The memory split, three weeks later: four scores and the tape since

We do not give buy tips, and this is not one. What we do is read the conversation around thousands of companies and score how strong and credible it is.

On August 12 we published a piece arguing that memory was not one trade. The coverage treated Micron, SanDisk, Western Digital and Seagate as a single AI-shortage story; the accounts we track had split on the names inside it, and the four signal scores in print that morning ran 86, 83, 37 and 35. If you want the setup as it looked at the time, start with the original piece. This is the follow-up; a published read only counts as a record if we come back and grade it.

What the four scores turned into

From the August 11 close, the last price in that piece, to the September 3 close:

The two names scored in the 80s gained double digits. The two names scored in the 30s went nowhere: Western Digital added less than a percent, Seagate gave a little back. The gap between the best and worst performer is 25 points of return in three weeks, inside a group the sector coverage was still describing as one trade.

The scores did not rank the winners perfectly. SanDisk, the second-highest score, beat Micron, the highest. The claim that held is the split itself: high scores on one side, low scores on the other, and the returns landed on the same sides.

The deadlock that broke

SanDisk is the interesting one, because on August 12 the raw crowd was no help at all. Across the published window it split 140 bullish against 126 bearish, a coin flip, the loudest and least agreed-upon name in the sector. A mention counter had nothing to give you.

The score read 83 anyway. It weights each call by the author's graded track record rather than counting heads, and the accounts that had been right before sat on the bullish side of that deadlock.

Then the deadlock broke their way. From August 12 through September 3 the accounts we track produced 223 SanDisk mentions running 164 bullish to 27 bearish, and the graded subset went 64 bullish to 3 bearish across 15 accounts. The stock ran from $1,271.05 to a peak of $1,786.85 on August 17, three sessions after the piece went out.

Price

$1554.99+22% over the window+22% off the low

Aug 11Sep 3

The part that has not paid

That chart also shows the honest half of the story. SanDisk gave back 17% from the August 17 peak to $1,480.77 on August 25, and at $1,554.99 it still sits 13% below that high. Micron is 5.3% off its own August 17 peak of $1,011.75. Anyone who bought either name at the August 17 top is down today, and our table does not console them.

We are not measuring from the top: the returns above are scoped from the day the read went out, which is the only day anyone could have acted on it. What we can add is what the graded room did during the fade. It did not flip. From August 18 through September 3, graded accounts made 27 bullish SanDisk calls against 2 bearish. The room that read the deadlock bullish at $1,271 kept its position through the pullback, and that hold is on the record whether the next leg rewards it or not.

Who made the calls

The graded accounts with the most SanDisk calls since August 12:

This conversation lives on the X side of our coverage: 150 of SanDisk's 223 mentions in the window came from X accounts, posting under their own handles with their records attached. Micron ran the same way, 259 mentions splitting 166 bullish to 52 bearish, with the graded subset at 50 to 7.

The quiet side of the split

Western Digital and Seagate tell the other half. On August 12 they carried the two low scores, and a week later, when Jim Cramer named all four stocks as memory picks, we ran the same split again: our trusted voices backed Micron and SanDisk, doubted Western Digital, and had no read on Seagate. Since that piece's August 20 close, Western Digital has fallen 5.9% and Seagate 6.1%, while the two backed names are roughly flat.

One scoping note. The accounts we track have largely moved on from both names: 26 Western Digital subject mentions since August 12, and 4 for Seagate. Those counts describe our coverage, and the claim they support is narrow: when the room did have a directional read on these two, it was a doubtful one, and three weeks of tape have not contradicted it.

What this is evidence of

One three-week window on four stocks proves nothing on its own, and we will keep grading windows that run against us with the same arithmetic. This one does show a thing a sector headline cannot: on the same day, from the same conversation, the credibility-weighted read put four names in the same industry on two different sides, published it, and the tape sorted them the same way.

As of this morning, Micron carries a 7-day Quantral score of 85 and SanDisk 83. Those are readings of the current conversation, not predictions, and the next check-in on them will use the same yardstick as this one. See the live reading on either name in Quantral: the score, every call behind it, and the track record of the account that made it.

---

Mention counts, sentiment splits, and credibility figures cover the accounts Quantral tracks, August 12 through September 3, 2026, bucketed by rolling 24-hour windows. Quoted August 12 scores and the 140-to-126 SanDisk split are as published in the August 12 piece. Scores are the current seven-day reading as of September 4, 2026, not historical values. Prices are closing prices via public markets through the September 3 close. Positions attributed to accounts are their stated views, not Quantral's. Quantral surfaces signals and context from public sources to support your own research. Nothing here is financial advice or a recommendation to buy or sell. Past signals are not indicative of future results.