9 projects. Billions in market cap. Almost none of them make money.
The thesis is compelling: decentralize AI compute, storage, and connectivity. The reality: out of 9 projects, only 2 have revenue that exceeds emissions.
AI & DePIN at a Glance
Wait — AI/DePIN is supposed to be the future, right?
The thesis is compelling: decentralize AI compute, storage, and connectivity. Let anyone contribute GPUs, hard drives, or wireless coverage and earn tokens. The problem is the same one that plagues L1 blockchains: almost every project pays more in token emissions than it earns in revenue. GPU providers earn tokens, not customer payments. Storage providers get block rewards, not rent. The tokens they earn get dumped to cover real-world costs (electricity, hardware), creating constant sell pressure. Out of 9 major AI/DePIN projects analyzed, only 2 have revenue that exceeds emissions.
The Scoreboard: Who Actually Makes Money?
The Cost to Earn $1
How much each project spends in inflation to generate $1 in fees
Sector Breakdown: Where's the Real Revenue?
The Hidden Sell Pressure
Some tokens have billions in locked supply that's slowly being released to early investors — who often sell
The Bittensor Paradox
Bittensor is the largest AI crypto project by mindshare. It has Bitcoin-like tokenomics (21M supply, halvings), 128 specialized subnets, and institutional backing from Polychain Capital ($200M+). The Dec 2025 halving cut emissions by 50%.
But here's the paradox: the network generates zero protocol revenue. Miners receive TAO emissions for running AI models, not payments from customers. Every TAO earned is a TAO that will be sold. The halving doesn't create revenue — it just reduces the speed of dilution. This is a venture bet on future monetization, not an investment in a productive asset.