Retailer Shares Failure Rates for GPUs, Motherboards, SSDs, More

GeForce RTX 4090 Gaming OC 24G
GeForce RTX 4090 Gaming OC 24G (Image credit: Gigabyte)

Hardware Sugar, a retailer in the Philippines, has shared the failure rates for the most popular hardware brands over its four-year operation. Although it's a small sample size, it's interesting to see which brands are returned more frequently. Aside from the relatively small sales area, also be aware that other factors, like user error, can impact this type of data. As such, treat it as interesting information and not an absolute. 

Graphics cards had the highest failure rate out of all the hardware components, specifically those from Gigabyte. The retailer's RMA data showed a 5% failure rate for Gigabyte-branded graphics cards. Graphics cards from MSI were seemingly more reliable, and only 1.5% of the sold units failed. However, Hardware Sugar sold 9.3% more Gigabyte graphics cards than MSI graphics cards, so that's important to consider. Sadly, the retailer didn't specifically reveal which models were more susceptible to failure. It would have also been appealing to determine whether Nvidia or AMD makes the best graphics cards.

Swipe to scroll horizontally
Header Cell - Column 0 SoldFailedFailure Rate
Graphics CardsRow 0 - Cell 1 Row 0 - Cell 2 Row 0 - Cell 3
MSI12921.5%
Gigabyte14175%
MotherboardsRow 3 - Cell 1 Row 3 - Cell 2 Row 3 - Cell 3
MSI470102.4%
Gigabyte38871.8%
Power SuppliesRow 6 - Cell 1 Row 6 - Cell 2 Row 6 - Cell 3
Corsair45110.22%
DeepCool4600%
Cooler Master15732%
Seasonic644121.8
CoolingRow 11 - Cell 1 Row 11 - Cell 2 Row 11 - Cell 3
NZXT15464%
Corsair4200%
DeepCool14900%
RAMRow 15 - Cell 1 Row 15 - Cell 2 Row 15 - Cell 3
G.Skill60140.66%
TeamGroup17921%
NVMe SSDsRow 18 - Cell 1 Row 18 - Cell 2 Row 18 - Cell 3
Samsung25700%
TeamGroup24331.2%

When it comes to motherboards, it was the other way around. MSI motherboards had a higher failure rate (2.4%) than Gigabyte motherboards (1.8%). But, again, we don't get specific information on whether Intel or AMD motherboards required the highest rate of RMAs.

The information also reveals that Seasonic power supplies are very popular in the Philippines, beating big names like Corsair and Cooler Master. For reference, Hardware Sugar sold 42.8% more Seasonic units than Corsair. However, Corsair was the more reliable brand overall, with less than a 1% failure rate. Other vendors, such as Seasonic and Cooler Master, were at 1.8% and 2%, respectively.

On the other hand, cooling solutions from Corsair and DeepCool didn't show any failure at the time of the retailer's video. However, NZXT cooling products, which sold the most, had a 4% failure rate. We don't know if that data is only for air cooling, liquid cooling, or both. We suspect the latter since CPU air coolers aren't very prone to failures. A fan may prematurely die from time to time, but it's not common to see a heatsink go bad.

Regarding memory, G.Skill was the preferred brand for DIY users and consumers in the Philippines. Hardware Sugar sold a whopping 235.8% more G.Skill memory kits than TeamGroup memory kits. Despite the enormous margin, G.Skill's failure rate was just 0.66%.

Samsung and TeamGroup were the two more prevalent brands for high-speed NVMe storage. The sale numbers were close, but Samsung was the dependable manufacturer with zero RMAs. In contrast, TeamGroup units reflected a 1.2% failure rate.

Zhiye Liu
RAM Reviewer and News Editor

Zhiye Liu is a Freelance News Writer at Tom’s Hardware US. Although he loves everything that’s hardware, he has a soft spot for CPUs, GPUs, and RAM.

  • geofelt
    I like the report.
    Of course it is a small sample.
    It shows, in general, that one should not expect perfection in buying new parts.

