that holds player money needs a clear answer to where that money sits, who can move it and what happens if something goes wrong.

So what should a better model look like?

If you put those six lessons side by side, they describe what a sustainable earning game would need. Here they are as a checklist:

1. Payouts come from money that already exists, so rewards don't depend on future players buying a token.
2. Getting in is cheap or free, so players aren't turned into investors waiting to break even.
3. The rules don't need tightening, because the economy can't leak in the first place.
4. Skill decides who earns, and timing doesn't.
5. The game is worth playing without money on the line, so people stay when they aren't winning.
6. Player money sits somewhere with clear, enforced rules about who gets it and when.

This is where skill-to-earn comes in, because it was designed with most of that list in mind.

How skill-to-earn answers the checklist

In a skill-to-earn game, there's no reward token. Players who want to compete for money put up an entry stake, usually in a stablecoin like USDC, and all the stakes go into one prize pool. The match is played, and the best performers split the pool under fixed rules, with a platform fee taken out.

That one change in where the money comes from covers most of the lessons at once. The payout is real money that players already put in, so there's nothing that needs future buyers to hold its value. The economy can't leak, because nothing is being minted, so the team never has to cut rewards to protect a token. And since the pool goes to whoever performs best in that match, the player who joined last week has exactly the same shot as the player who joined last year.

Here's the line worth remembering from all of this: play-to-earn asked the next player to pay you, and skill-to-earn asks the player you beat to pay you.

That one difference is the whole reason one model needed growth and the other only needs people to keep playing.

Six lessons, two answers

The questionPlay-to-earnSkill-to-earn

Where payouts come fromMinted tokens, bought by later playersStakes already in the pool

The cost to startAn NFT team, sneaker or heroesCheap or free

When the economy leaksRewards get cutNothing is minted, so nothing leaks

Who earns the mostThe earliest playersThe best players in that match

With no money on the linePlayers who came for money leaveA free mode tests that the game holds up

Where the money sitsInfrastructure players can't seeAn escrow with clear, enforced rules

A working example: RatPool

To make this concrete, here's how one skill-to-earn game currently being built handles each of those six lessons.

RatPool (@Ratpoolxyz) is a real-time multiplayer 2D arena shooter on Solana. Players control sewer rats in a side-scrolling arena with jetpacks, guns, grenades and a cutlass, and they score points from kills and from holding a moving zone on the map. Matches run 5 or 10 minutes and hold up to 40 players. Aim is manual, so there's no auto-aim deciding fights for you.

Going through the checklist one point at a time:

1. Payouts come from money that already exists. According to the RatPool whitepaper, players in a staked pool put up an entry in USDC, and once the room locks, the top three by score split 90% of the pool at 50%, 30% and 20%. The platform keeps 10%. The whitepaper is direct about tokens too: "RatPool has no token." So there's no SLP or GST equivalent waiting for buyers.

2. Getting in is cheap or free. You don't need to buy an NFT to start. The RatPool site says the pistol is free forever, and Free Play against bots is open at any time with no stake. Staked pools come in tiers, and on the current devnet test they start at 0.10 USDC and go up to 100 USDC, with each tier unlocked by playing two matches in the one below. Those devnet prices could change on mainnet, but the idea of a low starting tier stays the same.

3. The rules don't need tightening. Since the payout is a fixed split of a pool players filled themselves, there's no emission rate to cut and no daily earning target to lower. The split works the same for the 10th match as for the 10,000th.

4. Skill decides who earns. Only the finishing order decides the payout. The whitepaper says the store sells weapons with trade-offs but doesn't sell health, armour or a bigger share of the pool, and purchases don't change the payout formula. The site lists seven weapons, and spending money changes how you play without making you more likely to win.

5. The game is worth playing without money on the line. Free Play exists precisely so people can play for fun, practise and learn the zone without staking anything. If a player only enjoys the game when they're getting paid, that's a problem for any earning game, and keeping a free mode always open is the obvious way to test whether the game holds up on its own.

6. Player money sits somewhere with clear rules. Stakes go into an on-chain escrow program on Solana. When a match ends, the escrow pays the podium automatically, and if a match isn't settled within 30 minutes of the stakes being collected, every stake goes back in full with no fee taken, per the whitepaper. Players sign in with Google, get an embedded Solana wallet with no seed phrase, and never pay network fees, because the platform sponsors every transaction.

