Currently the largest NVMe drive is: Kioxia LC9 series drive with a massive capacity of 245.76 TB.

This enterprise-class drive is designed for data centers and AI workloads and is not available for general consumer purchase (prices unown but possibly between 5000 and 10000 USD)

so 4x of those drives = 1 Peta Byte… 8x for 1 PByte RAID10, while this is impressive, it could be possible to build even more dense storage systems from microsd cards X-D (largest current microsd card is 2TB so 512 are required at cost of 269 bucks = 137728 bucks holycow kind of costly

Q: how realistic are 1PByte of NVMe by the end of 2026? also: what would such a drive cost?

ChatGPT 5.0 mini: Short answer: Likely realistic for enterprise/datacenter NVMe (not consumer M.2) to see single‑drive capacities approaching 1 PB by end of 2026 as roadmap targets and demos already point that way, but commercial availability at scale and in multiple form factors will be limited and niche.

Why that’s plausible (brief):

  • NAND density: 2‑Tbit and larger QLC/TLC dies (BiCS8/Gen‑8 and beyond) and ~200–300+ layer 3D NAND make die capacities of 256–512 Gb and multi‑Tb die stacks feasible; vendors (SanDisk/others) have roadmap announcements toward 256–512 TB and 1 PB targets.
  • Controllers & packaging: New controllers (enterprise UltraQLC / “Stargate” style) and dense board/thermal designs can aggregate many high‑capacity dies into one EDSFF/U.2/E3/E2 style device.
    Form factors: Expect the first 1 PB devices in large enterprise EDSFF/2.5″ server form factors (high power/cooling);
  • consumer M.2 single‑card 1 PB is unlikely in 2026.
  • Traction: Press releases and demos in 2024–2025 already show 122–256 TB devices and vendor roadmaps explicitly target 512 TB → 1 PB in the near future.

What to expect on cost (ballpark):

  • Enterprise prototype/demo pricing will be extremely high initially — likely tens of thousands USD per drive for early units.
  • If/when a limited commercial product ships late 2026, expect MSRP roughly in the range $5,000–$15,000 depending on endurance class, warranty and vendor; cloud/hyperscaler negotiated prices could be lower per TB but still high.
  • By mid‑lifecycle (wider production / 2027–2028) per‑TB pricing likely falls; a rough normalized estimate: $5–$10/GB (i.e., $5,000 to $10,000 per 1 PB) early, trending down toward $1–$3/GB over a few years as volumes rise.

Practical caveats:

  • Endurance and write performance per TB for QLC‑heavy 1 PB drives will be lower than smaller TLC‑based drives;
  • suitability is primarily cold/large‑object or read‑heavy AI data‑lake workloads unless vendor techniques (direct‑write QLC, overprovisioning, host‑managed tiers) compensate.
  • Power, cooling and host platform (PCIe 5/6, EDSFF/E2 form factors) constraints matter — deploying 1 PB SSDs will require server/backplane changes.
  • Availability vs. roadmap: vendors often demo roadmap targets before broad availability; expect limited availability and high prices initially.

Q: list current vendor roadmaps and specific product announcements (SanDisk, Kioxia, Micron, Solidigm) with sources

  • SanDisk — Investor day / UltraQLC roadmap (2025–2026): previewed 128TB→256TB→512TB→1PB progression and demonstrated a 256TB NVMe SSD; roadmap mentions 1PB as a future target. Source: SanDisk investor-day coverage / press release — Blocks & Files / SanDisk FMS demo.
  • Sandisk FMS 2025 demo / PR (256TB): https://www.businesswire.com/news/home/20250805490958/en/
    • “Sandisk today demonstrated a high-capacity 256TB1 NVMe™ enterprise SSD, a breakthrough in storage capacity, performance and power efficiency, made possible by Sandisk’s new enterprise-grade UltraQLC™ platform”
  • Kioxia — roadmap & demos (2023–2025): Kioxia has published multi‑year NAND roadmaps (BiCS generations, >200‑layer stacks) and publicly discussed scaling to multi‑hundred‑TB and beyond for enterprise SSDs; demos and roadmap materials point toward 512TB+ enterprise targets. Source: Kioxia roadmap / press coverage.
    • Kioxia roadmap / tech pages (example): 2025: https://americas.kioxia.com/en-us/business/news/2025/20250220-1.html
    • “Everyone is fixated on the number of layers because it’s an easy way to define generations,” Nelson said. But lateral density is just as important because more layers add cost, he said. “For Kioxia, the number of layers is not important as it is the lateral scaling to minimize the cost.”
    • That’s where architecture comes into play, Nelson said. Kioxia’s CMOS directly Bonded to Array (CBA) architecture involves the production of a 3D NAND cell array and I/O CMOS on separate wafers, using optimal production nodes. He said this approach maximizes the bit density of the memory array and I/O performance because the CMOS circuitry is separated from the NAND array—each can be optimized on its own merit. (2024: https://www.eetimes.com/3d-nand-cant-change-the-laws-of-physics/)
  • Coverage noting capacity scaling: e.g., EE Times / storage news (example):
  • Micron — roadmap & prototypes (2024–2025): Micron’s 232‑layer announcements, high‑density QLC plans and enterprise SSD previews target multi‑hundred‑TB devices; Micron has signaled continued density increases that enable PB‑class devices in enterprise form factors. Source: Micron press & product roadmap.
  • Micron 232‑layer NAND / roadmaps: 2022: https://www.micron.com/about/blog/memory/nand/first-to-market-second-to-none-the-worlds-first-232-layer-nand
  • Micron enterprise SSD/product news (example): https://www.micron.com/resources/press‑releases
  • Solidigm — enterprise capacity announcements (2023–2025): Solidigm (previously Intel NAND group) shipped 122.88TB NVMe drives and published roadmaps toward 256TB+ enterprise devices; vendor materials and trade coverage show plans to scale to 512TB+ and arrange higher densities suitable for PB targets. Source: Solidigm product news / industry coverage.
  • https://forums.anandtech.com/threads/solidigm-d5-p5430-15-36-tb-2-5-internal-u-2-1050.2614197/

kind of obvious that those super cheap 2tb microsd cards on ebay are a scam, same with usb sticks: it is possible to program the firmware of usb and microsd storage devices so it would show a larger capacity than the device actually has…

as this vendor rating on ebay confirms: “The Samsung 2 TB SDXC cards are apparently counterfeits or defective cards. About 8 GB can be written in the delivery state (exFAT preformatted), after which there is an error message, all data is gone and the memory card no longer works. A new format with Windows or external programs is aborted with an error message.”

but at least 532 customers fell for the scam :/

ebay seems to like this kind of scam (whatever sells) because there is “report this” button on their completely catastrophic way too complicated platform

also Linus tested a “2TB” fake external drive, that turned out to be a 64GB SSD that not only has a modified firmware to show the fake 2TB but also DELETES old data in order to write new data X-D

holy heck… amazon and ebay do not care? 🙁

the scammers go as far as selling a product that rates good… then changing the product to something scam… with the advantage of leaving the good ratings in place X-D #wtf

liked this article?

  • only together we can create a truly free world
  • plz support dwaves to keep it up & running!
  • (yes the info on the internet is (mostly) free but beer is still not free (still have to work on that))
  • really really hate advertisement
  • contribute: whenever a solution was found, blog about it for others to find!
  • talk about, recommend & link to this blog and articles
  • thanks to all who contribute!
admin