    If a part fails during the build process, that is not so bad.
    Failure after the part is put into use is another thing.
    Puget systems keeps track of such and had a report for 2021:
    https://www.pugetsystems.com/labs/articles/Most-Reliable-PC-Hardware-of-2021-2279/
    Reply
  • ManDaddio
    I always thought 5% failure rate was the norm for Hardware.
    If all the numbers sold were equal then that would be a better assessment. But 5% is high still.
    Reply
  • Awesomeblackdude
    First let's get this out of the way why is it so difficult for users who use mobile to reply in this forum? Now onto Corsair I purchased a new PSU just before days of the covid-19 crisis and the country had shut down.

    About 4 - 6 months ago I was able to have a successful Corsair PSU installment. I power down my computer and finally added my GPU and the power supply died I contact Corsair and requested a RMA I sent proof of product but my order history on Newegg is not nowhere to be found.
    Reply
  • Elusive Ruse
    Been using Corsair PSUs and memory forever never had trouble with them; not surprised by the numbers here.
    Reply
  • samopa
    Can vouch for Samsung SSD, Gigabyte Motherboard, and Corsair PSU.
    My TeamGroup TForce memory are also good, my only complain that the timing is too rigid (CL40), can not alter at all, only speed that can be altered (succesfully hit 7000) with minor voltage increase.
    Reply
  • drivinfast247
    Awesomeblackdude said:
    First let's get this out of the way why is it so difficult for users who use mobile to reply in this forum? Now onto Corsair I purchased a new PSU just before days of the covid-19 crisis and the country had shut down.

    About 4 - 6 months ago I was able to have a successful Corsair PSU installment. I power down my computer and finally added my GPU and the power supply died I contact Corsair and requested a RMA I sent proof of product but my order history on Newegg is not nowhere to be found.
    You didn't get an email with the order details?
    Reply
  • King_V
    Awesomeblackdude said:
    First let's get this out of the way why is it so difficult for users who use mobile to reply in this forum? Now onto Corsair I purchased a new PSU just before days of the covid-19 crisis and the country had shut down.

    About 4 - 6 months ago I was able to have a successful Corsair PSU installment. I power down my computer and finally added my GPU and the power supply died I contact Corsair and requested a RMA I sent proof of product but my order history on Newegg is not nowhere to be found.
    That's weird... Newegg shows history searchable by year for about 10 years.

    You can even find stuff past 10 years, but have to do a keyword search for anything that old to come up.
    Reply
  • RichardtST
    The sample size and vendor coverage here is so small as to be completely useless. Why did you even bother publishing this? All the results here are easily within the margin of error, which is massive because the sample sizes are so small. I expect better quality data from you all.
    Reply
  • bit_user
    We don't know if that data is only for air cooling, liquid cooling, or both. We suspect the latter since CPU air coolers aren't very prone to failures. A fan may prematurely die from time to time, but it's not common to see a heatsink go bad.
    I wonder if any of the returns be due to customers not installing the CPU coolers properly. Maybe too much or too little thermal compound?

    I'm surprised to see so many Seasonic PSU failures. I have to wonder if they're lower-end models, due to Seasonic's reputation for quality.

    geofelt said:
    Puget systems keeps track of such and had a report for 2021:
    https://www.pugetsystems.com/labs/articles/Most-Reliable-PC-Hardware-of-2021-2279/
    I think data from integrators like Puget Systems is a little more reliable, since they're dealing in complete systems and won't log something as failed unless they're confident it actually failed.
    Reply
  • cryoburner
    RichardtST said:
    The sample size and vendor coverage here is so small as to be completely useless. Why did you even bother publishing this? All the results here are easily within the margin of error, which is massive because the sample sizes are so small. I expect better quality data from you all.
    This was exactly my thought as well. The sample size is far too low to be meaningful. Single-digit counts of hardware failures could easily be affected by whether or not the seller encountered a bad batch of hardware, or if they sold to a particular customer who misused multiple pieces of hardware in some way. And we don't even know what models these cover. One brand might be mostly represented by a handful of low-end models, while another may consist of high-end ones. Or one problematic model might be throwing off the numbers for a particular brand.

    It's also worth noting that this is in the Philippines, a tropical region that experiences very high temperatures and humidity, along with an unreliable power grid, meaning the conditions much this hardware is subjected to might not be representative of typical operating conditions.
    Reply