The ledger keeps score. And in the AI data center narrative, the scoreboard shows a 56% drawdown for CoreWeave between June and July. Yet Jim Cramer calls it a “rocket ship.”
Let me cut through the noise. On Wednesday, Cramer paraded six stocks—Nvidia, Intel, Supermicro, Lumentum, CoreWeave, Nebius—as the “AI data center trade” reborn. Nebius jumped 34%, CoreWeave 19%, and Supermicro 19%. The bulls cheered. But I’ve been auditing the code behind these narratives since 2017, and this smells like a short squeeze painted as a revival.
Context: The Hype Cycle
Cramer’s thesis is simple: AI infrastructure spending is unstoppable, and these six companies form the supply chain. He’s not wrong about the demand—but he’s selectively ignoring the cracks. The same stocks had been bleeding for weeks, underperforming financials and healthcare. Wednesday’s CPI data gave them a lifeline, but that’s not a fundamental shift. It’s a rate-expectation trade.
What’s buried in the fine print? Supermicro missed revenue expectations. Intel jacked up its stock offering from $15B to $20B—dilution disguised as demand. And CoreWeave’s “old GPU profitability” is being spun as a sign of strength, but it’s actually a warning: the market is still desperate for compute, but the depreciation clock is ticking.
Core: Systematic Teardown
Let me break this down by the numbers—cold, empirical, no fiction.
First, the volatility. CoreWeave dropped 56%, Supermicro 53%, Nebius 48%, Lumentum 43%, Intel 42% in just two months. The Nasdaq 100 fell only 11%. These are not stable assets; they’re high-beta lottery tickets. The 19% bounce on Wednesday is a textbook short squeeze, not a trend reversal. Code is truth. Intent is fiction. The price action tells me the market is still terrified of another rate hike.
Second, the business basics. Cramer called Supermicro’s earnings “better than expected.” The truth? Revenue missed estimates. Earnings per share may have beat, but that’s a shell game. Revenue is the real signal. Supermicro is a server integrator dependent on Nvidia’s GPU allocation. If Nvidia directs more supply to hyperscalers, Supermicro gets squeezed. Minted nothing, promised everything.
Third, Intel. The 200% YTD gain looks impressive, but it’s entirely about the Chips Act and foundry hype, not AI chip market share. Intel’s AI accelerators (Gaudi) are a rounding error compared to Nvidia. The $20B offering will dilute existing shareholders by ~5%. Cramer didn’t mention that. He just said “investors are hungry.” That’s a lie by omission.
Fourth, the GPU cloud plays: CoreWeave and Nebius. Their business models are simple: rent Nvidia GPUs, charge a markup. But they compete with AWS, Azure, and GCP—companies with infinite balance sheets. CoreWeave’s “old GPU retention” is a nuanced signal. Based on my experience analyzing GPU depreciation during the 2021 mining boom, I can tell you that when old hardware retains value, it means demand is shifting from training to inference. That’s bullish for the whole sector, but it also means the next-gen Blackwell GPUs will accelerate the write-down of current assets. The clock is ticking.
Fifth, Lumentum. The 13% pop on Wednesday is market pricing in more interconnects for larger clusters. That’s a real signal—AI data centers need more optical bandwidth. But Lumentum competes with Coherent and others. No moat, just a pass-through play.
Contrarian: What the Bulls Got Right
I’ll give credit where it’s due. The AI infrastructure buildout is real. The capital expenditure from hyperscalers is not a fiction. CoreWeave’s ability to keep old GPUs profitable suggests that inference demand is absorbing legacy capacity. That’s a positive structural shift. And Cramer’s framing of the six stocks as a basket does capture the diversity of the supply chain—from silicon to servers to networking to cloud services.
But the bulls are ignoring the timing. Gas fees don’t lie. People do. The market is pricing in three years of growth into six months of price action. If interest rates stay higher for longer, or if hyperscalers trim capex, these stocks will crater again. The 56% drop in CoreWeave wasn’t an anomaly—it was a preview.
Takeaway: Accountability Call
Cramer’s job is to sell excitement. My job is to read the blockchain of market data. The ledger shows that these six stocks are not a trade—they’re a bet on continued low rates and unlimited capital spending. When the music stops, the ones who bought the “rocket ship” narrative will be left holding the bag.
Ask yourself: if Supermicro missed revenue, why should I trust the narrative? If Intel is diluting shareholders, why celebrate? The market is a machine that rewards the prepared. The code is visible to anyone who looks past the TV screen.