Brian Reed: Eliza, hey — I've got a thought experiment for you. You're a portfolio manager, it's the morning of August 6th, and your screen shows three pending orders: sell Palantir, buy Nvidia, buy Block. All on the same day, all out of ARKK.
Eliza Ward: Okay, I'm picturing it.
Brian Reed: Those are Cathie Wood's actual trades on August 6th, 2026. $11.4 million out of Palantir — 70,259 shares. $17.6 million into Nvidia, 80,415 shares. And roughly $21 million into Block. One day, three moves, one fund.
Eliza Ward: Wait — all three on the same day? Because the Palantir sale, the Nvidia buy — those I'd half expect after a rally. But Block in the same session is the part that—
Brian Reed: That's the part that made me stop. And honestly, that's what today is — trying to figure out if those three trades together are Wood signaling a real conviction shift about where AI infrastructure value lives, or if this is a fund that needed to harvest something that moved.
Eliza Ward: And ARKK's performance context is going to matter for that read a lot.
Brian Reed: It really is. The fund's numbers make the motive question harder to dismiss.
Brian Reed: The numbers coming out of Palantir that week were genuinely good. Like, not fine. Good. US commercial revenue up 149% year-on-year.
Eliza Ward: Which is — wait, that's the exact segment Wood's whole thesis is supposed to reward.
Brian Reed: Right. So let me try to put this plainly, because I keep turning it over. Imagine a friend owns a house. It just jumped 29% in value. And the same week, the neighborhood gets its best school rating ever. She sells anyway. You can't tell if she's being smart about the price — or losing faith in the street.
Eliza Ward: That's ARK and Palantir right now. And the 29% rally — that happened before the exit. ARK sold $6.4 million on August 5, then another $11.4 million on August 6. Both days after the pop.
Brian Reed: So she trimmed everything expensive at once — Palantir, Shopify, Roblox, same window. Is that discipline or is she clearing the deck?
Eliza Ward: The Shopify and Roblox piece is — I mean, that's what makes me nervous about calling this a Palantir-specific thesis call. If it were just Palantir, you could say valuation discipline. But trimming high-multiple software names across the board? That reads more like... a category decision. Or honestly, rebalancing pressure.
Brian Reed: And Morningstar's been calling Palantir fairly valued to sell-rated at various points — so it's not like the skepticism is coming from nowhere.
Eliza Ward: No. And — actually, there's one more bearish signal sitting in this story that nobody's really sitting with. Michael Burry has a short position in Palantir. Burry. Actively betting against it while ARK was still long.
Brian Reed: So the question isn't whether the Palantir trade was right — it's whether Wood was reading valuation or reading her own fund's performance. Those land very differently.
Eliza Ward: And that motive question is actually where the full picture gets interesting — because the cash didn't just sit there. It went somewhere very specific.
Brian Reed: Right — so where did it go? Like, map it out.
Eliza Ward: Nvidia, SpaceX, CoreWeave. That's the stack. $17.6 million into Nvidia on August 6. Then SpaceX — roughly $26 million around August 3 and another $20 million on August 6. And CoreWeave, about $15.5 million, bought after the stock dropped something like 15% in the week of August 3. She bought it on a dip.
Brian Reed: Hang on — CoreWeave after a 15% drop? That's not momentum chasing. That's... I mean, that's someone saying the price got better, not worse.
Eliza Ward: It's the detail that complicates the performance-pressure reading. You don't buy a GPU cloud name on a drawdown because your fund needs a quick win. That's a conviction move — or at least it's structured like one. Semiconductors, launch infrastructure, cloud compute. It's a deliberate layer cake.
Brian Reed: But — wait, actually, here's what I keep snagging on. The April Nvidia buy. ARK already put $11.15 million into Nvidia on April 11, 2026. So the August 6 buy isn't Wood discovering AI infrastructure. She's already there. She's adding.
Eliza Ward: Which means the August move is a continuation, not a signal. And that matters because — okay, if you're framing this as 'Wood pivots to infrastructure,' you need the August buy to be new information. It's not. SpaceX's GPU commitments get cited as a specific reason for the Nvidia optimism, but the underlying position already existed.
Brian Reed: So is she building a thesis or chasing the only thing that's working? Because ARKK is down 2.75% year-to-date while the S&P is up 12.63%. Five-year annualized, it's -9.36% against 11.76%. That's not a small gap you just shrug at.
Eliza Ward: No. And — the Block trade is the part that's going to crystallize all of this. ARK sold Block around August 1st and 2nd, then bought roughly $21 million of it back days later, same week it was buying Coinbase and Circle. That sequence doesn't fit either story cleanly, and we need to get into it.
Brian Reed: So she's buying Coinbase and Circle — the cleaner, more regulated crypto-finance names — and then she turns around and buys Block back? In the same week she sold it?
Eliza Ward: Same week. And the sources can't even settle on the Block buy — it's either $21 million on August 6 or $22.5 million on August 7. That's not a rounding error, that's the record disagreeing with itself.
Brian Reed: Hold on — that's a $1.5 million gap on a single trade and nobody's pinned it down?
Eliza Ward: Nobody's pinned it down. And that fragmentation is actually — wait, I think it matters more than it looks. Because if you can't cleanly reconstruct the trade, you definitely can't reconstruct the thesis behind it. You're reading a blurry photograph and calling it a map.
Brian Reed: The part I don't get is — Block's Bitcoin Ecosystem gross profit is down 31% year-on-year. That's the arm everyone points to as the reason to own Block. And ARK bought it back anyway.
Eliza Ward: Right, and the framing is that it's confidence in Cash App's Bitcoin strategy specifically. But — I mean, that framing and a 31% gross profit decline in that exact segment don't sit next to each other easily. One of those is doing a lot of heavy lifting.
Brian Reed: So she's buying $18.6 million of Coinbase — 122,544 shares — and $12.9 million of Circle across three days, both purer crypto-finance plays, and simultaneously rebuying the fintech with the bruised Bitcoin arm.
Eliza Ward: Which is — actually, no, that's not a coherent rotation. A rotation exits one thing and enters the cleaner version. This is exiting Block, buying the cleaner versions, and then re-entering Block. That's three moves where two would've told the story. The extra step is what breaks it.
Brian Reed: So the Block trade might be the one place where the whole 'deliberate thesis shift' reading just — doesn't hold.
Eliza Ward: One number resolves most of this. Palantir's Q3 2026 US commercial revenue growth rate. That's it. That's the test.
Brian Reed: Say it plainly — what does each scenario actually mean for Wood's August call?
Eliza Ward: If that growth rate falls below 100%, the August exit looks prescient — she read the ceiling before the market did. But if Palantir holds 80%, 90% commercial growth while Nvidia and Block flatten out? Then what she did was trim something still accelerating and chase... I mean, we won't know yet whether she chased the right thing. The coin is still in the air. All of it — the SpaceX buy, the CoreWeave dip purchase, the Block reversal, the Coinbase and Circle accumulation — Cathie Wood and ARK Invest made every one of those moves inside a single week. The verdict hasn't come back.
Brian Reed: One week of trades. One future print. Yeah — that's actually the most honest place we can leave it. We'll know soon enough.
