SanDisk insiders did not call the top. The narrative circulating this week — that executives quietly sold into strength before the stock rolled over — collapses the moment you open the filing
SanDisk insiders did not call the top. The narrative circulating this week — that executives quietly sold into strength before the stock rolled over — collapses the moment you open the filings. Every September sale by SanDisk's chief executive, chief financial officer and chief legal officer was executed under a Rule 10b5-1 trading plan, and the two plans with disclosed adoption dates were put in place in March and May 2026, months before the run that is now being described as the thing insiders front-ran.
The arithmetic makes the point better than the legal framework does. Chairman and chief executive David Goeckeler's mid-September dispositions priced between $1,522.59 and $1,576.82. Days later, SanDisk debuted in the S&P 100 after a three-day run to $1,791.82. Whatever those sales were, they were not informed selling into an index-inclusion print — they were sales roughly 12% to 15% below it, which is precisely the signature of a pre-scheduled plan executing on a calendar rather than a view. Having read a great many Form 4s filed during memory up-cycles, I would put it this way: the insider data here is noise, and treating it as signal is costing people the actual story, which is a mix shift in NAND demand that management expects to reprice the entire addressable market inside a single calendar year.
Three executives filed Form 4s covering September sales. The detail matters, so here it is precisely.
David Goeckeler, chairman and chief executive, disposed of common stock in open-market sales on 14 September in batches ranging from 14,364 shares down to 359 shares, at prices between $1,522.59 and $1,535.96. He filed again for 17 September, with transactions from 40 to 9,604 shares priced between $1,567.26 and $1,576.82. Both filings cite a Rule 10b5-1(c) trading plan.
Luis Visoso, chief financial officer, sold 1,000 shares on 15 September at $1,568.83 under a Rule 10b5-1 plan adopted on 22 May 2026.
Bernard Shek, chief legal officer and secretary, sold 600 shares on 1 September at $1,525.60 under a plan adopted on 4 March 2026, disposed of 2,271 shares to tax withholding on 3 September, and sold 2,308 shares on 8 September at $1,567.33 under the same plan.
Two things follow. First, the tax-withholding disposition in Shek's filing is not a sale at all in any economically meaningful sense — it is shares surrendered to cover taxes on vesting equity, and it appears in insider-selling screens as a disposal because the filing taxonomy does not distinguish intent. Aggregators routinely fold these into "insiders sold X shares" totals. They should not be read as sentiment.
Second, and more importantly, the mechanics of Rule 10b5-1 are the entire point of the rule. An executive adopts a written plan while not in possession of material non-public information, specifying amounts, prices or dates in advance, and then has no discretion over execution. The March and May adoption dates for Shek and Visoso predate the September window by four and six months respectively. Under the amended rule, officers are also subject to a cooling-off period of at least 90 days between adoption and first trade — meaning the trades executing in September were locked in before the information environment that supposedly motivated them existed.
A genuine red flag would look different: a plan adopted and then terminated early, a cluster of discretionary open-market sales outside any plan, or a plan adopted immediately before a run with the minimum cooling-off period observed. None of that is present here.
While the insider filings were being misread, SanDisk and its peers were saying something genuinely significant about the shape of NAND demand.
In its fiscal fourth quarter, reported on 5 August, SanDisk posted revenue of $8.97 billion, up 51% sequentially, with GAAP net income of $6.90 billion, or $43.97 per diluted share. Datacenter revenue rose 103% sequentially to $2.977 billion. Management stated that it expects the datacenter share of the total NAND addressable market to expand from 30% in calendar 2025 to approximately 50% in calendar 2026.
Chief executive David Goeckeler framed the year as a structural repositioning rather than a cyclical bounce: "We closed fiscal 2026 with a leading technology portfolio, established datacenter as a key growth pillar, and deepened our customer partnerships... Our technology and products are well positioned to create value for our customers and generate growing and durable free cash flow."
