The market cycle spanning 2024 to 2026 has consolidated a structural correlation between digital assets and the development of artificial intelligence. Within that framework, two analysts with institutional backgrounds —Tom Lee, co-founder of Fundstrat, and Jordi Visser, former Chief Investment Officer of Weiss Multi-Strategy Advisors— have articulated independent positions that converge on the same conclusion: Ethereum is the best-positioned infrastructure to absorb value generated by autonomous agents and automated settlement processes. Both approaches, despite departing from distinct analytical frameworks, share a quantifiable core that is now decisive for any allocation thesis in digital assets.
Tom Lee’s reading: gas demand, tokenization, and deflationary pressure
Lee’s thesis rests on observable network flow metrics and supply dynamics. His starting point is that artificial intelligence agents —systems that execute transactions, pay for services, and manage collateral autonomously— need a programmable settlement currency with no sovereign counterparty risk and continuous availability.
ETH fulfills that role because of its dual function: it is the gas token required to execute any operation on the Ethereum Virtual Machine and, simultaneously, an asset that can be locked as collateral in staking and decentralized finance protocols.
The rise in non-human activity on the main network and on data availability layers that settle on Ethereum generates an increase in base fee burning through the EIP-1559 mechanism. Lee has projected in multiple analyses that the combination of a reduced issuance rate after The Merge and sustained transactional volume driven by AI agents can turn ETH into a net deflationary asset for prolonged periods. This is not a projection based on speculative assumptions, but on the elasticity of block space demand when algorithmic actors operating 24 hours a day intervene.
The second vector Lee incorporates is the tokenization of real-world assets accelerated by artificial intelligence. Money market funds, corporate bonds, private credit rights, and property titles have begun to be represented as tokens on Ethereum and on second-layer solutions. Artificial intelligence automates the origination, fractionalization, and restructuring of those instruments, reducing operational costs and multiplying issuance and settlement events.
Every smart contract deployed, every redemption of a yield token, and every collateral reallocation requires gas. The total tokenized value locked grows, and with it the structural demand for ETH as a computational resource and as a reserve asset in the treasuries of issuing protocols.
Added to this is the expectation of institutional flows through spot Ethereum exchange-traded funds (ETFs), approved in the United States in 2024. Lee has pointed out that the AI narrative provides a valuation anchor that transcends previous speculative cycles, because it links network utility to a macro trend of business productivity.
In that scheme, demand for regulated exposure to ETH translates into buying pressure on the underlying asset, which is in turn withdrawn from circulation for staking or as collateral in decentralized derivatives markets, intensifying the effect on liquid supply.
Jordi Visser and the “AI Flippening”: settlement, identity, and the displacement of trust
Jordi Visser has formulated a thesis of crypto asset market capitalization reordering that he calls the “AI Flippening.” The central argument holds that artificial intelligence will cause a shift in the value hierarchy between Bitcoin and Ethereum because the autonomous agent economy demands a programmable settlement layer that Bitcoin cannot provide without modifying its base architecture.
Visser structures his reasoning on three pillars. The first is the settlement of payments in stablecoins. More than 60% of the supply of fiat-backed stablecoins resides on Ethereum and its rollups. AI agents will settle exchanges of digital goods, API access, parametric insurance, and subscription payments mostly in stablecoins, but every operation requires ETH as the gas unit.
The volume of automated payments expected by the end of the decade turns that technical friction into a permanent demand vector. Visser points out that this flow does not depend on market cycles, but on the expansion of computing infrastructure and multimodal language models that generate transactions on behalf of users or corporations.
The second pillar is verifiable identity and reputation for non-human agents. The operation of autonomous economic agents requires cryptographic authentication mechanisms, on-chain reputation accumulation, and permission management that do not depend on a central entity. Ethereum, through account abstraction standards (ERC-4337), the Ethereum Name Service (ENS), and verifiable credential formats, offers a decentralized identity layer that can be used by software agents without human intervention.
The third pillar is a displacement of institutional trust toward verifiable code. In an environment where artificial intelligence automates treasury functions, loan origination, and capital allocation, participants will demand transparent and real-time auditable execution rules. Smart contracts on Ethereum allow verifying compliance with conditions without relying on intermediaries that can modify policies unilaterally.Â
Visser argues that this neutrality differential will cause a gradual migration of value from instruments that depend on sovereign jurisdictions toward decentralized protocols governed by open-source code. The base asset backing that infrastructure is ETH, which functions as the reserve asset of an algorithmic settlement economy.
