As per Fundstrat’s head of analysis, Tom Lee, capital related to the bogus intelligence sector is now flowing into Ethereum as an alternative of chip makers. In a put up he made on X on July 21, Lee implied that buyers are transferring away from investing in AI {hardware} corporations towards investing in digital infrastructure, a improvement that may profit Ether from institutional funding perspective.
The latest statements come on the heels of Lee’s constant optimistic imaginative and prescient on Ethereum. Prior to now, Cryptopolitan has disclosed that Lee believes that the community stands to achieve from the rising necessities of AI options for decentralized settlement, tokenization and on-chain infrastructure.
Lee pointed to the rising efficiency hole between Ether and a fund monitoring memory-chip producers, deeming it a sign that the “AI downstream” commerce is selecting up steam.
The AI downstream commerce continues to strengthen. Over the previous month, $ETH has outperformed the Roundhill Reminiscence ETF (DRAM) by 7,200 foundation factors.
The 72-point hole Lee is pointing to
As per Lee, Ether climbed by 24% during the last month whereas the Roundhill Reminiscence ETF (ticker: DRAM) fell by 38%, making $ETH outperform the latter by 7,200 foundation factors or 72%.
Fundstrat offered a graph displaying the 2 property getting into separate instructions. It didn’t, nevertheless, provide fund-flow statistics to again up the notion that cash is leaving semiconductor shares in favor of Ethereum. A idea about “rotation” is subsequently based mostly extra on market efficiency than precise flows of capital.
This comparability is intentional. DRAM was launched in April 2026, and it’s the first ETF centered fully on reminiscence chip makers, together with people who produce Excessive Bandwidth Reminiscence (HBM) chips, DRAM chips and NAND Flash chips. The chips have assumed nice significance within the coaching and operation of enormous AI fashions, which makes the ETF an excellent marker for investments in AI infrastructure.
Why reminiscence chips are the comparability
One of many largest beneficiaries of the AI increase has been reminiscence chips. In keeping with IDC, worldwide spending on AI is predicted to succeed in $758 billion by 2029 and its enterprise infrastructure monitoring revealed that storage methods designed particularly to assist AI purposes grew by 20.5% in Q2 of 2025, as companies expanded their AI capabilities.
The prospects for the trade stay optimistic. TrendForce’s July report estimates a rise within the contract costs for standard DRAM, anticipated to rise 13 – 18% within the third quarter of 2026, whereas NAND Flash costs will develop by 10 – 15% due largely to demand from AI servers.
That makes the latest downturn within the DRAM ETF rather more pronounced. The lower appears to stem from buyers both cashing of their earnings or making changes to their semiconductor inventory holdings and never from any downturn available in the market for reminiscence chips.
Lee’s put up didn’t make clear this concern, and it remained unclear whether or not Ethereum is benefiting from the funds being withdrawn from AI {hardware} or is just profiting at a time when the chip shares are consolidating.
What it means for the crypto market
For crypto buyers, Lee’s argument affords a brand new method to take a look at Ethereum. Somewhat than viewing $ETH solely because the spine of decentralized finance, the thesis casts the community as infrastructure that might profit from the subsequent part of AI funding.
The idea nonetheless must be examined. A month’s price of outperformance isn’t sufficient data to conclude that the capital exiting reminiscence chip equities goes straight into Ether. Whether or not or not the development is indicative of a bigger institutional shift will rely on Ethereum funding merchandise’ efficiency sooner or later and steady flows into $ETH in the course of the restoration part of semiconductor shares.
In any case, the “AI downstream” thought by Lee seems to be an fascinating market occasion, though it nonetheless requires extra proof earlier than it may be known as a long-lasting shift.
