UK’s trusted IT infrastructure partner since 2003
Servnet
FinanceToolsConfiguratorGet in Touch
Storage

QLC vs TLC vs MLC NAND Flash Explained

Servnet Editorial · IT infrastructure analysis9 min read

Every SSD in your estate stores data as electrical charge trapped in microscopic cells, and the single biggest lever on price, capacity and lifespan is how many bits each cell holds. This guide explains QLC vs TLC vs MLC (and SLC) in plain terms: how packing more bits per cell trades write endurance for raw density, why QLC now underpins the world's largest drives at 122TB and beyond, and when cheap QLC is exactly the right call versus when it wears out early on the wrong workload.

Bits per cell: SLC to QLC and the voltage-state explosion
5SLC · 1-bit · 2 states50k–100k P/E — cache & extreme write4MLC · 2-bit · 4 states3k–10k P/E — legacy high-endurance3TLC · 3-bit · 8 states1k–3k P/E — mainstream workhorse2QLC · 4-bit · 16 states100–1,000 P/E raw — high-capacity1PLC · 5-bit · 32 states~100 P/E — experimental / future

Bits per cell: the one idea that explains everything

A NAND flash cell is a tiny charge trap. To store data, the controller pushes a precise amount of electrical charge into the cell; to read it back, it measures the resulting voltage. The whole SLC/MLC/TLC/QLC hierarchy comes down to one question: how many distinct voltage levels are we asking a single cell to hold?

The maths is unforgiving. To store n bits per cell you need 2 to the power n distinguishable voltage levels. SLC (single-level cell) stores 1 bit using just 2 levels, charged or not. MLC (multi-level cell) stores 2 bits across 4 levels. TLC (triple-level cell) stores 3 bits across 8 levels. QLC (quad-level cell) stores 4 bits across 16 levels. The emerging PLC (penta-level cell) would push to 5 bits and a punishing 32 levels.

Every time you double the number of levels, the voltage window that separates one value from the next gets narrower. That makes the cell slower to program, because the controller has to place charge far more precisely, harder to read reliably, and far more dependent on heavy error correction. It is the root cause of every trade-off that follows: more bits per cell means more capacity and lower cost per terabyte, at the direct expense of endurance, write speed and error resilience.

The endurance penalty: why more bits means fewer writes

Flash wears out. Every program/erase (P/E) cycle degrades the oxide layer that holds charge in place, and the narrower your voltage windows, the sooner that wear starts corrupting data. This is why raw endurance falls off a cliff as bits per cell climb. Industry figures for raw cell endurance are roughly: SLC 50,000 to 100,000 P/E cycles; MLC 3,000 to 10,000; TLC 1,000 to 3,000; and QLC just 100 to 1,000 P/E cycles.

Read those QLC numbers and it is tempting to write the technology off. Do not. Raw cell endurance is not the endurance you actually get from a shipping enterprise drive. Modern 3D QLC stacks the cells vertically (Kioxia's latest LC9 uses 8th-generation BiCS flash), and pairs them with powerful LDPC error correction, generous over-provisioning and wear levelling. The result is far more usable life than the bare cell suggests. Solidigm's 122TB-class D5-P5336 QLC drive is rated at up to 134.3 petabytes written and up to 0.60 DWPD (IU-aligned 32KB random write) over a five-year warranty.

DWPD (drive-writes-per-day) is the number that matters for buyers. A 0.58 DWPD rating on a 30TB drive means you can overwrite roughly 17TB every single day for five years before the warranty endurance is exhausted. For read-dominant storage that is a colossal margin. For a write-thrashing transactional database, it may not be. The trick is matching the drive to the write intensity of the workload. If you are trying to squeeze more life out of drives already in service rather than refresh early, extending support cover through third-party maintenance is often the cheaper path than a forced upgrade.

Capacity is QLC's superpower

The upside of four bits per cell is density, and density is why QLC has quietly taken over the top of the capacity charts. Four bits per cell gives roughly 33% more raw capacity than TLC from the same silicon, and that advantage compounds with vertical 3D stacking and multi-die packaging.

