The Game That Paid Wages

Axie Infinity employed a small country's worth of players, made $1.3 billion in a year, and lost 90% of its users when the pay dried up. It was the biggest test ever of paying people to use a product, and the result is final.

By AnkhLabs Research, Growth research team··5 min read

For about a year, the most successful consumer product in crypto was a job. Axie Infinity, a game where cartoon creatures battle each other, reached 2.7 million daily players at its 2021 peak, but most of them weren't really playing. They were working. Analysts at the time estimated around 55% of players were in the Philippines, where grinding the game's SLP token often paid better than local jobs, and people formed "scholarship" groups that lent starter teams to new workers for a cut of their earnings. The money was enormous, with $364 million in revenue in August 2021 alone and $1.3 billion across the year.

But look at where that revenue came from, and the machine explains its own ending. Most of it was breeding fees, paid by existing players creating new creatures to sell to new players. In 

other words, today's workers were being paid by tomorrow's workers' entry fees, and that kind of economy has one requirement, which is that it must keep growing. The moment new players stopped arriving, everything ran backwards. The wage token collapsed, eventually falling 99.8% from its peak, and the workforce did what any workforce does when the pay stops. It quit. Daily players fell about 90% from the peak, with a $624 million hack of the game's bridge speeding up an exit that was already underway.

When the reward is the product, every user is an employee, and employees quit when the pay stops.

The pattern, repeated on schedule

If Axie were one badly designed economy, it would just be a story. It's a law, and the copies prove it. StepN paid people to walk and ran the same curve at the same speed, going from over 700,000 monthly users at its May 2022 peak to about 46,000 nine months later, a 94% collapse. The whole genre followed. By 2025, quarterly investment in crypto gaming had fallen to $73 million, the sector's market value had dropped by more than two-thirds, and the standard retention stat was 60% of players gone within 30 days.

Academic research adds the precise version. A study of Axie player behavior found that money rewards do keep players around, but the effect weakens over time and is best described as temporary. That's the entire genre in one sentence. Paying users works, genuinely and briefly, at a cost that keeps growing, and the effect fades exactly when you need it most.

Why this matters far beyond gaming

Play-to-earn is the extreme version of a pattern that runs through everything we've written about. Points programs pay in promises, liquidity mining pays in token emissions, tap-to-earn pays in airdrops, and play-to-earn paid something close to a salary. The costume changes but the physics don't. A payment can attract users faster than any product improvement ever will, and it holds them only as long as the payments keep growing. The moment your growth depends on paying people, your retention has a payroll, and payrolls compound while novelty doesn't.

What makes the gaming version so valuable is that it found the ceiling. Axie wasn't a small test. It was a national labor phenomenon with a billion-dollar income statement, built by a company worth $3 billion at its peak. If paying users could ever replace wanting the product, it would have worked there. It didn't, and that settles the question at every smaller scale too.

What we'd tell a founder

Use payment to fund discovery, never routine. A reward that gets someone to try your product once is acquisition spend with a clear job. A reward that pays them to keep doing the same thing every day is a wage, and once you pay wages you're an employer, competing with every other wage your user could earn. Know which type every incentive in your product is.

Find your zero-subsidy floor before you scale. Somewhere in your data is the group of users who act as if the rewards didn't exist, using the product at full price because they want to. The size of that group is your actual business, and everything above it is payroll. An honest growth plan grows the floor, not the payroll.

Watch the direction of the money, not the volume. Axie's dashboards looked incredible while every dollar of "revenue" was new players' entry money funding old players' exits. The test is one question. If new users stopped arriving tomorrow, does the current user's experience still work? If the answer is no, your growth chart is measuring a fuse.

If you're building a game, make the game the reward. The studios still standing moved to fun-first design for the oldest reason in entertainment, which is that fun is the only retention tool with zero ongoing cost. It's also the only one a competitor can't steal by offering a higher wage, because it isn't a wage.

The most expensive lesson, fully paid for

Someone was always going to run this experiment. What happens if you simply pay people, in real money, at massive scale, to use your product? Crypto ran it with millions of workers and billions of dollars, and it bought the industry a final answer. Payment rents behavior and products earn it. Every incentive budget written since is either spending with that result in mind, or paying to learn it again.

