Case Study 2: Axie Infinity's Dual-Token Death Spiral — When Tokenomics Meet Reality

Background

Axie Infinity, developed by Vietnamese studio Sky Mavis, was the poster child of the "play-to-earn" (P2E) movement. At its peak in November 2021, Axie had 2.7 million daily active users, a market capitalization exceeding $10 billion for its governance token (AXS), and generated more revenue than any other blockchain application. In the Philippines alone, tens of thousands of players earned more from playing Axie Infinity than they could from minimum-wage employment.

Eighteen months later, daily active users had dropped below 400,000, the in-game economy had collapsed, and the game's utility token (Smooth Love Potion, or SLP) had lost over 99% of its value. The Axie Infinity story is the most instructive case study in tokenomics failure in the history of cryptocurrency — not because the team was incompetent, but because the tokenomics contained a structural flaw that was invisible during growth and catastrophic during contraction.

The Dual-Token Model

Axie Infinity used two tokens:

AXS (Axie Infinity Shards): The governance and staking token. Fixed maximum supply of 270 million. Used for governance votes, staking rewards, and as a premium currency in the marketplace. AXS was the "store of value" token designed to appreciate.

SLP (Smooth Love Potion): The utility token. Unlimited supply — new SLP was minted every time a player won a battle or completed a quest. SLP was required to breed new Axies (the in-game creatures). SLP was the "medium of exchange" token designed to be used and circulated.

The two tokens interacted through the breeding mechanism: - Breeding a new Axie required burning both SLP and AXS - The SLP cost increased with each subsequent breed (first breed: 300 SLP; seventh breed: 2,100 SLP) - Breeding was the primary sink for SLP — the mechanism that removed it from circulation

This dual-token model was considered sophisticated at the time. The governance token (AXS) was insulated from in-game inflation by its fixed supply. The utility token (SLP) had a flexible supply that could expand with the game's economy. The breeding mechanism created a natural sink for SLP, theoretically balancing supply and demand.

The Growth Phase (2020-2021)

During the growth phase, the tokenomics appeared to work brilliantly:

Demand exceeded supply. New players joining the game needed Axies, which required SLP for breeding. As user growth was exponential (from 30,000 daily actives in April 2021 to 2.7 million in November 2021), demand for SLP consistently outstripped the rate at which existing players earned it. SLP's price rose from $0.03 in early 2021 to $0.40 in July 2021 — a 13x increase.

The play-to-earn economy thrived. At $0.40 per SLP, a skilled player could earn 150-200 SLP per day, translating to $60-80 in daily income. In the Philippines, where the minimum wage was approximately $10 per day, this was transformative. "Scholarship" programs emerged where wealthy players ("managers") lent Axie teams to players ("scholars") in exchange for a share of SLP earnings, creating an entire shadow labor market.

The breeding market boomed. As new players needed Axies, existing players bred and sold them for hundreds or thousands of dollars each. A competitive Axie team cost $500-$1,500 to assemble. The Axie NFT marketplace generated over $3.5 billion in secondary sales volume in 2021.

Everything reinforced everything else. More players meant more SLP demand, which raised SLP's price, which attracted more players, which raised demand further. The flywheel was spinning, and observers hailed Axie as the future of gaming.

The Structural Flaw

The fatal flaw was hidden in plain sight: Axie Infinity's economy required perpetual exponential growth to sustain SLP's price.

Here is the arithmetic:

SLP inflows (daily): Every active player earned SLP through gameplay. With 2.7 million daily actives earning an average of 150 SLP per day, the daily SLP emission rate was approximately 405 million SLP. At $0.10 per SLP, this represented $40.5 million in daily new supply.

SLP outflows (daily): SLP was burned through breeding. But breeding was a one-time event for each Axie, and the Axie population was growing faster than new breeding demand. By late 2021, the daily SLP burn rate was approximately 120-150 million SLP — less than half the daily emission rate.

The deficit: Every day, approximately 250-280 million more SLP were created than destroyed. This excess supply had to be absorbed by new players entering the market and buying SLP to breed their first Axies. As long as new player inflows exceeded the daily SLP surplus, the price held. The moment new player growth slowed, the excess supply had nowhere to go.

