🚀 She Sold AMD. For This.

92% of VCs use AI to find companies. 58% of buyers start there, too. If you're not showing up in AI answers, you're invisible before the conversation even starts. Join HubSpot for Startups, Anthropic, and Marketing Against the Grain on July 30 (11 am ET) for a live AEO teardown. Real startup. Real recs. Register and unlock the free Startup Visibility Bundle.

Register Now

Cathie Wood dropped $56.9M on SpaceX in a single week.

That was ARK's biggest buy of July 13-17.

Then she went shopping for nuclear reactors and stablecoins. 👇

After SpaceX $SPCX ( ▼ 3.34% ), the money went to X-Energy $XE ( ▲ 11.7% ) at $21.6M and Circle Internet Group $CRCLON ( ▲ 12.86% ) at $13.5M.

Space, nuclear, and crypto rails. Basically a bingo card of things that either 10x or go to zero.

Smaller adds rounded it out:

The selling was bigger than most of the buying.

AMD $AMD ( ▲ 1.58% ) led the chopping block at $39.2M, followed by Deere $DE ( ▼ 1.88% ) at $20.6M, 10x Genomics at $18.5M, Twist Bioscience $TWST ( ▼ 6.82% ) at $16.4M, and Robinhood $HOOD ( ▼ 0.68% ) at $14.8M.

More trimming across Roku $ROKU ( ▼ 0.21% ) ($9.4M), Iridium $IRDM ( ▼ 0.58% ) ($6.7M), Baidu $BIDU ( ▲ 2.4% ) and Cloudflare $NET ( ▼ 1.89% ).

Genomics took a beating too: Personalis ($12.6M), Atai ($8.5M), Illumina and Natera ($5.3M each). Caterpillar got clipped alongside Deere.

Read the flows and the trade is obvious. Money moving out of the stuff that already worked, into the stuff that hasn't happened yet.

AMD and Robinhood have had a run. SpaceX and small modular reactors are still promises with a pitch deck.

The genomics exit is the one to watch. That was the ARK thesis for years, and it's now funding the space and nuclear bets.

Does your State Treasury Department have money owed to you?

Search your name and state to check if you may appear in unclaimed property records, including funds from accounts, refunds, or deposits.

Search for Money

Bitcoin is down 26.34% this year. Gold is down 7.2%.

That makes them the two worst-performing major assets of 2026.

Both are sold as protection against financial chaos. Right now the fire extinguishers are the thing on fire.

Creative Planning strategist Charlie Bilello points out that Bitcoin and gold have never both finished a calendar year as the two worst performers.

Since 2011, at least one of them has always posted a positive year. Usually both.

The year isn't over. But investors who bought both essentially purchased an umbrella and sunscreen, then got sunburned in the rain.

Here's how fast the story flipped:

Anyone holding both in 2024 felt like they'd cracked investing. Then gold kept partying while Bitcoin checked whether its invite got cancelled.

Now they're sharing a corner at the loser's table.

The maths on Bitcoin is uglier than it looks. A 26.34% drop needs a 36% gain just to get back to flat. Gold's 7.2% dip is a scratch by comparison, though it's still strange after a 64% year.

Bitcoin and gold get shoved into the same "alternative asset" bucket, but they normally move for completely different reasons.

Gold responds to interest rates, currency moves and demand for safety. Bitcoin behaves like a highly caffeinated risk asset, chasing liquidity, leverage and appetite.

So why are both bleeding at once?

Because correlations aren't loyal. Assets that hedged each other beautifully last year can absolutely fall over together this year.

The bigger takeaway: labels don't protect portfolios. Calling something a "store of value" or an "inflation hedge" is marketing, not a guarantee it behaves that way over any given six months.

There's still time on the clock. Both could recover before year-end, or they could make Bilello's observation official.

Either way, the safety trades currently aren't offering much safety.

Elon Musk just torched a $52 billion story with four words.

"This is fake news," he posted on X, responding to a report that SpaceX had ordered a mountain of Nvidia AI servers from Foxconn. The original post has since vanished.

The Economic Daily said Foxconn (formally Hon Hai Precision) had landed the contract, with SpaceX preparing to deploy over 13,000 server racks built on Nvidia's GB300 platform.

Here's the fun part. That $52B figure wasn't a contract value at all.

It came from multiplying 13,000 racks by an assumed $4 million per rack. Someone did napkin maths and the internet ran with it.

Because if true, it would have been a genuine shake-up in who builds the world's AI hardware.

Citrini analyst Jukan framed it as Foxconn breaking the Dell and Super Micro duopoly, winning its first contract to manufacture SpaceX's next-gen AI servers.

He also noted the order wasn't in Hon Hai's existing revenue guidance. AKA free upside nobody had modelled yet.

Combine that with strong results from subsidiary FII, and Jukan reckoned expectations for Hon Hai's second half had jumped significantly.

Great story. One problem: the CEO says it didn't happen.

This is a masterclass in how AI supply chain rumours move money.

A single trade publication report, an estimated price tag, and one analyst's take can reprice a $50B+ narrative before anyone confirms a thing.

Notice the chain of assumptions. Rack count was reported, price per rack was assumed, total was calculated. None of it was a stated contract value.

Worth remembering when the next eye-watering AI order number hits your timeline. The number often exists because someone estimated it, not because someone signed it.

And a denial from a CEO isn't the end of the story either. It's just the loudest voice in it so far.