What the lessons still demand of skill-to-earn

It would be easy to stop there and call it solved, but the honest version of this article has to show the other side, because skill-to-earn has its own hard problems.

The first is that most players in a skill-to-earn match don't get paid. In a full RatPool room of 40, three players split the pool and 37 lose their stake for that match. That's the trade-off for real money that doesn't collapse. It means skill-to-earn is a way for good players to earn, and anyone who expects everyone to earn is going to be disappointed. The free mode matters for exactly that reason, because it lets people get good before they put anything in.

The second is fair play. If cheaters can win, the best cheater takes the pool and the model breaks. RatPool's approach is to let the room server control health, scores, the clock and the zone, check every hit against the shot that caused it, and flag only physically impossible actions like moving or firing faster than the game allows. That avoids punishing players for simply being very good, and it will need to hold up once real money is involved.

The third is trust in the result. The escrow handles the money, but it can't watch the match. The room server reports the finishing order, and the whitepaper openly says "the escrow trusts them." That's normal for real-time games, but it means the server's integrity is something players are relying on. The whitepaper also says the upgrade key is currently a single key held offline, with a multisig planned before or at mainnet.

The fourth is the law. Staking money on the result of a game, even one decided by skill, is treated differently around the world. RatPool's whitepaper notes that access to paid pools depends on jurisdiction and applicable law, so players should check where they live before playing for real money once mainnet arrives.

And finally, it's early. RatPool is live on Solana devnet with test USDC, which has no cash value, and mainnet with real USDC comes after an independent audit of the escrow and payments system. It's still a public test, which makes it a good time to try the game without anything real at stake.

The real takeaway

Here's the flag I'll plant, softly: play-to-earn didn't prove that earning from games doesn't work. It proved that you can't pay players with money that hasn't arrived yet.

Axie, STEPN and Thetan Arena each attracted millions of people, which shows the demand to earn from games is real and huge. What they got wrong was the order of things. They paid first and hoped the money would come later. Every competition that has paid people sustainably, from local tournaments to professional esports, does it the other way around, collecting the pot first and letting the result decide who takes it.

And here's a claim you can check over time. Within the next two years, any new crypto game that pays players in a token it mints for playing, without a real outside revenue source, will see that token lose most of its value once its user growth slows, the same way SLP and GST did. If that turns out wrong, it'll mean someone found a way to make minted rewards hold value without new buyers, and that would be worth learning from too.

Skill-to-earn is the model that took those lessons seriously. If you want to see what that looks like in a real game, RatPool has Free Play open at ratpool.xyz/play, the staked pools are running on test USDC during the devnet phase, and updates go out on @Ratpoolxyz.

For the full head-to-head comparison of the two models, read Play-to-Earn vs Skill-to-Earn: Which Is Better?

Quick facts

• Play-to-earn: games that pay players in tokens the game mints, with token value depending mostly on new players buying in. Examples: Axie Infinity (SLP), STEPN (GST), Thetan Arena (THC).
• Skill-to-earn: games where players put entry stakes into a shared prize pool and the best performers win it under fixed rules, with no minted reward token.
• Axie Infinity: about 2.5M daily users in November 2021, about 1.2M in April 2022; SLP down 99% from its February 2022 peak; Ronin network hacked for about $625M in March 2022.
• STEPN: GST peaked at $9.36 in April 2022 and fell about 98% in roughly two months; basic sneaker earnings fell from up to $37 a day to about $0.72.
• Thetan Arena: 10M downloads in its first two months; an early-2022 simulation found earners losing about $0.61 per match on average.
• The six lessons: rewards need a real source of money, high entry costs turn players into investors, leaking economies force reward cuts, timing shouldn't beat skill, games must be fun without money, and player funds need clear rules.
• RatPool: a skill-to-earn 2D arena shooter on Solana. No token, free Play mode, Google sign-in, sponsored fees, top 3 split 90% of the pool (50/30/20), 10% platform fee, up to 40 players, current devnet pools from 0.10 to 100 USDC (may change on mainnet).
• Status (November 2026): RatPool is live on Solana devnet with test USDC; mainnet follows an independent audit.