The iPhone 18 Pro Max went up for preorder September 12, 2026, and landed on shelves September 18 with a marquee spec bump: four storage tiers running from 256GB up to 2TB, an A20 Pro chip built on a 2nm process, and NVMe-based flash that Apple markets as best-in-class. Nine days after general availability, independent lab testing tells a messier story. Benchmarks published by Notebookcheck show that the 1TB configuration’s storage chip, once its cache runs dry under sustained heavy writes, can fall to throughput levels below a mid-range microSD card. For a phone that starts at $1,299 and climbs to $2,499 at the top tier, that is not a small footnote.
Two Notebookcheck editions, the German-language site and its Swedish counterpart, ran the numbers independently and reached the same conclusion within days of each other. The German headline reads, translated, “Partly slower than SD cards: iPhone 18 Pro Max 1TB in storage benchmark.” The Swedish edition put it more bluntly: “Slower than an SD card under heavy load: iPhone 18 Pro Max’s QLC storage tested.” Both point to the same underlying cause, a shift from TLC to QLC NAND flash in the higher-capacity models, and both frame the slowdown as conditional rather than constant.
Don't miss new tech stories on Google
Add Tech Insider once in the Google app and our stories appear in your news suggestions.
What Notebookcheck actually found
The core finding is narrower than the headlines suggest, and that nuance matters. Under normal, everyday conditions, the iPhone 18 Pro Max’s storage performs close to its previous-generation TLC counterpart. Notebookcheck’s testing, echoed by a separate technical breakdown from Chinese outlet Sina, describes real-world read and write speeds reaching roughly 3 GB/s while the drive’s dynamic SLC cache is active. That is the number Apple would want quoted in a press release.
The problem shows up at the edges. Once a user writes a large enough continuous block of data, big video files, an iOS backup restore, a bulk photo import, the SLC cache empties out. At that point the controller has to write directly to the QLC cells, which store four bits per cell instead of three. Four-bit cells are inherently slower to program and erase than three-bit TLC cells, because the controller has to distinguish between sixteen possible voltage states instead of eight. Sequential write throughput then drops sharply, and Notebookcheck’s testing found that in this specific worst-case scenario, sustained writes on the 1TB model fell below the sequential write speed of the microSD card used in its comparison matrix, generally in the 60-90 MB/s write range for the class of cards the outlet typically tests alongside phones.
Sina’s reporting adds a second variable that compounds the issue: available storage. As a device’s storage utilization climbs toward roughly 80% full, the pool of NAND blocks available for the dynamic SLC cache shrinks, since that pool is itself carved out of the same physical chip. A phone that is nearly full has less cache to fall back on, so it hits the QLC penalty sooner and more often. That is a familiar pattern to anyone who has owned a phone with any flash storage, near-full devices always feel more sluggish, but QLC narrows the margin considerably compared to TLC.
QLC vs. TLC: the technical distinction that matters
NAND flash chips store data in cells, and how many bits each cell holds determines both cost and speed. Single-level cell (SLC) stores one bit per cell and is the fastest and most durable, but also the most expensive per gigabyte, so it survives today mainly as a caching layer rather than primary storage. Triple-level cell (TLC) stores three bits per cell and has been the mainstream choice in flagship phones, including every iPhone Pro model before this generation. Quad-level cell (QLC) pushes to four bits per cell, which lets manufacturers fit more capacity into the same physical die area and cuts the cost per gigabyte, but at a real cost to sustained write speed and to the number of program/erase cycles the cell can survive before it degrades.
According to Sina’s analysis of Apple’s storage strategy, the company is not applying QLC uniformly across the iPhone 18 Pro lineup. The 256GB and 512GB configurations reportedly continue to use TLC NAND, the same technology as the iPhone 17 Pro Max generation. The switch to QLC applies specifically to the 1TB and 2TB tiers, the configurations aimed at users who shoot a lot of ProRes video, keep large offline media libraries, or simply want headroom. Apple is running the QLC chips in what Sina describes as a TLC-emulation mode with a dynamic SLC cache layered on top, which is precisely the arrangement that produces strong burst performance but a hard fall-off once that cache is exhausted.
