Samsung SSD Deals - 990 Pro Family Verdicts, Live

What a verdict here actually means

Every row above is compared against its own tracked price history, in its own condition, and nothing else. Not against a list price, not against what the drive cost at launch, and not against a competitor. The question the verdict answers is narrow on purpose: for this exact listing, is today's price near the bottom of its own recent range, in the middle of it, or above it.

That narrowness is the point. A discount measured against a manufacturer's suggested price tells you about the suggested price, which in this category is frequently a number nobody has charged in months. A discount measured against the drive's own recent trading range tells you whether you are catching it low, which is the only version of the question that affects what you pay.

The comparison is per exact ASIN and per condition, which matters more in this family than in most. A 990 Pro 2TB with a heatsink and a 990 Pro 2TB without one are separate listings with separate stock and separate price histories, and they routinely diverge. Judging one against the other's range would produce confident verdicts that describe the wrong product. The same applies to condition: a used listing is judged against used history, never against new.

Where a listing is too new to have a meaningful range, the verdict reads as building history rather than as a judgement. That is common in this family right now because the 9100 Pro line is recent, and we would rather show you an honest absence than compute a verdict from a handful of observations and present it with the same confidence as one built on months of data.

The market these verdicts sit in

Consumer NVMe prices rose sharply through the 2025-2026 NAND shortage. Coverage across that period repeatedly described popular consumer drives roughly doubling from their earlier lows, and this family was not exempt. Every price on this page should be read against that: the absolute numbers are high by any historical standard, and a verdict of at or near low means low within the current market rather than low against what you remember.

This is why we do not print a percentage off anything. A percentage requires a reference, and every available reference is compromised. List prices in this category were in many cases reset upward during the shortage, so a discount against list can be large while the absolute price remains poor. Launch prices belong to a different market. Historical lows from before the shortage are real but unreachable, and quoting them would imply an achievable target that does not exist.

The verdict framework sidesteps that by refusing to reference anything outside the listing's own recent behaviour. It cannot tell you whether the current market is fair — that is a judgement about the whole industry, not about one drive. It can tell you whether, within this market, you are buying at the good end or the bad end of where this listing has been trading.

One consequence worth internalising: in a rising market, at or near low verdicts become rarer and more valuable, and elevated verdicts become more common without any seller doing anything unusual. The distribution of verdicts across the table is itself a reading of the market's direction.

Why deal articles cannot do this

The standard alternative to this page is a deal article: someone notices a price, writes it up, and publishes. Those articles are useful at the moment of publication and misleading immediately afterwards, because the price they name is a snapshot and the article does not know when it stops being true. Many of them acknowledge this obliquely by linking to a price-tracking site, which is an admission that the number in the article has a shelf life measured in days.

The structural problem is that deal coverage is episodic and buying is not. You do not need to know that a drive was cheap last Tuesday; you need to know whether it is cheap now, at the moment you are ready to spend. An article can only answer the first question, and only for the drives someone happened to write about.

A continuous verdict answers the second question for every listing in the family simultaneously, including the ones nobody wrote about. That coverage difference is as important as the freshness difference. Deal articles cluster around the popular capacities and the popular models, which means the 1TB and the 8TB and the bare-versus-heatsink distinctions — exactly the places where the odd pricing lives — go uncovered.

None of which means a verdict is a recommendation. It is one input. A listing at its own recent low is still a listing from whatever seller is behind it, at whatever condition is stated, and those checks remain yours to make.

The sibling arbitrage

Because the table above carries all three models together, it exposes something a single-model deals page cannot: the moments when the sibling ordering inverts. A 990 EVO Plus sitting at its own recent low, against a 990 Pro sitting elevated against its own range, is a specific and temporary situation in which the value drive is not merely cheaper but unusually cheap while the premium drive is unusually dear. The gap at that moment is wider than the product hierarchy would ever justify.

The same logic runs across the generational line. The 9100 Pro at a low against a 990 Pro at an elevated verdict can put the newer Gen5 drive below the older Gen4 one — which is not a hypothetical in this family, because the gap between those two lines already inverts at some capacities without any verdict divergence at all.

Reading for that pattern is straightforward: scan the verdict column for a low on one sibling and an elevated on another at the same capacity, then check the actual prices rather than assuming the verdicts imply an ordering. A low verdict on an expensive drive and an elevated verdict on a cheap one can still leave the expensive drive dearer. The verdicts tell you about each listing's own range; the prices tell you what you pay.

This is also the reason the comparison pages exist alongside this one. The verdict answers whether a listing is well-priced against itself. The gap tables answer which drive is cheaper right now. Those are different questions and a purchase usually needs both.

Sales events and reinflated discounts

Large retail events produce a burst of deal coverage and a burst of pricing that looks like deals. Distinguishing the two is a methodology question rather than a matter of watching more closely, and the methodology is the same one this page runs continuously: compare the event price against the listing's own recent trading range, not against whatever reference price the event is advertising itself against.

The failure mode to watch for is a price that rises in the period before an event and then falls to a level that is presented as a discount but is at or near where the listing was trading beforehand. Against a list price that looks like a substantial reduction. Against the listing's own recent range it reads as typical, which is what our verdict will say. This is not an accusation about any particular seller; it is a description of how a reference-price discount can be technically accurate and practically empty.

The corollary is that a genuine event low will show up here as at or near low, and it will do so without any special handling, because the framework does not know or care that an event is happening. If a price is genuinely at the bottom of its own recent range, the verdict says so on an ordinary Tuesday and during a sale alike.

The practical advice is dull and effective: during an event, check the verdict rather than the banner. A drive showing elevated during a sale is a drive whose sale price is above its own recent normal, however the discount is framed.

Frequently asked questions