The numbers are striking. Solidigm's D5-P5336 shipped in 122.88TB in the first quarter of 2025, the first single drive to break the 100TB barrier in volume. Kioxia followed with the LC9 at 245.76TB, built on a 32-die stack of 2-terabit BiCS8 QLC dies with a PCIe 5.0 interface delivering up to 12GB/s sequential reads, sampling to customers in 2025 ahead of wider 2026 availability. SK hynix, Samsung and Micron are all lining up 245TB-class drives for 2026, and Samsung has publicly committed to a 512TB PCIe Gen6 drive for 2027. None of this happens on TLC economics; it is QLC that makes petabyte-dense storage practical.

This matters most for the workloads driving today's data-centre build-out: AI training corpora, vector databases, media archives and warm data lakes that are written once and read endlessly. If you are sizing storage for those, our AI server data study puts the capacity and power figures in context.

The 2026 economics: QLC vs TLC vs HDD

Here is the reality check every UK buyer needs in 2026: high-capacity flash is not cheap, and the gap over spinning disk has blown wide open. Storage vendor VDURA reported in April 2026 that a 30TB QLC enterprise SSD cost about $15,121, or roughly $504 per terabyte, while a 30TB hard drive cost about $668, roughly $22 per terabyte. That is a 22.6x price multiple for QLC flash over HDD, up from just 4.9x a year earlier.

What changed was not the technology but the market. An AI-driven NAND shortage sent prices vertical: the same 30TB QLC drive cost only $2,450 in the second quarter of 2025 before climbing more than 470% in under a year. Spot prices for 1-terabit TLC wafers more than doubled from around $4.80 in July 2025 to $10.70 by November 2025 as fabs diverted capacity to high-margin enterprise and data-centre parts. These figures are volatile and should be treated as indicative snapshots, not fixed list prices.

Against TLC, QLC's advantage is narrower than many expect. In the same April 2026 data, a 30TB TLC enterprise SSD ran about $17,500 (roughly $583/TB) versus QLC's ~$504/TB, so QLC saved around 13% per terabyte at that tier. The bigger QLC win is fitting more capacity into fewer slots, fewer drives, less rack space and lower power. With flash this expensive, it pays to model the capital outlay properly; our IT finance calculator helps compare buy-versus-finance options, and refurbished server platforms can absorb these high-capacity drives without a full new-build premium.

Endurance vs cost per terabyte by NAND type
MLCTLCQLCBits per cell234Endurance (P/E)3k–10k1k–3k100–1,000Cost per TBindicative$583 (30TB)$504 (30TB)Best fitLegacy ent.Write-intensiveRead capacity

When cheap QLC is fine, and when it wears out early

The engineering answer is simple: match the drive's endurance to the workload's write intensity. QLC is the correct, cost-effective choice for read-dominant and write-once-read-many data. Think media and content repositories, backup and archive tiers, AI and analytics datasets, object storage, streaming, and warm data that is read far more often than it is written. On these profiles QLC delivers TLC-like read performance at lower cost per terabyte and vast capacity, and the modest DWPD rating is never the bottleneck.

QLC wears out early when you point sustained, random, small-block writes at it: heavy transactional databases, write-back caching, high-churn logging, busy virtual desktop (VDI) estates and metadata-intensive workloads. For those, choose TLC (typically 1 to 3 DWPD) or mixed-use drives, and keep QLC for the capacity tier behind them. A quick sanity check: estimate your daily write volume, then compare it to the drive's DWPD multiplied by its capacity. If a 30TB QLC drive rated at 0.58 DWPD gives you ~17TB/day of headroom and your workload writes 3TB/day, QLC is comfortable; if it writes 40TB/day, it is the wrong tool.

This is exactly the calculation that belongs in the design phase, not after deployment. When you are specifying a platform, our storage solutions guidance and server configuration pages help pair the right NAND tier to each workload rather than defaulting to a single drive type across the estate.

A UK buyer's checklist for NAND tiers

Use this checklist when you specify or refresh flash so the tier matches the workload rather than the marketing headline.

  • Read the endurance spec, not just the capacity. DWPD and TBW (terabytes written) tell you how hard the drive can be written; capacity alone tells you nothing about lifespan on a write-heavy workload.
  • Tier deliberately. Use QLC for the bulk capacity and read-heavy tiers, TLC or mixed-use for write-intensive front-ends, and reserve SLC-cache or high-endurance parts for the rare extreme-write cases. A blended estate almost always beats a single-drive-type standard.
  • Time your buying around price volatility. With NAND prices swinging sharply through 2025 and 2026, refresh timing and lead times materially change the bill; do not assume last year's per-terabyte figure still holds.
  • Do not refresh purely to escape endurance fears. Enterprise QLC endurance is far higher than the raw-cell numbers imply, and drives approaching end of warranty can often be kept safely in service. Check real wear indicators and consider extending cover; our storage end-of-life guidance explains how to judge when a drive genuinely needs replacing versus when the vendor calendar simply says so.

