The Selectivity Trade: Why SpaceX and AMD Got Punished, Palantir Got Rewarded, and Records Happened Anyway

A day that produces record closes and falling AI stocks in the same 24 hours isn't confused — it's specific. Tuesday delivered exactly that combination, and the specificity is the actual story. The S&P 500, Dow, and Nasdaq all closed at fresh records, lifted by genuine macro relief as oil tumbled on Iran deal optimism. In the same session, two companies with genuinely strong quarterly numbers — SpaceX and AMD — watched their stocks fall anyway, while a third company, Palantir, surged 29% on largely the same broad "AI trade" that supposedly connects all three. That's not a market confused about what it wants. That's a market that has become considerably more selective than the phrase "AI stocks" usually implies, and Tuesday drew the dividing line with unusual clarity.

Two Genuinely Strong Reports, Two Falling Stocks

Start with the numbers, because they were good — not marginally good, genuinely strong by almost any conventional measure. SpaceX's first earnings report as a publicly traded company showed revenue climbing 92% to $7.8 billion, with the company narrowing its net loss to $541 million and disclosing a $47.5 billion order backlog. Its AI-related revenue specifically jumped 250% year over year, and Starlink's subscriber base crossed 12 million, pushing connectivity revenue up 66%. By almost any reasonable standard, this is the kind of debut report that should have sent a newly public company's stock climbing.

AMD's report told an equally strong story from a different angle. The company posted record quarterly revenue of $11.5 billion, with adjusted earnings per share of $1.66, driven by Data Center sales that jumped 107% year over year to $6.7 billion — a segment directly tied to the AI infrastructure buildout that has dominated market attention all year. Both companies beat on the headline numbers that typically move stocks. Both stocks fell anyway.

What the Market Was Actually Punishing

The explanation in both cases wasn't the revenue — it was what sat beneath it. SpaceX's stock came under pressure specifically because the company disclosed higher-than-expected spending on its AI business, a detail that overshadowed the otherwise strong top-line growth. AMD's issue was more subtle: its forward outlook, despite record current-quarter results, simply failed to generate the kind of forward excitement that would justify continued upward momentum in the stock. In both cases, the market wasn't questioning whether these companies could grow. It was questioning whether the pace and structure of their spending, or the credibility of their forward guidance, justified paying up for that growth right now.

• SpaceX's stock fell despite a 92% revenue increase specifically because disclosed AI spending came in above what investors had modeled, echoing the exact pattern that punished Meta's strong quarter two weeks earlier
• AMD's record Data Center revenue wasn't enough to offset a forward outlook investors found underwhelming, showing markets are now weighting guidance credibility as heavily as current-quarter execution
• Both reactions confirm that strong historical numbers alone are no longer sufficient to move AI-infrastructure-adjacent stocks higher — the market now demands clarity on both spending discipline and forward trajectory

Palantir's 29% Answer to the Same Question

Set directly against those two reactions, Palantir's 29% single-session surge looks, at first glance, like it belongs to an entirely different market. It happened in the same trading session, under the same broad macro backdrop, technically within the same loosely defined "AI trade" that connects nearly every major technology stock this year. And yet investors treated it as a completely different proposition than SpaceX or AMD.

The distinction worth drawing out here is structural rather than sentimental. Palantir sits closer to the application and software layer of the AI stack — a company selling AI-powered analytics and decision-making tools directly to enterprise and government customers, with revenue that maps fairly directly to specific, deployed use cases. SpaceX and AMD, by contrast, sit closer to the infrastructure layer — building the physical and computational capacity that AI applications eventually run on, spending heavily in advance of confirmed, monetizable demand. When markets reward one layer of this stack while punishing another in the same session, it's a genuinely useful signal about where investor patience currently sits.

