Turns out building a Ponzi with better graphics doesn't make it a game.
The Summary
- Over 90% of Web3 games launched during the 2021-2022 boom have shut down, with most never attracting actual gamers beyond initial token speculators
- Gaming's share of Web3 venture funding collapsed from 63% in 2022 to single digits by 2025 as capital fled to AI agents, RWA tokenization, and L2 infrastructure
- The $15 billion deployed into Web3 gaming revealed a fundamental category error: building financial instruments disguised as games, then being shocked when gamers didn't show up
The Signal
The numbers tell the story of one of crypto's most expensive learning experiences. According to Caladan's analysis, gaming absorbed $15 billion in Web3 venture capital between 2021 and 2023, peaking at 63% of all Web3 funding in 2022. By 2025, that share had cratered to single digits. More than 90% of the games launched during this period are now defunct.
The autopsy is simple: these weren't games. They were DeFi protocols wearing game skins. The core loop wasn't "do something fun, get rewarded." It was "buy tokens, recruit new buyers, extract before collapse." Axie Infinity, the poster child, peaked at 2.7 million daily active users in 2021. By 2024, it had fewer than 150,000. That's not player churn. That's a Ponzi unwinding.
"Web3 gaming confused financial speculation with player engagement, then wondered why nobody wanted to grind for tokens worth less each day."
Real gamers have jobs. They play *Elden Ring* and *Baldur's Gate 3* because those games respect their time and intelligence. Web3 games asked players to become yield farmers, tax accountants, and exit liquidity. The friction was absurd:
- Set up a non-custodial wallet
- Buy gas tokens on an exchange
- Bridge assets to a game-specific L2
- Mint an NFT character
- Start "playing" (grinding repetitive tasks for token rewards)
- Track cost basis across multiple tokens for tax purposes
- Time your exit before the token price collapses
Meanwhile, *Fortnite* requires an email address and lets you play in 90 seconds. The Web3 gaming thesis assumed people would tolerate complexity for ownership. They wouldn't. They wanted fun.
The capital rotation tells you where the smart money landed after learning this lesson. AI agents, real-world asset tokenization, and L2 infrastructure, the categories now dominating Web3 funding, solve actual problems. AI agents automate work humans don't want to do. RWA tokenization makes illiquid assets liquid. L2s make blockchains usable. These aren't financial games pretending to be products. They're products that might use tokens.
The survivors in Web3 gaming are the ones who forgot about tokens first and built games second. *Pixels*, *Shrapnel*, and a handful of others focused on actual gameplay, then added optional on-chain elements for players who cared. Their retention metrics look like game metrics, not DeFi metrics. Daily active users who return because they enjoy the experience, not because they're underwater on their NFT investment.
"The lesson cost $15 billion: you can't financialize fun, you can only ruin it."
The Implication
The Web3 gaming collapse is a preview for every other category trying to jam tokens into products that don't need them. If your core value proposition requires users to care about token prices, you're building a financial instrument, not a product. Financial instruments need different distribution, different users, and different expectations.
Watch where the $15 billion is going now. AI agents that trade, manage wallets, and automate DeFi strategies. Tokenized real estate, commodities, and private credit that unlock liquidity without requiring users to become crypto natives. Infrastructure that makes blockchains invisible to end users. The Fourth Web isn't about making everything a token. It's about using tokens where they actually create value, then getting out of the way.