This is a Ponzi structure — not in the fraudulent sense (Sky Mavis did build a real game), but in the mathematical sense: existing participants' earnings depended on a continuous inflow of new participants' capital. Without exponential growth, the system was in permanent deficit.

The Collapse (2022)

The unraveling followed a predictable sequence:

Phase 1: Growth Stalls (January-February 2022)

After the initial excitement subsided, user growth plateaued. The number of daily active users stabilized around 2.5 million but stopped growing. The game's novelty had worn off, competition from other P2E games fragmented attention, and the $500+ entry cost became a barrier to new players as the broader crypto market cooled.

With growth stalled, daily SLP emissions still exceeded daily SLP burns. The surplus began accumulating. SLP's price, already declining from its $0.40 peak, fell below $0.03.

Phase 2: The Earning Crisis (March-April 2022)

At $0.03 per SLP, a player earning 150 SLP per day made $4.50 — less than the Philippines' minimum wage. The entire economic proposition of "play-to-earn" collapsed. Scholars began leaving scholarship programs. Managers stopped recruiting new scholars. The player base contracted.

But the contraction made the problem worse, not better. Fewer players meant less demand for new Axies, which meant less breeding, which meant less SLP burned. The daily SLP surplus — the gap between what was earned and what was burned — actually widened as the player base shrank. This is the death spiral: contraction does not cure the supply imbalance; it accelerates it.

Phase 3: The Ronin Bridge Hack (March 29, 2022)

In the midst of the economic decline, Axie Infinity's Ronin bridge was hacked for $620 million — the largest DeFi hack in history at the time. The hack, attributed to the North Korean Lazarus Group, was not directly related to the tokenomics failure, but it destroyed remaining confidence in the ecosystem. AXS dropped 25% in the days following the hack disclosure.

Phase 4: Emergency Tokenomics Redesign (April-August 2022)

Sky Mavis attempted several interventions:

Reduced SLP emissions. In February 2022, Sky Mavis dramatically cut SLP earnings from the Adventure (PvE) mode, reducing daily emissions per player. This was necessary but painful — it directly reduced players' income, further accelerating player departures.

Introduced new SLP sinks. New features like crafting, runes, and charms were added that required SLP. But these sinks were insufficient to offset the accumulated supply overhang.

"Origin" relaunch. Sky Mavis launched Axie Infinity: Origin, a revamped version of the game with free-to-play elements and reduced dependence on SLP. But the damage was done — the community had moved on.

Phase 5: The Long Aftermath (2023-2025)

By early 2023, SLP traded below $0.003 — a 99.2% decline from its peak. Daily active users fell below 100,000. The scholarship economy dissolved. The Axie NFT floor price dropped from over $300 to under $2.

AXS, the governance token with a fixed supply, fared somewhat better in relative terms (declining approximately 95% from its all-time high rather than 99%), but its value was inextricably linked to the game's user base and economic activity.

The Lessons

Lesson 1: Ponzi Mathematics Cannot Be Outgrown

The fundamental lesson is mathematical, not moral. Any system where existing participants' returns depend on new participants' capital requires exponential growth to sustain itself. Exponential growth always ends. It does not matter how good the product is, how dedicated the team is, or how enthusiastic the community is. If the tokenomics require perpetual exponential growth, the system will eventually fail.

The test is simple: Can the system sustain current token prices with zero new user growth? If the answer is no, the tokenomics are structurally unsound. For Axie Infinity, the answer was unambiguously no — daily SLP emissions exceeded daily SLP burns by a factor of 2-3x even at peak user counts.

Lesson 2: Dual-Token Models Require Robust Sinks

The dual-token model is not inherently flawed, but it requires the utility token to have sinks (mechanisms that remove tokens from circulation) that are proportional to sources (mechanisms that create tokens). Axie's only meaningful SLP sink was breeding, which was a one-time action per Axie. Once a player had their team, they had little reason to breed more Axies unless they were selling to new players.

