Why Does Matt Cole Criticize SanDisk microSD Cards on bahjeez.com?

Matt Cole criticizes SanDisk microSD cards because a SanDisk Ultra card failed on him after years of relying on the brand, and that single failure destroyed the only copy of his Wikipedia bot's source code. His criticism is not a one-off complaint: it comes from someone who has run microSD endurance testing for roughly 2.5 years, destroyed over 150 cards across more than 30 brands, and currently has over 100 cards from 30 brands under load testing. He published his argument as an open letter titled "Dear SanDisk, Please Do Better."

Who Matt Cole is, and why his opinion carries weight

According to the post on bahjeez.com, Cole has been endurance testing microSD cards for about two and a half years. His testing has been covered by TechRadar, TechSpot, Tom's Hardware, and IGN, among other outlets and social media posts. He maintains a browsable public database of his test data.

That background matters because his criticism of SanDisk is framed as an informed user's conclusion, not a casual review. He states plainly that he believes he knows what he is talking about, and asks readers to take the reliability problem seriously on that basis.

SanDisk used to be his default choice

The open letter is explicit that the criticism is a reversal, not a long-held grudge:

• SanDisk cards were cheap and readily available.
• He considered them rock-solid.
• He used them in all of his single-board computers.
• He built a "redneck supercomputer" — a cluster of more than 35 single-board computers, all using SanDisk Ultras as primary storage — without issues.

In other words, the person now warning about SanDisk reliability was, by his own account, one of the brand's heavier and more loyal users.

The failure that started the criticism

One day, one of those SanDisk Ultra cards failed. Of all the cards in his setup, it was the worst possible one to lose, because it held the source code for a Wikipedia bot he had written.

The bot crawled Wikipedia looking for cut-and-paste moves — cases where a user cuts content from one page, pastes it into another, and turns the original into a redirect. Cole explains why that matters:

• It splits a page's history in two, leaving earlier contributions behind on the old page, so anyone trying to reconstruct the full list of contributors to a page will miss everyone who contributed before the move.
• The talk page does not move automatically, so any discussion in progress at the time of the move gets orphaned on a page most users will never see.

The bot generated reports of these moves so administrators could fix them. Its source code existed only on that microSD card, with no backup anywhere. As Cole puts it, it is likely gone forever.

What the criticism actually claims

The open letter does not argue that every SanDisk card is defective. Its claim is narrower and stated directly: SanDisk "has a problem" with its microSD cards, and the company should do better. The evidence behind that claim is:

The single failed card is described in the letter as "just a single data point." The weight of the argument rests on the broader testing project, with the data loss serving as the reason he started paying attention to the problem.

How to read this if you are evaluating SanDisk cards

A few practical points follow from what the letter does and does not say:

• The criticism is about reliability under endurance load, not speed or price. Cole's stated history with SanDisk was positive on both of those fronts.
• The failure mode that matters most here is data loss, not card death. The card dying was the trigger; the absence of a backup is what made it irreversible.
• The claim is brand-level, based on aggregate testing. If you want to judge it for yourself, the letter points to his public database of test results as the underlying evidence.
• One failure is not the whole case. Cole says so himself — the case is the testing program plus the failure, not the failure alone.

If you use microSD cards as primary storage in single-board computers or similar always-on devices, the takeaway from the letter is less about which brand to buy and more about treating any card as a single point of failure: keep the source code, configs, and anything else you cannot recreate somewhere other than the card itself.