The Game That Paid Wages

Axie Infinity employed a small country's worth of players, made $1.3 billion in a year, and lost 90% of its users when the pay dried up. It was the biggest test ever of paying people to use a product, and the result is final.

By AnkhLabs Research, Growth research team··5 min read

For about a year, the most successful consumer product in crypto was a job. Axie Infinity, a game where cartoon creatures battle each other, reached 2.7 million daily players at its 2021 peak, but most of them weren't really playing. They were working. Analysts at the time estimated around 55% of players were in the Philippines, where grinding the game's SLP token often paid better than local jobs, and people formed "scholarship" groups that lent starter teams to new workers for a cut of their earnings. The money was enormous, with $364 million in revenue in August 2021 alone and $1.3 billion across the year.

But look at where that revenue came from, and the machine explains its own ending. Most of it was breeding fees, paid by existing players creating new creatures to sell to new players. In 

other words, today's workers were being paid by tomorrow's workers' entry fees, and that kind of economy has one requirement, which is that it must keep growing. The moment new players stopped arriving, everything ran backwards. The wage token collapsed, eventually falling 99.8% from its peak, and the workforce did what any workforce does when the pay stops. It quit. Daily players fell about 90% from the peak, with a $624 million hack of the game's bridge speeding up an exit that was already underway.

When the reward is the product, every user is an employee, and employees quit when the pay stops.

The pattern, repeated on schedule

If Axie were one badly designed economy, it would just be a story. It's a law, and the copies prove it. StepN paid people to walk and ran the same curve at the same speed, going from over 700,000 monthly users at its May 2022 peak to about 46,000 nine months later, a 94% collapse. The whole genre followed. By 2025, quarterly investment in crypto gaming had fallen to $73 million, the sector's market value had dropped by more than two-thirds, and the standard retention stat was 60% of players gone within 30 days.

Academic research adds the precise version. A study of Axie player behavior found that money rewards do keep players around, but the effect weakens over time and is best described as temporary. That's the entire genre in one sentence. Paying users works, genuinely and briefly, at a cost that keeps growing, and the effect fades exactly when you need it most.

Why this matters far beyond gaming

Play-to-earn is the extreme version of a pattern that runs through everything we've written about. Points programs pay in promises, liquidity mining pays in token emissions, tap-to-earn pays in airdrops, and play-to-earn paid something close to a salary. The costume changes but the physics don't. A payment can attract users faster than any product improvement ever will, and it holds them only as long as the payments keep growing. The moment your growth depends on paying people, your retention has a payroll, and payrolls compound while novelty doesn't.

What makes the gaming version so valuable is that it found the ceiling. Axie wasn't a small test. It was a national labor phenomenon with a billion-dollar income statement, built by a company worth $3 billion at its peak. If paying users could ever replace wanting the product, it would have worked there. It didn't, and that settles the question at every smaller scale too.

What we'd tell a founder

Use payment to fund discovery, never routine. A reward that gets someone to try your product once is acquisition spend with a clear job. A reward that pays them to keep doing the same thing every day is a wage, and once you pay wages you're an employer, competing with every other wage your user could earn. Know which type every incentive in your product is.

Find your zero-subsidy floor before you scale. Somewhere in your data is the group of users who act as if the rewards didn't exist, using the product at full price because they want to. The size of that group is your actual business, and everything above it is payroll. An honest growth plan grows the floor, not the payroll.

Watch the direction of the money, not the volume. Axie's dashboards looked incredible while every dollar of "revenue" was new players' entry money funding old players' exits. The test is one question. If new users stopped arriving tomorrow, does the current user's experience still work? If the answer is no, your growth chart is measuring a fuse.

If you're building a game, make the game the reward. The studios still standing moved to fun-first design for the oldest reason in entertainment, which is that fun is the only retention tool with zero ongoing cost. It's also the only one a competitor can't steal by offering a higher wage, because it isn't a wage.

The most expensive lesson, fully paid for

Someone was always going to run this experiment. What happens if you simply pay people, in real money, at massive scale, to use your product? Crypto ran it with millions of workers and billions of dollars, and it bought the industry a final answer. Payment rents behavior and products earn it. Every incentive budget written since is either spending with that result in mind, or paying to learn it again.