Tom Lee Says AI Capital Now Favors Ethereum Over Memory Chips

Tom Lee, the head of research at Fundstrat, believes that AI-related capital is shifting towards Ethereum rather than memory chip manufacturers. Over one month, ETH gained 24% while the Roundhill Memory ETF, or DRAM, fell by 38%.

Why Tom Lee talks about a rotation of AI towards Ether

The market has already begun to treat Ether as a more cyclical asset than before. Tom Lee relies on this same strength to support his view. 

In his post shared by Cryptopolitan, he observes a gap of 7,200 basis points between ETH and the Roundhill Memory ETF over one month.

Lee’s message is simple. When ETH advances 24% and DRAM falls 38%, he sees a shift of the “AI downstream trade” towards Ethereum. In other words, the capital that benefited from the AI infrastructure boom would be looking for a relay further down the value chain.

Tom Lee wrote on X: “The AI downstream trade continues to strengthen. Over one month, ETH outperformed the Roundhill Memory ETF (DRAM) by 7,200 basis points.” 

This interpretation gives Ethereum a place in the AI narrative without reducing it to a mere speculative asset. It also has a clear limitation. Tom Lee does not show inflows towards ETH here, he only shows a performance gap between two assets operating in very different universes.

The memory chip ETF is not a perfect thermometer for AI

DRAM is not an ETF like others. The fund, launched on April 2, 2026, presents itself as the first ETF entirely dedicated to memory manufacturers, notably DRAM, NAND, and high-bandwidth memory, or HBM. It thus serves as a proxy for part of the AI infrastructure, not for the entire sector.

The problem is that memory component prices do not always tell the same story as stocks. TrendForce predicted, on July 3, an increase of 13% to 18% in contractual DRAM prices in the third quarter of 2026, as well as a 10% to 15% increase in NAND Flash prices. IDC, for its part, still anticipates $758 billion in global AI-related spending by 2029. Thus, the fund remains promising, even if the DRAM ETF fell in the observed period.

In other words, the decline in the memory fund does not prove an exodus from AI. It may also reflect profit-taking, sector reallocations, or a simple rebalancing after a too rapid phase. Here again, Tom Lee reads a capital movement. The visible data mainly shows a difference in stock market behavior.

Why Ethereum still benefits from the AI narrative

The case remains interesting for another reason. Ethereum is no longer limited to decentralized finance alone. Its settlement layer, its capacity to carry tokenization, and its role as on-chain infrastructure give it a new macro reading, especially when investors seek concrete uses beyond mining and trading.

Cryptopolitan also recalls that Lee sees Ethereum as a network likely to benefit from the growing AI needs in decentralized settlement, tokenization, and digital infrastructure. This thesis does not depend solely on the price of Ether. It also depends on the network’s ability to remain relevant when liquidity cycles change.

The market likes this type of narrative because it links two powerful themes, AI and crypto. However, one must not confuse narrative and evidence. For now, Ethereum mainly benefits from a favorable context, renewed interest in crypto ETFs, and a return of the programmable assets theme.

In short, Tom Lee’s interpretation says something useful about the market, but not yet about the flows themselves. ETH benefits from a broader narrative, memory remains under stock market pressure, and the rotation thesis towards Ethereum needs more solid data to be validated. 

At this stage, the subject is still read as a market signal, not as definitive proof. To extend the reflection, Cointribune also covered tokenization as the next major project of finance.