Cathie Wood’s ARK buys Amazon stock, sells AMD and Rocket Lab

ARK Invest disclosed large daily portfolio moves on April 24, 2026, led by a $92.5M purchase of 4.02M shares of X-Energy and a $71.5M purchase of 280,450 Amazon shares. The firm also sold $65.8M of AMD, along with reductions in Rocket Lab, Teradyne, Caterpillar, and Iridium, signaling active repositioning across tech and industrial exposures. The article is primarily a flow/positioning update rather than a fundamental catalyst, so the likely market impact is limited to the individual names involved.

Analysis

The important signal is not the headline flow itself, but the divergence between what ARK is funding and what it is funding out of. Buying a capital-intensive, policy-sensitive energy-related name while trimming hardware-adjacent semis and launch/satellite exposure suggests a late-cycle repositioning toward “optionality with catalysts” rather than pure momentum. That usually means the portfolio is becoming more selective about second-order beneficiaries of AI infrastructure, favoring names with asymmetric rerating potential over names whose upside is increasingly dependent on flawless execution.

AMD’s sale matters less as a verdict on the company and more as a tell on factor crowding: if a growth allocator is reducing a visible semi winner, it implies concern that near-term upside may already be priced relative to the next catalyst window. That can pressure the broader AI supply chain because money tends to rotate from the “picks and shovels” complex into the infrastructure layer only after compute demand visibly translates into monetizable capex. In that sense, TER and even CAT reductions read as a hedge against a slower industrial spend conversion cycle, not simply a stock-specific call.

The contrarian read is that the market may be overestimating the durability of momentum in the names being sold and underestimating the path dependence of the name being bought. If the AI buildout narrative pauses, semis and test equipment can de-rate quickly over 1-3 months because their multiple support is highly flow-dependent. Conversely, if the newly favored name is a long-duration story, the real payoff is 6-18 months out and requires patience through volatility; the near-term risk is that crowded retail enthusiasm accelerates the move before fundamentals catch up.

Key Decisions for Investors

• Reduce tactical exposure to AMD into strength over the next 1-2 weeks; use a 3-6 month horizon because the risk is multiple compression, not fundamental collapse. Prefer harvesting premium via covered calls rather than outright selling if existing long-term conviction remains.
• Pair trade: long a diversified AI infrastructure beneficiary basket vs short a semi beta basket (e.g., AMD/TER) for 1-3 months. Thesis: if capex breadth narrows, the market will reward the most direct monetizers and punish crowded second-tier hardware names; target 1.5-2.0x downside capture on the short leg.
• If holding RKLB, tighten risk and consider a call-spread overwrite. This is a sentiment-sensitive name where flow reversals can cut 15-25% in days; downside is faster than upside once the momentum tape turns.
• For AMZN, wait for either a pullback or evidence of accelerating margin leverage before adding. Best risk/reward is through staged entry over 2-4 weeks, not chase-buying a flow-driven bid; upside is steadier but less explosive than the more crowded AI names.
• Avoid chasing the newly accumulated name after a one-day institutional print; if buying, do it only on post-flow consolidation with a 6-12 month view. The key risk is that the market extrapolates one allocator’s positioning into a durable re-rating before adoption and execution milestones confirm it.