After dominating markets through 2026, AI-linked memory and semiconductor stocks are starting to lose momentum, prompting speculation that capital could be rotating back into bitcoin.
This year’s standout performers have been chip and memory companies tied to the AI buildout, which attracted strong inflows as investors crowded into the sector and largely sidelined crypto.
That leadership may now be fading as semiconductor shares cool and bitcoin rebounds from levels near a two-year low.
ETF performance highlights the gap. The Roundhill Memory ETF (DRAM) more than doubled in the first half of the year, while the VanEck Semiconductor ETF (SMH) gained around 60%, both fueled by demand for AI infrastructure. In contrast, BlackRock’s iShares Bitcoin Trust (IBIT), the largest bitcoin ETF, has fallen roughly 30%, mirroring bitcoin’s decline.
The AI rally has been driven by companies such as Sandisk (SNDK), which has surged more than 530% on demand for NAND flash memory used in AI servers and data centers, and Micron Technology (MU), up over 230% as a major supplier of DRAM and high-bandwidth memory for AI systems.
More recently, however, signs of cooling have emerged across the trade.
The Roundhill Memory ETF has dropped around 25% from its June 22 peak, while the VanEck Semiconductor ETF has declined roughly 12%. At the same time, bitcoin briefly dipped below $58,000 on July 1 before recovering above $61,000.
The selloff in AI-related stocks intensified after reports that Meta Platforms (META) is creating a “Meta Compute” unit to monetize excess GPU capacity, raising concerns about supply-demand dynamics in AI infrastructure.
The move hit “neocloud” companies particularly hard—firms providing GPU-based computing services, including former bitcoin miners such as IREN, Cipher Digital (CIFR), and TeraWulf (WULF), all of which have fallen at least 20% from recent highs.
While still early, the combination of weakness in semiconductor leaders and bitcoin’s rebound may be an initial sign that investors are beginning to rebalance exposure between AI infrastructure and digital assets.
