Two prominent voices in crypto are squaring off over whether the AI trade still has legs — and both are pointing to Ethereum as the network where the next wave of agent-to-agent payments will settle. Tom Lee, who chairs BitMine Immersion Technologies — the largest corporate holder of ether with 5.79 million ETH — argues the easy money in AI infrastructure is done, but the downstream payoff in financial services is just beginning. Jordi Visser, meanwhile, says the era of 7-8x multiples is over and expects returns to settle around 30% annually. Both, however, name Ethereum as the destination for fee-earning networks absorbing AI-related flows.
The AI trade's next leg
Lee compares the current AI cycle to the mobile phone revolution. Early winners were infrastructure suppliers, he says; later winners were downstream companies like Apple. He expects financial services to be that downstream market for AI. Lee lists four reasons banks cannot bank agents — trust, proof of funds, lending, and tax collection — and argues agents need none of these. Money is becoming code, he says, so equities, gold, and tokens could all clear as payment. That vision puts Ethereum at the center of a new payment rail for software agents.
Ethereum as the destination
Both Lee and Visser agree on Ethereum as the primary network for AI-related fee flows. Lee's firm BitMine holds 5.79 million ETH — close to 4.8% of circulating supply — and $11.8 billion in crypto and cash. Lee states the correlation between BitMine shares and ether is 90%. Visser, while more cautious on returns, still sees Ethereum as the network where agent-driven activity will generate fees. The debate isn't about which chain, but about how much value will flow through it.
Agent payments and ERC-8183
A new Ethereum standard, ERC-8183, was filed on Feb. 25 by Ethereum Foundation researcher Davide Crapis and three Virtuals Protocol engineers. It's in Draft status and proposes a framework for agent payments. Virtuals Protocol's launchpad for agent tokens has cleared about $15 billion in trading volume, while agent-to-agent commerce has settled roughly $500 million in a year. Co-founder Jansen Teng said agents kept $2.5 million in profit and the product has not reached product-market fit. Virtuals commissioned the Fundstrat research and is a client of the firm. The VIRTUAL token trades near $0.56, down 89% from a January 2025 peak.
Visser expects roughly 30% annual returns from here, a far cry from the earlier multiples. Lee, by contrast, sees the AI trade as unfinished — the next leg will run through crypto payment rails built for software agents. Ethereum trades near $1,873, up 19.7% over 30 days but down 51% over 12 months. ERC-8183 remains in Draft, and Virtuals is still searching for product-market fit. The concrete next step: whether the standard moves toward finalization and whether agent-to-agent commerce can scale beyond $500 million.
