The reports we flagged landed after the bell. SpaceX beat by nearly a billion and AMD beat and raised, and both fell about seven percent. Only Nvidia rose.  ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌
 
THE INVEST HAVEN
August 5, 2026  •  Morning edition  •  No hype, just perspective.
SpaceX and AMD Both Beat Last Night. Both Fell Anyway.
Yesterday afternoon we flagged that the record was leaning partly on a paper gain, and that SpaceX and AMD were both due to report after the bell. They reported. SpaceX beat revenue by nearly a billion dollars, and AMD delivered a double beat with raised guidance, the kind of night that would have sent both stocks higher a year ago. Instead both fell roughly seven to nine percent within minutes. The one large AI name that rose was Nvidia, the company collecting the checks the other two are writing.
Sponsored Brownstone Research

Editor’s Note: What if you could claim a stake in a technology set to become 100X bigger than Bitcoin… starting with just $2? Click here to see the details from investing legend and Washington D.C. insider Jeff Brown — the man who picked Bitcoin, Tesla, and Nvidia before they exploded higher. Or read more below.


Dear Reader,

If you missed out on Bitcoin… this could be even bigger.

It might sound crazy…

But according to the man overseeing $13 trillion — BlackRock CEO Larry Fink…

A new technology could be 100 times bigger than Bitcoin.

(Yet he believes it’s still “like where the internet was in 1996.”)

In 2015, I pointed my readers to Bitcoin when it was trading for just $240.

Some had the chance to see gains as high as 52,000%.

Enough to turn every $100 into $52,100.

So if the world’s largest money manager is right…

And I believe he is…

This could be the greatest investing opportunity of the century.  

We have so much to look forward to,

Jeff Brown
Founder & CEO, Brownstone Research

The Scoreboard
SpaceX: revenue of $7.8 billion beat estimates by nearly a billion, up 92 percent from a year ago, yet the stock fell about seven percent after hours on capital spending of $18.37 billion for the quarter, roughly $15.8 billion of it AI.
AMD: a double beat and a raised outlook, record revenue of $11.54 billion and Q3 guidance near $13 billion, and the stock still dropped about eight to nine percent after hours on a gross-margin miss tied to its AI buildout.
Who rose: Nvidia gained about two percent after hours after SpaceX said it would build its AI systems exclusively on Nvidia’s Vera Rubin chips. Palantir had closed the day up about 29 percent on its own blowout.
The tape: before the after-hours drops, the S&P 500 closed at a record 7,737 and the Dow at a record 54,086, a broad risk-on session on easing oil and strong daytime earnings.
The pattern: SpaceX and AMD cap two weeks in which Alphabet, Meta, Microsoft and Amazon all reported with their AI spending under the same microscope.
Details
The Market Re-Sorted the AI Trade After the Bell
The bet we flagged yesterday reported after the close, and for a few minutes it looked like vindication for the bulls. SpaceX put up revenue of $7.8 billion, up ninety-two percent from a year ago and nearly a billion above what analysts expected, with a loss cut roughly in half. AMD followed with record revenue, adjusted earnings above forecasts, and third-quarter guidance well ahead of the Street. Prediction markets had given SpaceX only about a one-in-three chance of beating. It cleared that bar with room to spare. Both stocks then turned red and stayed there.
The number that undid SpaceX was its own ambition. The company spent $18.37 billion on capital projects in a single quarter, close to six times a year ago, with roughly $15.8 billion of it poured into AI compute. It earned $7.8 billion and spent more than twice that building the machinery meant to earn more later. AMD’s stumble was quieter but the same in spirit: a gross margin that came in a couple of points light, weighed down by the cost of ramping its new AI hardware. A year ago, spending this aggressively on AI was the surest path to a higher stock. Last night it was the reason two companies that beat expectations both fell.
The rule that just broke. For most of the last two years the AI build-out ran on a simple market rule: announce a bigger number and your stock went up. Bigger data centers, bigger chip orders, bigger multi-year commitments. Spending read as ambition, and ambition got rewarded. That is the rule that gave way overnight. SpaceX announced the biggest number of all and got punished for it, on the same day Palantir was bid up nearly thirty percent for selling software rather than building the plumbing underneath it.
Who the market paid instead. Look at what rose while SpaceX and AMD sank. Nvidia, which will collect a large share of the tens of billions those two are spending, gained after hours. Palantir, which sells finished AI software, closed the day near a record. The money stayed inside the AI trade and moved down the invoice, toward the names collecting the spending and away from the ones doing it. That is a rotation, not a retreat, and it is happening inside the index most readers already own.
Why this is a macro story. The S&P 500 closed at a record 7,737, and its heaviest weights are split between the companies spending fortunes on AI and the companies being paid those fortunes. For two years the index treated both groups as one trade. This week the market began charging different prices for each. If you hold a broad index fund, that re-sort happened inside your account whether you watched it or not.
The discipline underneath. None of this runs on rockets. A one-year Treasury pays about four percent right now for taking no view at all, and the ten-year sits near 4.6 percent, with the long bond close to levels last seen before the 2008 crisis. When cash is that expensive, the market stops extending spenders the benefit of the doubt and starts asking when the spending turns into cash back. SpaceX said it holds $100 billion and expects a $100 billion annual revenue run rate by December. The market chose to price the $18 billion it spent this quarter and wait on the promise about next year.
Where that leaves you. Two dates sit directly ahead. Thursday, the first major post-IPO lockup expires and early SpaceX backers can sell for the first time, so the tape gets a second read on the same story within forty-eight hours. Friday at 8:30, the July jobs report lands after June printed a startling 57,000. Treat each report as a verdict on that one company, remember that today’s record-chasing crowd was selling this same trade a week ago, and note that the risk-free four percent pays you to wait for the answer rather than guess it.
A year ago the market paid for the promise of the AI build-out. Last night it started pricing the cost of it, and it did so to two companies that beat their numbers. The boom is not over; it is being audited, one earnings report at a time, and the grade now turns on who collects the spending. Friday’s jobs report is the next number that matters. Until then, the question worth sitting with is which side of that invoice your own index fund is standing on.
Harold Winston
Thirty years advising individual investors. Now reads markets for a living.
No hype, just perspective.