That tiering decision is itself a signal. Apple’s most storage-hungry customers, the ones most likely to write large sustained files, are the ones who bought into the tier most exposed to the QLC penalty. It is also the tier commanding the highest markup: Apple charges $1,899 for the 1TB iPhone 18 Pro Max and $2,499 for 2TB, a $600 jump between the two top configurations, according to Apple’s own newsroom pricing table and confirmed independently by MacRumors’ breakdown of every storage tier’s cost. MacRumors also flags that the $2,499 price tag on the 2TB model is $500 higher than the 2TB iPhone 17 Pro Max carried at launch, meaning the generation-over-generation increase lands disproportionately on the exact tier now reported to be running QLC instead of TLC.
| iPhone 18 Pro Max tier | NAND type (per reports) | US price | Cached read/write (reported) | Sustained write after cache exhaustion |
|---|---|---|---|---|
| 256GB | TLC | $1,299 | ~3 GB/s | Not flagged as an issue |
| 512GB | TLC | $1,499 | ~3 GB/s | Not flagged as an issue |
| 1TB | QLC | $1,899 | ~3 GB/s | Falls below tested microSD card speed under heavy load |
| 2TB | QLC (reported) | $2,499 | ~3 GB/s (reported) | Same QLC cache-exhaustion pattern expected |
Notebookcheck’s own reporting frames the drop as compounding, not flat: the outlet’s smartphone news feed adds that “as storage fill-level increases, performance of the iPhone 18 Pro Max QLC storage drops even more significantly,” reinforcing that the slowdown is tied to sustained load and high utilization together, not to casual daily use like app switching, browsing or streaming.
Why this happened now: the DRAM and NAND supply context
Apple’s move to QLC in its top storage tiers does not exist in a vacuum. The memory market has been under sustained pricing pressure through 2026, with DRAM and NAND contract prices climbing as AI data center demand competes with consumer electronics for the same fabs. TechInsider has covered DRAM inventory falling below ten days of supply and handheld makers passing DRAM costs on to buyers throughout the year. Apple itself signed a multi-year NAND supply agreement to lock in capacity, a deal TechInsider detailed in its report on Apple’s NAND memory supply arrangement. QLC is measurably cheaper per gigabyte than TLC to produce, and in a memory market where every additional gigabyte of TLC costs more than it did a year ago, a cost-down move at the top-capacity tiers, where margins are already highest, is a rational business decision even if it introduces a worst-case performance regression.
That framing does not excuse the tradeoff, it explains it. Apple has not commented publicly on the QLC-versus-SD-card comparison as of this writing. The company’s iPhone 18 Pro Max newsroom announcement lists storage capacities and describes the NVMe architecture in favorable, general terms, but makes no mention of the underlying NAND cell type or of the worst-case degradation independent testers have now measured.
How this compares to what came before
The iPhone 17 Pro Max, and every Pro-tier iPhone before it, used TLC NAND at every capacity, meaning buyers of the base 256GB model and buyers of the top 2TB model got the same underlying flash technology, just more of it. That consistency is part of what made “buy more storage, get more speed headroom” a safe assumption for iPhone buyers for years. The 18 Pro Max breaks that pattern for the first time at the top tiers. It is also a reversal of the usual assumption that paying more for a flagship product buys uniformly better hardware throughout; in this case, the $600 premium for the 2TB model over the 256GB model buys more room but, under worst-case conditions, a storage subsystem that can dip below the performance of a card costing a small fraction of that upgrade fee.
It is worth noting Apple is not the first vendor to lean on QLC in a flagship device, and the industry has generally treated it as an acceptable tradeoff in laptops and external drives where large sequential writes are rarer. TechInsider’s own testing of NVMe versus SATA SSD performance and SSD versus HDD throughput shows just how wide the performance band inside “flash storage” as a category actually is, cache-backed QLC can look identical to TLC in a five-second benchmark and fall dramatically behind in a five-minute one. What is new here is seeing that gap surface inside a phone marketed on performance, at a $1,899-and-up price point, rather than in a budget laptop or an external backup drive.
Who actually hits this bottleneck
The realistic exposure here is narrower than “your new iPhone is slow.” Casual users who take photos, scroll social apps, and stream video are unlikely to ever exhaust the SLC cache; those workloads are small, bursty writes well within the cache’s normal operating range. The users who will notice are the ones the 1TB and 2TB tiers are explicitly marketed toward: videographers shooting extended 4K or 8K ProRes clips directly to the device, people restoring a full iCloud or local backup onto a new phone, users transferring large offline media libraries, and anyone running storage-heavy pro apps that write continuously for extended periods. Those are precisely the workloads that empty an SLC cache and expose the raw QLC write ceiling, and precisely the customers who paid the most to avoid exactly this kind of limitation.