Why the Application Layer Is Winning the Infrastructure Layer's Trust Right Now

Infrastructure spending is, by its nature, a bet made in advance of confirmed returns — you build the data centers, the chips, and the launch capacity before you know with certainty how much revenue that capacity will eventually generate. Application-layer companies like Palantir operate on a shorter, more immediately verifiable cycle: revenue growth tends to track more directly with specific, named contracts and deployed use cases that investors can point to and confirm today, rather than infrastructure capacity investors have to trust will eventually be filled. In an environment where markets have grown newly skeptical of unconfirmed infrastructure bets, that shorter, more verifiable cycle is currently commanding a real premium.

• Palantir's revenue growth ties more directly to specific, currently deployed enterprise and government contracts, giving investors more immediately verifiable evidence than infrastructure capacity spending offers
• The application layer of the AI stack is increasingly being rewarded for near-term revenue clarity, while the infrastructure layer is being scrutinized for spending that outpaces confirmed near-term demand
• This divergence within a single session suggests investors aren't broadly bearish or bullish on "AI" as a category — they're actively discriminating between different layers of the same technology stack based on how quickly each converts spending into verifiable revenue

Records and Falling AI Futures — How Both Happened at Once

The third layer of Tuesday's contradiction sits at the index level. The S&P 500, Dow, and Nasdaq all closed at record highs, driven by a genuinely significant macro catalyst: Treasury Secretary Scott Bessent said publicly that the US and Iran were in active talks, with a chance of a deal within a day to reopen the Strait of Hormuz. Oil responded immediately and sharply, falling nearly 6% on the session, and the broader market rallied on the combination of reduced geopolitical risk and cheaper energy costs — a genuinely broad-based, macro-driven rally rather than a narrow tech move.

But by early Wednesday trading in Asia, Nasdaq 100 futures had already turned lower, specifically on the overnight SpaceX and AMD reactions. This is the part worth sitting with: a powerful, genuinely positive macro catalyst was strong enough to lift the entire market to record closes, but it wasn't strong enough to prevent AI-infrastructure-specific caution from reasserting itself within hours. That's a meaningfully different market condition than one where a single dominant narrative — whether optimism or fear — is driving every asset in the same direction. Instead, macro sentiment and AI-specific sentiment are now operating as two semi-independent forces, each capable of dominating different parts of the market simultaneously.

Why Two Separate Sentiment Tracks Matter More Than One Combined Mood

Markets are easier to read, and arguably easier to trade, when a single dominant narrative explains most price action on a given day. What Tuesday and Wednesday morning demonstrated instead is a market running on two separate tracks that don't automatically move together anymore — broad macro sentiment, currently buoyed by Iran de-escalation hopes, and AI-infrastructure-specific sentiment, which is being judged on its own separate, stricter set of criteria regardless of what's happening with oil prices or geopolitical risk. Investors and analysts trying to read overall market direction from a single catalyst, whether the Iran talks or an earnings report, are increasingly likely to miss half the actual picture.

• A genuinely significant macro catalyst — Iran de-escalation hopes and a sharp oil price drop — was strong enough to lift broad indices to record closes despite simultaneous AI-specific caution
• Nasdaq 100 futures turning lower within hours of the broader market's record close shows AI-infrastructure sentiment now operates as a largely independent force from broad macro sentiment
• This divergence suggests markets are entering a phase where reading a single dominant catalyst is no longer sufficient to explain overall direction, since macro and AI-specific sentiment can move in opposite directions simultaneously

What Tuesday Actually Taught the Market

Three companies beat expectations by conventional measures, and the market delivered three different verdicts based on a much more specific set of criteria than "AI" as a single category would suggest. SpaceX and AMD, both sitting closer to the infrastructure layer, were punished for spending patterns and guidance that didn't yet clearly justify their pace. Palantir, sitting closer to the application layer with more immediately verifiable revenue, was rewarded handsomely in the same session. And the broader market, riding a genuine and significant Iran-related macro tailwind, posted records anyway — before AI-specific caution reasserted itself within hours, on an entirely separate track. The lesson isn't that AI enthusiasm is fading. It's that the market has become considerably more precise about which part of the AI story it's actually willing to pay for, and Tuesday drew that line more clearly than perhaps any single day this earnings season.