SNDK ends September with short sellers continuing to scale back, cost to borrow collapsing, and a fresh analyst initiation above the current price, all ahead of a Q1 print due in less than a month.
The short covering that defined September has continued through the final week. Short interest as a percentage of the free float now stands at 3.6%, down from 3.7% noted in the previous note and well below the mid-August peak above 5.5%. In raw terms, roughly 5.31 million shares are short, compared with nearly 8.1 million at the start of September. The more striking move is in borrow cost: the cost to borrow has fallen 58% over the week to just 0.27%, giving back most of the spike that briefly took it to 0.74% on September 21. Availability tells the same story from the other direction. With availability running at nearly 4,923%, there are roughly 49 shares available to borrow for every one currently on loan. The lending market is as relaxed as it has been all year, and no squeeze mechanics are visible at current levels. The ORTEX short score has drifted lower to 31.3, down from 32.7 a week ago, consistent with a position that is gradually unwinding rather than rebuilding.
The most notable development on the Street this week was Rosenblatt initiating coverage with a Buy rating and a $2,400 price target, roughly 39% above Tuesday's close of $1,729.76. That sits at the bullish end of a range that stretches from RBC Capital's $1,600 at the cautious end through JP Morgan's $2,250 and Cantor Fitzgerald's $2,900, with the Street consensus firmly at Buy across 17 analysts and a mean target near $2,137. The bull case rests on Sandisk's New Business Model contracts, which cover roughly 65% of FY28 production with minimum contracted revenue of $93.9 billion and $16.5 billion of financial guarantees, plus a datacenter revenue line that leapt from $960 million in FY25 to $5.15 billion in FY26. Bears point to the cyclical nature of NAND pricing, the concentration risk from eSSD qualification at a handful of hyperscalers, and a valuation that may already reflect peak-cycle margins given that FY26's 84.6% non-GAAP gross margin and 79.2% operating margin were heavily pricing-driven. On EV/EBITDA the multiple has compressed slightly over the past 30 days to 5.5x, while PE on trailing figures sits near 6.9x. Factor scores reflect the split: 90-day EPS momentum ranks in the 98th percentile, analyst recommendation divergence in the 94th percentile, and days-to-cover in the 88th percentile, but 12-month forward EPS growth ranks in only the 15th percentile, underscoring the Street's caution about replicating this year's earnings quality.
The stock pulled back 8.3% over the week to $1,729.76 despite recovering nearly 1% on Tuesday, a meaningful correction but modest context given the year-to-date move of roughly 629%. Closest peers were mixed. STX slipped 0.7% on the week and WDC fell 2.4%, both far narrower moves than Sandisk's drawdown, which reflects continued high-beta behaviour in a name where the multiple expansion has been extreme by any measure.
One ownership detail worth noting is that CEO David Goeckeler filed sales totalling approximately $26.4 million on September 17, all executed under a pre-arranged 10b5-1 plan. Plan-driven sales are a scheduled mechanism rather than a discretionary signal, but the size of the transaction drew attention given his retained stake of around 383,000 shares. Separately, Vanguard's 13D/G register shows a split between The Vanguard Group, which has reduced its disclosed stake to effectively zero, and Vanguard Capital Management, which filed a fresh 13G in April showing a 7.17% position. As with all 13D/G filings, these stakes are event-driven disclosures around the 5% threshold and may not reflect current holdings.
The last two earnings prints produced a 16.9% gain on the day in August 2026 and a 22.1% gain in a separate August event, though the print on August 5 resulted in an 11.8% one-day decline. With Q1 results scheduled for October 29, the pattern points to outsized moves in both directions. The central question heading into that date is whether contracted revenue visibility can offset any softening in NAND spot pricing, and whether datacenter demand remains on the trajectory implied by the $2.98 billion 4QFY26 reading.
What to watch: the October 29 earnings release will be a direct test of whether the contract pricing floor holds in Q1 and whether the datacenter revenue run-rate established in Q4 has continued to expand.
See the live data behind this article on ORTEX.
Open SNDK on ORTEX →ORTEX Market Intelligence content is generated by AI from a snapshot of ORTEX's proprietary data. Content is informational only and does not constitute investment advice.