Convergence points and empirical evidence accumulated through 2026
Both analysts agree that AI agents will use public and permissionless networks, discarding enterprise solutions with controlled validators. Unified liquidity, token standards (ERC-20, ERC-721, ERC-4626), and lending and money market protocols already deployed with years of security history make Ethereum the only platform capable of absorbing demand without excessive fragmentation.
On-chain data collected from early 2025 to mid-2026 support several of these assumptions. The percentage of gas consumed by addresses associated with autonomous agent frameworks (such as ELIZA, GAME, ubc, and others) rose from less than 0.5% in January 2025 to exceed 6% of total gas on Ethereum mainnet and certain rollups in June 2026, according to Dune Analytics dashboards and Nansen reports.Â
This growth does not come from a single protocol, but from a fragmented ecosystem of agents executing loan settlement transactions, yield position rebalancing, prediction market arbitrage, and data verification for decentralized oracles.
ETH burning metrics correlated with peaks in agent activity show that days with higher gas consumption by these systems coincide with increases in the net negative issuance rate. The Dencun upgrade (2024) and the introduction of blobs for data availability reduced execution costs on L2s, allowing AI agents to operate with transaction costs low enough to justify automated business models.
The volume increase on execution layers translates into higher settlement frequency toward the settlement layer, where ETH is burned as part of the batch inclusion process.
In parallel, real-world asset (RWA) tokenization records monitored by rwa.xyz and The Block indicate that the total value of financial instruments represented on Ethereum and its L2 extensions exceeded 22 billion dollars in the second quarter of 2026, an increase of nearly 300% since the beginning of 2025.
The participation of asset managers such as BlackRock, Franklin Templeton, and WisdomTree in issuing tokenized funds has been publicly documented and leverages existing smart contract infrastructure, reducing settlement times and allowing automated composition with DeFi protocols.
The algorithmic management of these funds by autonomous treasury agents adds an additional layer of predictable gas demand.
The ETH/BTC correlation has shown a relevant behavioral shift. Between 2022 and 2024, movements of that pair depended almost exclusively on macro flows and risk appetite within the crypto ecosystem.
Starting in 2025, the introduction of ETH ETFs in the United States and the rise of AI and tokenization narratives have generated episodes where the pair appreciates coinciding with announcements of advances in artificial intelligence models or favorable stablecoin regulation, even in sessions where Bitcoin trades sideways.
While causality is complex to isolate, 90-day rolling correlation data between the AI equity index (BOTZ, AIQ) and the ETH/BTC pair have moved from levels close to 0.2 to surpassing 0.6 in some stretches of 2026, a value that research teams from FalconX and Kaiko have begun to monitor as part of the post-halving correlation regime.
Implications for valuation and risk allocation
The convergence of these two theses is not limited to a narrative exercise. If gas consumption by autonomous agents maintains the observed growth rate and asset tokenization reaches the volumes projected by market infrastructure analysts, network fee revenue could become a variable independent of the crypto asset price cycle.
Valuation models based on revenue multiples, such as the relationship between network value and annualized fees (NVT adjusted), allow projecting valuation ranges that differ from purely monetary approaches applied to Bitcoin.
The risks to this thesis lie in the technical execution of the scalability roadmap, competition from alternative chains with parallel execution architectures and differentiated fee models, and the possibility that agent frameworks migrate part of their settlement activity to layer-3 solutions or to execution environments that do not inherit Ethereum’s security. So far, network effects, liquidity depth, and token standardization have acted as effective entry barriers, but monitoring the market share of TVL and active developers remains necessary.
The investment case posed by Lee and Visser does not rest on vague expectations of mass adoption, but on metrics of non-human economic activity already underway. Ethereum’s capacity to function as financial infrastructure for autonomous agents introduces a demand vector that did not exist in previous cycles and that may modify the supply dynamics and institutional perception of the asset in the coming years.Â
In that context, the discussion about whether Ethereum can surpass Bitcoin in market capitalization ceases to be tribal speculation and becomes a testable hypothesis with on-chain activity data, stablecoin growth, and the expansion of volumes settled by algorithmic agents.
