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

Cathie Wood’s ARK ETFs made notable shifts on Aug. 25, including buying 93,290 shares of Cerebras Systems (CBRS) for $17.3M across ARKK/ARKW. The firm also sold Twist Bioscience (TWST) 44,062 shares for $6.2M, reduced AMD by 7,600 shares for $3.5M, and sold Tempus AI (TEM) 57,819 shares for $3.8M, signaling portfolio rebalancing in tech/AI. These moves likely support modest, stock-specific positioning rather than a broad market shift.

Analysis

The flow reads more like a sentiment map than a fundamental verdict: capital is being nudged toward names with a cleaner AI monetization story and away from long-duration, research-heavy franchises that still need proof. That favors CBRS on the margin, while AMD, TEM, and TWST are the places where multiple compression can persist if investors keep preferring “clear revenue now” over “optional revenue later.” In the next few sessions, the effect is mostly technical; in the next 1-3 months, it becomes a validation trade around who can show conversion of hype into bookings, margins, or repeat demand.

Second-order, this reinforces a relative-value wedge inside innovation: AI infrastructure and compute-adjacent assets should continue to outperform application-layer AI/biotech names until the market sees hard data. The risk is that this becomes a crowded factor rotation rather than a stock-specific insight, which would make the move self-limiting once liquidity desks fade the retail/quant flow. AMD is the cleanest falsifier on the upside: if the company proves accelerator share gains or better datacenter mix, the recent sentiment drag should reverse quickly.

The contrarian read is that ARK-style selling is often mistaken for informed conviction when it is frequently just portfolio maintenance and liquidity management. That matters because the most fragile names here are the ones with thin public-market proof points; CBRS may get the benefit of the doubt near term, but if it fails to translate attention into measurable commercial traction by the next earnings cycle, the flow premium can unwind fast. For TWST/TEM, the burden is less on story and more on whether the next 1-2 quarters show enough operating leverage to justify staying in the long-duration bucket.

Key Decisions for Investors

• Long CBRS vs short TWST as a 2-6 week pair trade: express the market’s preference for AI compute over genomics optionality; size modestly because the edge is flow-driven, not yet fundamental.
• Fade rallies in TEM and TWST over the next 1-3 months if multiple expansion continues without a visible step-up in bookings or margin guidance; use a tight stop on any earnings-driven re-acceleration.
• Watch AMD into the next earnings/datacenter update rather than shorting mechanically; if AI accelerator commentary improves, the recent selling becomes a contrarian buy signal, but if share/guidance stay ambiguous the stock can lag SMH for another quarter.
• No-trade on RBLX, SLMT, and FRNM unless liquidity or corporate action changes; the current size is too small to infer durable positioning, so treat as noise unless the names gap on volume for multiple days.