Fundstrat’s head of research, Tom Lee, claims that capital from the artificial intelligence sector is now flowing into Ethereum rather than into chipmakers. In a post on X on July 21, Lee suggested investors are shifting away from AI hardware companies toward digital infrastructure, a move that could benefit Ether from an institutional investment standpoint.
The statement aligns with Lee’s consistently bullish outlook on Ethereum. He has previously argued that the network stands to gain from the growing need for decentralized settlement, tokenization, and on‑chain infrastructure driven by AI solutions. Now Lee points to a widening performance gap between Ether and a fund tracking memory‑chip manufacturers as evidence that the “AI downstream” trade is accelerating.
Over the past month, Ether climbed 24% while the Roundhill Memory ETF (ticker: DRAM) slumped 38%, resulting in a 72‑percentage‑point outperformance. Fundstrat provided a chart showing the two assets diverging, though no fund‑flow data was supplied to confirm a direct rotation. The comparison is deliberate: DRAM, launched in April 2026, is the first ETF focused solely on memory‑chip makers producing High Bandwidth Memory, DRAM, and NAND Flash chips — all critical for training and running large AI models.
The memory‑chip industry remains strong. IDC forecasts global AI spending will hit $758 billion by 2029, and enterprise storage for AI grew 20.5% in Q2 2025. TrendForce expects conventional DRAM contract prices to rise 13–18% and NAND Flash 10–15% in Q3 2026, driven by AI server demand. Thus the DRAM ETF’s decline likely reflects profit‑taking or portfolio adjustments rather than a fundamental slowdown.
Lee’s “AI downstream” thesis casts Ethereum as infrastructure primed to benefit from the next wave of AI spending. However, a single month of outperformance is not conclusive proof of a lasting institutional rotation. The trend will need to be confirmed by sustained inflows into Ethereum investment products and further divergence during chip‑stock recoveries.