Cramer’s Career-Long Trading Rule Just Got Shattered by SanDisk’s Parabolic Run

On CNBC’s Squawk on the Street this morning, Jim Cramer said something he has spent a career telling viewers never to do. He said that a stock that has risen several hundred percent can still be bought, that the run does not mean the opportunity is gone, and that the memory chip cycle has changed enough to make chasing it defensible.

His words were, “I have never seen it pay to come in on top of a 500 percenter. And yet I actually think it’s going to work.” The stock he was defending is SanDisk (NASDAQ:SNDK | SNDK Price Prediction), the flash memory business spun out of Western Digital (NASDAQ:WDC) in early 2025, which is up 591.34% year to date through August 14 and rose another 8.88% on Monday to $1,786.85. Cramer’s argument is bigger than a single call. He is saying a rule he has followed his entire career, that buying after a parabolic move destroys returns, does not apply here because the industry itself has been remade. That claim deserves to be taken seriously and tested.

What Actually Changed Inside the Memory Business

The structural argument rests on contract discipline that did not exist in prior cycles. SanDisk closed fiscal 2026 with $20.25 billion in revenue, up 175.3%, $11.49 billion in free cash flow, and a GAAP gross margin of 84.6% in the fourth quarter.

More importantly, CEO David Goeckeler has signed five New Business Model agreements, multi-year customer engagements backed by firm financial commitments. Micron (NASDAQ:MU) has gone further, telling investors on its June 24 earnings call that it has signed 16 Strategic Customer Agreements covering roughly 25% of revenue today and eventually about half, with $22 billion in customer deposits and financial commitments and approximately $100 billion in remaining performance obligations across 14 of those agreements. Sanjay Mehrotra said the agreements are structured as take-or-pay with binding volume commitments and floor prices that produce gross margins well above any prior peak.

Prior memory cycles ended because customers overbought during shortages and suppliers overbuilt in response. Take-or-pay contracts with binding volume and floor pricing, designed to protect margins below current levels, break the historical cycle if the contracts hold.

Mehrotra also told analysts that DRAM and NAND demand continues to significantly exceed supply and that tight conditions will persist beyond calendar 2027, citing fab construction lead times, skilled labor shortages, and wafer pressure from HBM’s trade ratio. If those constraints are real, the ceiling on supply is being held by physics and permitting rather than discipline alone. That is the substantive case for treating this as a regime change rather than a late-cycle top. You can read Micron’s account in SanDisk’s fiscal Q4 press release filed with the SEC.

The Tepper Point, and the Rate Caveat

Cramer also pushed back against selling SanDisk after David Tepper trimmed the position, saying, “If you’re selling SanDisk because Tepper was selling SanDisk, I’m willing to go as far as to say you’re an idiot.”

On that narrower point, he is correct. A 13F reports positions as of a date already well in the past, tells you nothing about hedges, and is a poor basis for a retail selling decision. His caveat on interest rates is the one that actually matters. He said “interest rates disrupt everything”, and the 10-year Treasury yield sits at 4.63%, in the 92.7th percentile of the past year. A high-multiple structural story is exactly what a rising long end pressures most, and SanDisk trades at a P/E of 22x against consensus expectations that already price in continued outperformance.

Cramer is likely right that the contract structure represents a genuine change, and likely wrong that this makes it reasonable for most investors to chase a 591% year-to-date move. Western Digital, which he mentioned as up 195% and possibly not done, is up 195.58% year-to-date to $536.01 on a comparable structural narrative for HDDs. Riding a move like this can still pay if you plan the exit, which is the entire subject of our free bubble survivor’s handbook.

The condition that would prove the structural thesis wrong is specific and observable: a single major hyperscaler renegotiating an SCA below its floor price, or a supplier granting relief on take-or-pay terms during a demand air pocket. If either happens, the contracts are a marketing artifact rather than a business model, and the multiple compresses fast. Until one of those breaks, the case Cramer is making is defensible. Chasing it here is a different question, and on that one his career-long rule was probably right the first time.

Contact [email protected] for any questions or corrections.