Tom Lee, head of research at Fundstrat Global Advisors, stated that capital from the artificial intelligence sector is increasingly moving toward Ethereum, shifting away from semiconductor companies specializing in AI hardware. In a recent post on X, Lee described this development as a sign that digital infrastructure, particularly Ethereum, may be gaining traction with institutional investors.
AI sector capital shifts to digital infrastructure
Lee has maintained a bullish outlook on Ethereum for some time. He has previously highlighted that the network could benefit from the rising demand for decentralized settlement, tokenization, and on-chain infrastructure as AI solutions become more widespread.
Comparing Ethereum’s recent performance to that of memory chip manufactures, Lee drew attention to a significant divergence. He referred to this as the “AI downstream” trade, suggesting money might be rotating from traditional AI hardware to blockchain-based platforms like Ethereum.
The AI downstream trade continues to strengthen. Over the past month, ETH has outperformed the Roundhill Memory ETF (DRAM) by 7,200 basis points.
Over the past month, Ethereum has climbed 24%, while the Roundhill Memory ETF—focused on memory chip companies—declined by 38%. This amounts to a performance gap of 72% in favor of ETH. Lee shared a chart showing the two assets moving in opposite directions, but did not provide concrete fund flow data to confirm capital is leaving chip stocks for digital assets.
| Asset | 1-Month Performance |
|---|---|
| Ethereum (ETH) | +24% |
| Roundhill Memory ETF (DRAM) | -38% |
| Performance Gap | 72% |
Memory chip ETFs and the AI investment boom
The Roundhill Memory ETF (DRAM) was launched in April 2026 and stands as the first exchange-traded fund focused exclusively on memory chip manufacturers. These companies produce components, such as High Bandwidth Memory (HBM), DRAM, and NAND Flash chips, which are critical to supporting large-scale AI models and powering the sector’s ongoing growth.
According to IDC, global spending on AI is projected to reach $758 billion by 2029. The organization also reported that storage systems built for AI applications grew by 20.5% in the second quarter of 2025, reflecting rapid enterprise infrastructure development.
Mini dictionary: Roundhill Memory ETF (DRAM), an exchange-traded fund launched in April 2026 to give investors exposure to memory chip manufacturers supporting AI infrastructure.
Industry forecasts remain strong. TrendForce reported that contract prices for conventional DRAM are expected to rise 13–18% in the third quarter of 2026, while NAND Flash prices are projected to grow by 10–15%, primarily due to heightened demand from AI servers.
Despite positive forecasts, the recent drop in DRAM ETF prices appears to reflect profit-taking or portfolio adjustments by investors, rather than a downturn in the core memory chip industry. Lee did not clarify if funds exiting semiconductor stocks are directly moving into Ethereum, or if ETH is simply rallying while chip equities consolidate.
Implications for Ethereum and digital assets
For the cryptocurrency market, Lee’s observation raises the prospect that Ethereum may serve as a digital backbone for emerging AI technology, beyond its established role in decentralized finance.
Still, it is uncertain if these recent trends mark the beginning of a lasting institutional allocation to ETH at the expense of chip equities. Further evidence, such as sustained inflows into Ethereum investment products, would be needed to confirm a broader rotation.
Lee’s “AI downstream” narrative introduces a new angle for market participants to consider, but more data are required to determine whether this is a temporary market move or the start of a significant shift in capital between tech sectors.
