Cathie Wood Has Been Selling Palantir Into Every Rally Since Its 29% Earnings Pop

Palantir shares surged 29.5% after Q2 results, with revenue up 93% YoY to $1.94B and third-quarter revenue guided to ~$2.16B alongside a full-year outlook near $8.15B (about +82% vs 2025). Despite the strong fundamentals, ARK funds sold 109,492 PLTR shares (~$17M) on Aug. 4-5 and sold another 11,525 shares (~$2M) after the follow-on jump, while buying ~$26.6M of Nvidia. The mixed flow suggests ARK is responding to valuation/position sizing rather than deteriorating business performance, but the ~150x trailing earnings multiple keeps the stock valuation-sensitive.

Analysis

This reads more like a marginal-buyer signal than a fundamental call. When a software name is marked at a triple-digit earnings multiple, the stock becomes exquisitely sensitive to portfolio rebalancing: the first sellers can be “smart money” without implying the business is broken, but they do tell you the next leg depends on fresh estimate upgrades, not just momentum. In that regime, PLTR’s risk/reward shifts from asymmetrically positive to brittle — any deceleration, even from very high growth, can trigger multiple compression faster than the earnings base can catch up.

The second-order effect is relative-value pressure across AI software. PLTR now sets the hurdle rate for adjacent high-multiple names; if it holds this valuation, other AI applications vendors will be forced to defend their own premiums with either faster growth or clearer FCF conversion. That tends to compress dispersion inside software while widening dispersion versus AI infrastructure, which is why NVDA is the cleaner “quality at a reasonable price” expression when investors rotate away from crowded software beta.

Contrarianly, the market may be underestimating how much of PLTR’s upside is already in the price, but also underestimating the power of a continued beat-and-raise cycle. Over the next 1-3 months the key catalyst is not sentiment, it’s whether bookings and commercial momentum stay hot enough to justify the multiple; over 6-18 months, the test is whether growth can stay above the threshold needed to outrun dilution of the valuation. Falsifier: another quarter of sustained >70% revenue growth with expanding margins and raised guidance would keep the bull case intact; a guide that merely meets expectations would be enough to start de-rating the stock.

Key Decisions for Investors

• Initiate a 1-3 month long NVDA / short PLTR pair trade: NVDA offers better earnings visibility and a much lower multiple; target 10-15% relative outperformance if AI capex stays firm. Stop if NVDA guidance disappoints or PLTR keeps re-rating on successive estimate raises.
• For short-only accounts, buy PLTR 1-2 month put spreads on further strength rather than chasing outright shorts. Risk/reward is better on a volatility fade because the stock can stay elevated, but the downside accelerates if the next update is merely good instead of exceptional.
• Use any PLTR-led AI software rally to rotate toward infrastructure names (NVDA first, then SMH exposure) over the next 1-3 months. This is the cleaner expression if the market starts paying up for capex beneficiaries over story stocks.
• Set an alert for the next PLTR earnings print and guidance revision. If revenue growth slips below the current elite band or margins stop expanding, the multiple can compress sharply; if the company raises again, cover any tactical short exposure immediately.