Cathie Wood’s ARK sells AMD stock, buys Cerebras Systems

ARK ETFs reported a notable rebalance on May 26, 2026, buying 35,128 shares of Cerebras Systems for $9.0M and 2,451 shares of Twist Bioscience for $148.5K, while selling 5,301 shares of AMD for $2.48M and 16,998 shares of Teradyne for $6.09M. The activity signals continued portfolio rotation toward Cerebras and Twist Bioscience and reduced exposure to AMD and Teradyne. The trades are meaningful for the affected names but are likely more reflective of ARK positioning than a broad market catalyst.

Analysis

The flow pattern suggests ARK is leaning into the most capital-intensive AI infrastructure bottleneck while trimming more mature semiconductor exposure. Accumulating CBRS alongside selling AMD and TER implies a preference for “optionality on compute scarcity” over owning the broad picks-and-shovels beta that already has more consensus ownership and a clearer cyclical revenue path. That is usually a sign the fund expects the market to keep rewarding perceived displacement winners even before the economics are fully proven.

The second-order effect is that this can become self-reinforcing in small/mid-cap AI infrastructure names: persistent fund demand tightens float, amplifies moves on limited incremental volume, and can force systematic followers to chase momentum. But it also raises the failure mode: if enterprise AI spend slows or model training capex shifts to a few hyperscalers, the names with the weakest near-term revenue visibility can de-rate hardest once narrative beta fades. AMD’s relative underweighting versus the broader AI trade may reflect concern that the market is already discounting a lot of accelerator upside, leaving less room for multiple expansion absent a true share-gain surprise.

TWST is the more interesting contrarian tell. The buying suggests ARK is positioning for a platform re-rating in biotech tools, where any improvement in sequencing economics or clinical adoption could compound quickly, but the market is still treating the space as a slow monetization story. If risk appetite rolls over, TWST should outperform high-beta software less because of fundamentals and more because it has a cleaner idiosyncratic catalyst path and lower direct exposure to capex cycles than semiconductor peers.

Near term, the main reversal triggers are a pullback in AI spending guidance, a crowded-trade unwind in speculative growth, or any sign that custom silicon and packaging capacity are no longer the bottleneck. Over months, the key question is whether these names can convert flow-driven valuation support into durable earnings revisions; without that, ARK’s buying becomes a liquidity event rather than a fundamental one.

Key Decisions for Investors

• Go long CBRS versus short AMD in a 1-3 month pair trade: thesis is that narrative-driven AI scarcity premium can persist longer than broad accelerator beta, but cap the short if AMD re-accelerates guidance or announces meaningful share gains.
• Buy TWST on pullbacks over the next 2-6 weeks with a tight stop: favorable for a high-beta idiosyncratic rebound if biotech tools sentiment improves; risk/reward is attractive because positioning is still light relative to semiconductor crowding.
• Sell out-of-the-money AMD calls or run a call spread into the next catalyst window: implied upside is likely capped unless the market gets a genuine earnings revision cycle, so monetize the gap between strong AI sentiment and crowded ownership.