The word doing the work in that sentence is "durable." NAND has historically been the least durable business in semiconductors — a commodity with brutal capacity cycles, where every up-cycle is followed by oversupply as everyone adds fabs into the same demand signal. Goeckeler is arguing that a demand base weighted to datacenter rather than consumer handsets and PCs behaves differently, because hyperscale buyers contract capacity ahead and do not vanish when consumer sentiment turns.
The rest of the industry is positioning on the same thesis. Samsung intends to more than double HBM4 output in 2027, which we covered in Samsung Plans to More Than Double HBM4 Output in 2027, and Micron reports on 30 September against a $50 billion guide with an extra week in the period — the subject of Micron Earnings Date Sept 30: The Extra Week in the $50bn Guide. Three large suppliers are simultaneously telling the market that AI infrastructure has changed the demand mix. That is either the most important structural shift in memory since the smartphone, or it is the setup for the most synchronised oversupply the sector has ever engineered.
Here is the data synthesis that the headline numbers conceal. SanDisk disclosed that its 51% sequential revenue growth came roughly one-third from higher volumes and two-thirds from higher pricing.
That ratio is the single most important number in the release, and it is the one least discussed. Volume growth is demand. Price growth is scarcity. Scarcity is a function of supply discipline, and supply discipline in NAND has a documented half-life measured in quarters, not years. Two-thirds of the sequential improvement rests on a pricing environment that every supplier in the industry is currently investing to relieve.
Run it forward against guidance. Fiscal Q1 2027 revenue is guided to $10.30–10.80 billion against $8.97 billion delivered — another 15% to 20% sequential step. If the same two-thirds pricing mix holds, roughly 10 to 13 percentage points of that guided growth is price. If pricing merely flattens while volumes continue, the guide requires volume growth of a magnitude that has not yet been demonstrated.
Then there is the index effect, which is the thing investors should actually be uneasy about — not the insiders. SanDisk joined the S&P 100 following a three-day run to $1,791.82, as we reported in SanDisk Stock Joins the S&P 100. Index inclusion produces mandatory, price-insensitive buying from passive funds, concentrated into a short window. It is demand that carries no information about the business whatsoever, and it mechanically inflates the print that everyone then anchors to. The mirror image of this ran in the same rebalance, when Nike dropped out of the S&P 100 and index funds had to sell.
What looks like a warning but is not What should actually concern you Executives sold shares in September Two-thirds of sequential growth came from pricing, not volume CEO sold across two separate days Samsung, Micron and SanDisk are all adding capacity into the same signal Multiple officers sold in the same month The $1,791.82 print included price-insensitive index buying Sale prices near all-time highs Sales occurred 12–15% below the index print, consistent with scheduled plans Disposals appear in insider-selling screens Some of those disposals are tax withholding on vesting, not salesThe contrarian conclusion is straightforward. The insider-selling story is the market looking for a reason to be nervous and reaching for the wrong evidence. There is a perfectly good reason to be nervous sitting in the same earnings release, disclosed voluntarily by management, in the sentence about pricing mix.
It is worth being precise about why 10b5-1 plans carry the evidentiary weight they do, because the rule was materially tightened and much of the commentary still reflects the old regime.
The Securities and Exchange Commission amended Rule 10b5-1 in December 2022 after sustained criticism that the original safe harbour was being gamed — plans adopted with material non-public information, immediate first trades, multiple overlapping plans, and quiet cancellations when the view changed. The amendments imposed cooling-off periods of 90 to 120 days for directors and officers, restricted overlapping plans, required director and officer certifications of good faith, and mandated disclosure of adoption and termination in periodic reports. The practical consequence is that a post-2023 10b5-1 plan is a far stronger exculpatory fact than a pre-2023 one. When SanDisk's chief legal officer adopts a plan on 4 March 2026 and trades execute in September, the cooling-off requirement and the certification regime stand behind that sequence.
