AI-linked equities led in the first half of 2026
According to ChainCatcher, market demand in the first half of 2026 strongly favored companies viewed as direct beneficiaries of rising AI demand. Among the standout names, Sandisk gained more than 530% on a cumulative basis, while Micron advanced more than 230%. These moves underscored how aggressively capital was positioned around memory, semiconductor, and AI infrastructure-related themes during the period.
The scale of those gains suggested that investors were not simply buying broad technology exposure, but specifically targeting companies tied to the supply chain expected to benefit from AI-driven hardware demand. Memory and chip names were among the clearest expressions of that trade.
Memory and semiconductor ETFs then retreated from recent highs
After the strong run-up, the trade lost momentum. The report said that the Roundhill Memory ETF (DRAM) had fallen 25% from its June peak. Meanwhile, the VanEck Semiconductor ETF (SMH) declined by about 12%. The pullback indicates that the market has become less willing to extend valuations at the same pace seen earlier in the year.
For professional market participants, the retreat in these benchmark sector products is notable because it shows that weakness was not isolated to a single stock. Instead, it affected broader baskets tied to memory and semiconductor exposure, pointing to a more general cooling in the AI demand trade.
Meta-related GPU report added pressure to AI infrastructure names
The report also noted that selling accelerated in AI infrastructure stocks and computing service providers after Meta was reported to be moving forward with sales of excess GPU capacity. That headline appears to have reinforced concerns about the near-term pace of infrastructure demand and capital deployment across the AI ecosystem.
Even without additional numbers on the magnitude of the sell-off, the sequence is clear in the source report: a market previously rewarding AI-linked hardware and compute exposure turned more cautious once signals emerged that excess capacity might be entering the market.
Bitcoin rebounded as cross-asset performance diverged
In contrast to the weakness in AI infrastructure equities, Bitcoin rebounded from a two-year low. The source did not specify the size of that rebound, so no further price move should be inferred. Still, the divergence is significant: while AI-related equities and compute providers came under pressure, Bitcoin moved higher from depressed levels.
That split in performance suggests that short-term risk allocation was shifting across asset classes. At minimum, the report highlights an important market development: the unwind in parts of the AI hardware and infrastructure trade coincided with a recovery in Bitcoin, creating a clear contrast between the equity and crypto segments of the broader risk market.