News 300TB Flash Drives Due in 2026, Supplier Claims

[Admin]

Pure Storage expects its solid-state drives to increase capacity by six times in a few years.

300TB Flash Drives Due in 2026, Supplier Claims : Read more

 

[Alvar "Miles" Udell]

Meanwhile going on three years later, we're still waiting for this

New X-NAND Tech Detailed: SLC Speed at QLC Capacity and Pricing | Tom's Hardware (tomshardware.com)

 

[domih]

300TB SSD for between $30,000 and $300,000 🤣

 

[Amdlova]

Great maybe now can storage my corn collection.

 

[bit_user]

Sounds like a way to lose data at unprecedented speeds.

Also, even using PCIe 5.0 x4, it would take at least 5.5 hours to copy a 300 TB storage drive. That's starting to get into the realm of hard drives, which is interesting to think about.

 

[bit_user]

Meanwhile going on three years later, we're still waiting for this

New X-NAND Tech Detailed: SLC Speed at QLC Capacity and Pricing | Tom's Hardware (tomshardware.com)

Maybe he was trying to charge too much to license the IP, and that put off flash memory makers. Or, maybe there were problems with actually implementing those concepts in modern flash, made on modern process nodes.

It's hard to say, because NEO Semiconductor doesn't make complete memory devices or IP. If X-NAND were announced by an actual memory maker, then you could decry it as vaporware. But, because NEO doesn't actually make or design full flash memory products, it's quite plausible they're just a form of patent troll that's trying to create an IP minefield around where they think the industry is headed.

 

[peachpuff]

Consumers will only receive $200 1tb pcie5 ssds...

 

[kerberos_20]

Maybe he was trying to charge too much to license the IP, and that put off flash memory makers. Or, maybe there were problems with actually implementing those concepts in modern flash, made on modern process nodes.

It's hard to say, because NEO Semiconductor doesn't make complete memory devices or IP. If X-NAND were announced by an actual memory maker, then you could decry it as vaporware. But, because NEO doesn't actually make or design full flash memory products, it's quite plausible they're just a form of patent troll that's trying to create an IP minefield around where they think the industry is headed.

x-nand is on gen2 now, it is available for storage companies who wants to build it, its main target is embeded devices, AI, and some high computing (cloud/nas/data centers), probably no data on whos using it now

 

[digitalgriffin]

Huge Boon for database backups administrator everywhere.

 

[jeremyj_83]

300TB SSD for between $30,000 and $300,000 🤣

If you have to ask how much it costs you cannot afford it. However, in all seriousness these drives are going to be used by companies like Google. A 2U server with 24 U.2 drive bays populated with 15.36TB drives only has 368TB of raw storage. Basically one of these drives is able to replace an entire 2U server for space. In that same 2U a server with 24x 300TB drives will have a raw storage capacity of 7.2TB. In 1 2U they can replace the storage for an entire 42U rack and that level of consolidation saves A TON of money.

 

[korekan]

meanwhile for consumer we will be stuck with 512GB and 1 TB for at least next 10 years.😆

 

[sabicao]

Sounds like a way to lose data at unprecedented speeds.

Also, even using PCIe 5.0 x4, it would take at least 5.5 hours to copy a 300 TB storage drive. That's starting to get into the realm of hard drives, which is interesting to think about.

How would you fill 300TB worth of hard drives in 5.5 hrs?

 

[jeremyj_83]

How would you fill 300TB worth of hard drives in 5.5 hrs?

PCIe 5.0 x4 is 128Gb/sec transfer rate or 16GB/sec. That means it can transfer 57.6TB/hr so 300TB would take 5.2 hours. Do note this is sustained max theoretical speed which you won't hit but is a nice numbers game to put thing in perspective. For reference at SATA 3 link speeds this would take >111 hours.

 

[bit_user]

x-nand is on gen2 now, it is available for storage companies who wants to build it, its main target is embeded devices, AI, and some high computing (cloud/nas/data centers),

Source?

probably no data on whos using it now

I find that rather suspicious. I saw nothing in the Press section of their website to indicate that anyone at all is using it. Usually, young companies are very eager to tout design wins, partnerships, or at least provide gross stats on market adoption.

 

[bit_user]

meanwhile for consumer we will be stuck with 512GB and 1 TB for at least next 10 years.😆

If you read the article, they're partially basing these projections on a 3-4x increase in 3D NAND layers. While I expect the price of NAND will not stay constant as layers are added, it should increase slower than the density. In that case, you'll definitely see better capacity/$.

 

[bit_user]

How would you fill 300TB worth of hard drives in 5.5 hrs?

That seems an obvious misreading of what I said. I simply meant that copying an entire hard drive takes in the realm of hours to days, and that has logistical implications. These SSDs would finally reach a size where you'd have to make the same sorts of considerations.

BTW, I wonder if this company is taking into account the amount of additional cooling that would be needed to sustain peak PCIe 5.0 or 6.0 transfer rates. Because some enterprise SSDs already burn quite a lot of power, and additional cooling will come at the expense of density. And if you limit transfer speeds, either directly or due to thermal throttling, then these drives would take even longer to copy or fill.

 

[kerberos_20]

Source?


I find that rather suspicious. I saw nothing in the Press section of their website to indicate that anyone at all is using it. Usually, young companies are very eager to tout design wins, partnerships, or at least provide gross stats on market adoption.

https://www.tomshardware.com/news/x-nand-gen-2-doubles-flash-write-speed

 

[bit_user]

https://www.tomshardware.com/news/x-nand-gen-2-doubles-flash-write-speed

Oops, my bad. I thought you said memory makers in those markets were using it, and I was wondering how you knew.

Okay, so I stand by my original skepticism, based on the fact that their Press page contains zero announcements of partnerships or design wins. So, I really don't know if they're anything more than a Rambus-style patent factory.

I think we can stop discussing NEO Semiconductor and giving them more free publicity. They're irrelevant to this article.