The residual tension — and it is a fair one — is that executives still choose when to adopt a plan, and adoption itself can be informed. An executive who understands the shape of a demand cycle better than the market can schedule an exit twelve months out and never touch the plan again. Nothing in the rule prevents that, and nothing in the filings would reveal it. That is a legitimate criticism of the safe harbour as a construct. It is not evidence of anything specific about SanDisk.
For compliance officers at brokers and platforms fielding client questions on insider data, the workable standard is narrow: treat plan adoption dates as the material field, not transaction dates; strip tax-withholding dispositions before aggregating; and flag early terminations, which remain the most informative disclosure in the regime and the one aggregators most often ignore.
Prediction one: the pricing mix, not the insider filings, drives the next repricing. SanDisk has guided to $10.30–10.80 billion for fiscal Q1 2027. If the accompanying commentary shows pricing contributing materially less than two-thirds of sequential growth while the revenue number still lands, the durability thesis strengthens sharply and the multiple can expand. If pricing remains the dominant contributor, the stock is a leveraged bet on NAND spot prices wearing a datacenter narrative, and it will trade like one.
Prediction two: Micron's 30 September print is the read-across that matters for SNDK this month. Micron reports first, against a $50 billion guide containing an extra week. Its datacenter mix commentary will either corroborate SanDisk's 30%-to-50% TAM shift or undercut it. A corroborating print lifts the whole complex; a print that shows datacenter strength offset by weak consumer NAND tells you the mix shift is real but the aggregate is not growing as fast as the mix suggests.
Prediction three: capacity announcements in the next two quarters set the 2027 cycle, and they will be aggressive. With Samsung more than doubling HBM4 output and every supplier posting record datacenter growth, the industry is behaving exactly as it has in every prior up-cycle. The causal chain is well documented: record margins fund capacity, capacity arrives eighteen to twenty-four months later, and pricing breaks before the last fab is qualified. The structural argument that datacenter demand is different is plausible. It is also the argument made at the peak of every memory cycle, and it has not been correct before.
For a fuller view of where the valuation bands sit, our SanDisk price prediction with a $2,520 bull and $1,240 bear case maps both paths. The insider filings belong in neither.
Not in any way that implies foreknowledge. Every disclosed September sale by SanDisk's CEO, CFO and chief legal officer was executed under a Rule 10b5-1 trading plan, with the two disclosed adoption dates falling in March and May 2026. The sales priced between $1,522.59 and $1,576.82, roughly 12–15% below the $1,791.82 the stock printed into its S&P 100 debut days later.
It is a written trading plan adopted while an executive is not in possession of material non-public information, specifying share amounts, prices or dates in advance so the executive has no discretion over execution. Following the SEC's December 2022 amendments, directors and officers face cooling-off periods of 90 to 120 days between adoption and first trade, restrictions on overlapping plans, and good-faith certification requirements.
Revenue was $8.97 billion, up 51% sequentially, with GAAP net income of $6.90 billion, or $43.97 per diluted share. Datacenter revenue rose 103% sequentially to $2.977 billion. Management disclosed that roughly one-third of the sequential revenue growth came from higher volumes and two-thirds from higher pricing.
Revenue of $10.30 billion to $10.80 billion, with non-GAAP diluted earnings per share of $44.00 to $46.00 — implying a further 15% to 20% sequential revenue step from the $8.97 billion delivered in the prior quarter.
Management expects datacenter to expand from 30% of the total NAND addressable market in calendar 2025 to approximately 50% in calendar 2026. Datacenter buyers contract capacity ahead and are less volatile than consumer handset and PC demand, which is the basis for the argument that cash flows are more durable than in prior NAND cycles.
Pricing, not insiders. Two-thirds of last quarter's sequential growth came from price rather than volume, and SanDisk, Micron and Samsung are all adding capacity into the same demand signal. A separate technical risk is that the recent price includes mandatory, price-insensitive buying from passive funds following the company's S&P 100 inclusion.