Sources

All capacity, endurance and pricing figures in this guide are drawn from vendor specifications and industry reporting accessed July 2026.

QLC enterprise-SSD capacity roadmap, 2018–2027
5123842561280201820212023202520262027Year (2026–27 announced roadmap)Max single-drive TBQLC max capacity (TB)
Key takeaways
  • Bits per cell is the master trade-off: SLC (1 bit) to QLC (4 bits) each double the voltage levels a cell must hold, buying capacity and lower cost per terabyte at the cost of endurance and write speed.
  • Raw QLC endurance looks scary (100 to 1,000 P/E cycles) but shipping enterprise QLC drives deliver real usable life, up to 0.60 random DWPD and up to 134.3PB written on the 122TB Solidigm D5-P5336.
  • QLC owns the capacity crown: 122TB shipping in 2025, 245TB drives arriving through 2026, and 512TB planned for 2027, densities TLC economics cannot match.
  • Flash is expensive in 2026: a 30TB QLC SSD costs about 22.6x a 30TB HDD per terabyte, and only ~13% less per terabyte than TLC, so QLC's real win is density, not raw price.
  • Match the tier to the write pattern: QLC for read-heavy and write-once-read-many data, TLC or mixed-use for transactional and high-churn workloads.
Frequently asked

FAQs — QLC vs TLC vs MLC NAND Flash Explained

Is QLC SSD reliable enough for business use?

Yes, for the right workloads. While raw QLC cell endurance is low (100 to 1,000 P/E cycles), enterprise QLC drives add strong LDPC error correction, over-provisioning and wear levelling to deliver dependable life. Solidigm's 122TB-class D5-P5336 is rated up to 0.60 random DWPD and up to 134.3 petabytes written over a five-year warranty, which is ample for read-dominant and write-once-read-many data. It is only unsuitable for sustained heavy random-write workloads, where TLC or mixed-use drives are the better fit.

How long does a QLC SSD last?

It depends on how hard you write it, not on age alone. Endurance is measured in DWPD (drive-writes-per-day) or TBW (terabytes written). A 30TB enterprise QLC drive rated at 0.58 DWPD tolerates roughly 17TB of writes every day for five years. If your actual daily writes are well below that ceiling, as they are for archives, backups and AI datasets, the drive will comfortably reach and often exceed its warranty life.

QLC vs TLC: which should I buy?

Buy QLC for capacity and read-heavy or write-once-read-many data, where it offers TLC-like read speeds, higher density and lower cost per terabyte. Buy TLC (typically 1 to 3 DWPD) for write-intensive workloads such as transactional databases, caching, logging and busy VDI. In April 2026 pricing, QLC was only about 13% cheaper per terabyte than TLC, so the real reasons to choose QLC are density and slot, space and power savings rather than headline price.

Why are QLC SSDs so expensive in 2026?

An AI-driven NAND shortage. Fabs have diverted wafer capacity to high-margin enterprise and data-centre flash, sending prices sharply higher. A 30TB QLC SSD rose from about $2,450 in Q2 2025 to roughly $15,121 by Q1 2026, and 1-terabit TLC spot prices more than doubled in under six months. As a result, 30TB QLC flash cost about 22.6x the equivalent 30TB hard drive per terabyte in early 2026. These figures are volatile snapshots, so verify current pricing before budgeting.

What is PLC NAND and is it coming?

PLC (penta-level cell) stores 5 bits per cell using 32 distinct voltage levels, pushing density even higher than QLC. In theory it lowers cost per terabyte further, but the extremely narrow voltage windows make endurance, write speed and error correction far harder to manage. PLC remains largely experimental and is not yet a mainstream enterprise product, so for planning purposes QLC is the densest tier you can realistically deploy today.

Related

Continue reading

More in Storage

Got a question this article didn't answer?

One conversation with an engineer who's done this before. No sales script.

Talk to Servnet →

Talk to a UK specialist

Get expert advice or a no-obligation quote — servers, storage, networking, maintenance, finance and cloud. We reply the same working day.

or call 0800 987 4111