A more robust design might include: - Recurring sinks: Maintenance costs, equipment upgrades, or entry fees that require ongoing SLP expenditure - Sink scaling: Sinks that automatically increase as the player base grows (e.g., higher crafting costs when the total SLP supply exceeds a threshold) - Burn-on-transfer: A small percentage of every SLP transfer destroyed automatically

Lesson 3: Play-to-Earn Is Play-to-Earn-from-Newer-Players

The "play-to-earn" narrative implied that value was created by gameplay. In reality, value was redistributed from newer players (who bought Axies and SLP at high prices) to older players (who earned SLP through gameplay and sold it). The game did not generate external revenue sufficient to fund player earnings — the money came from other players.

This is not inherently wrong (traditional games also redistribute value from buyers to sellers in secondary markets), but the P2E narrative obscured the redistribution mechanism. Players believed they were "earning" from the game; in reality, they were earning from the next generation of players. When the next generation stopped arriving, earnings evaporated.

Lesson 4: Real Yield Must Come from Real Revenue

Sustainable token economics must eventually be funded by real revenue, not token emissions. A protocol that generates $10 million in annual fee revenue can sustainably distribute $10 million to token holders indefinitely. A protocol that distributes $100 million in annual token emissions is simply diluting existing holders unless those emissions generate at least $100 million in new protocol revenue.

For Axie Infinity, real revenue came from the 4.25% marketplace fee on Axie NFT sales. At peak volumes, this generated significant revenue. But the tokenomics distributed far more value (through SLP emissions) than the marketplace fees generated, creating a permanent subsidy gap.

Lesson 5: "Just Fix the Tokenomics" Is Harder Than It Sounds

Sky Mavis's post-crisis interventions — reducing emissions, adding sinks, redesigning the game — were all rational responses. But they came too late and faced a coordination problem: any change that reduced SLP earnings drove players away faster, which reduced demand for new Axies, which reduced SLP burns, which made the surplus worse. The system had entered a state where every intervention had the opposite of the intended effect.

This is a common pattern in failing tokenomic systems. The same mechanisms that create positive feedback loops during growth (more users leads to more demand leads to higher prices leads to more users) create negative feedback loops during contraction (fewer users leads to less demand leads to lower prices leads to fewer users). Once the flywheel reverses, there is no easy fix within the existing token framework.

Broader Implications for Token Design

The Axie Infinity case reveals several principles that apply far beyond gaming:

Model the contraction scenario. Most tokenomics whitepapers model the growth scenario in detail and ignore the contraction scenario entirely. Any token model that looks great during growth but collapses during contraction is not robust — it is fragile.

Distinguish between value creation and value redistribution. A token that redistributes value between participants (zero-sum or negative-sum) is fundamentally different from a token that distributes value created by the protocol (positive-sum). The former requires growth; the latter can sustain itself in steady state.

Beware of tokens that subsidize their own demand. When token emissions are used to pay users who are the primary source of token demand, you have a circular system that can only expand or collapse. There is no stable equilibrium.

The dual-token model isolates but does not solve. Separating the governance token from the utility token protects the governance token from in-game inflation, but it does not prevent the utility token from collapsing. And when the utility token collapses, the economic activity that gives the governance token its value also collapses. Isolation is not immunity.

Discussion Questions

  1. At what point, if any, could Sky Mavis have changed the SLP tokenomics to prevent the death spiral? What specific changes would you have made, and when?

  2. The scholarship model (managers lending Axie teams to scholars for a share of earnings) was celebrated as financial innovation in 2021 and criticized as exploitative in 2022. Was the scholarship model inherently problematic, or was it only problematic because the underlying tokenomics were unsustainable?

  3. Some argued that Axie's real value was the community and the technological innovation, not the token economics. Is it possible to separate a game's cultural value from its economic model? Can a game survive if its economy collapses but its gameplay is compelling?

  4. Compare Axie Infinity's dual-token model to MakerDAO's dual-token model (MKR + DAI). Why did MakerDAO's model survive while Axie's did not? What structural differences in the sink mechanisms explain the divergent outcomes?

  5. If you were designing a play-to-earn game today, what tokenomics model would you use? Would you use a token at all, or would you distribute value through a different mechanism?

  6. The Axie Infinity story disproportionately affected low-income players in the Philippines, Venezuela, and other developing countries who could least afford the losses. Does the token designer have an ethical obligation to protect vulnerable participants, or is this the responsibility of the participants themselves?