There is also a compounding risk for anyone who keeps their phone nearly full, a common habit among users who buy the largest storage tier specifically so they never have to manage space. Ironically, filling a 1TB or 2TB QLC iPhone to its intended capacity is the condition most likely to shrink the SLC cache and trigger the slowdown Notebookcheck measured, which somewhat undercuts the pitch for buying maximum storage in the first place.
Market and competitive impact
For Apple, the immediate financial exposure is limited. Most iPhone 18 Pro Max buyers choose the 256GB or 512GB configurations, both of which reportedly retain TLC NAND and are not implicated in Notebookcheck’s findings. But the story lands at an awkward moment for Apple’s premium-tier pricing argument. The company has spent years positioning storage upgrades as a straightforward value trade, more money for more space and, implicitly, no worse performance. A documented, independently verified case where the most expensive configuration underperforms a cheap microSD card in a specific but realistic scenario gives critics and competitors a concrete data point rather than a hypothetical one.
Android flagship makers have faced similar scrutiny before over QLC and eMMC-class storage in budget tiers, and the discourse around “storage type” as a spec worth disclosing on the box has been building in enthusiast circles for a few years. This story could accelerate calls for phone makers, Apple included, to disclose NAND cell type the way SSD manufacturers increasingly do, rather than bundling it into a vague “storage capacity” line. It also reinforces a broader 2026 trend TechInsider has tracked: as component costs rise industry-wide, manufacturers across categories, from gaming handhelds to laptops to phones, are increasingly substituting cheaper memory technology at the high end rather than eating margin compression, often without prominent disclosure.
| Metric | iPhone 17 Pro Max (prior gen) | iPhone 18 Pro Max (current) |
|---|---|---|
| NAND type, all tiers | TLC across 256GB-2TB | TLC on 256/512GB; QLC reported on 1TB/2TB |
| Storage interface | NVMe | NVMe |
| Base price (US) | Prior-gen pricing, sub-$1,300 range | $1,299 (256GB) |
| Top-tier price (US) | Prior-gen 2TB pricing | $2,499 (2TB) |
| Chip | A19 Pro | A20 Pro (2nm) |
| Worst-case sustained write concern | Not reported | Falls below tested microSD card, per Notebookcheck |
The A20 Pro angle: fast chip, throttled pipe
There’s a particular irony in where this bottleneck lands. The iPhone 18 Pro Max ships with the A20 Pro, a 2nm chip that TechInsider covered in depth when it launched, including benchmarks showing it running a 27-billion-parameter on-device AI model at roughly twice the speed of the A19 Pro. On-device AI workloads, especially generative image and video tools, tend to read and write substantial amounts of scratch data to local storage during processing. A chip fast enough to run billion-parameter models locally is only as useful as the storage pipe feeding it, and a QLC cache-exhaustion event during a sustained AI workload, not just a video export, is a realistic scenario for exactly the power users the 1TB and 2TB tiers are pitched to. Apple built a genuinely faster processor and paired the priciest configurations with a storage subsystem that can bottleneck it under load.
What Apple could do next
Apple has several paths available, none of which it has confirmed pursuing as of September 21, 2026. A firmware-level adjustment to how aggressively the SLC cache is allocated and reclaimed could soften the worst-case scenario, at some cost to available capacity. A public specification disclosure, listing NAND type per storage tier the way it already lists RAM and chip generation, would at least let buyers make an informed choice before paying the $600 premium for 2TB. Apple could also, in a future hardware revision, reserve QLC for a lower tier and keep TLC in the 1TB and 2TB configurations specifically because those are the tiers most likely to see sustained-write workloads, essentially the inverse of its current tiering. None of these are confirmed; they are the range of responses available to a company that has stayed silent on the specific benchmark claim so far.
Predictions
- Expect Apple to stay quiet on the specific “slower than SD card” framing publicly, while likely adjusting SLC cache management in a future iOS point release, a common pattern for storage-controller tuning that doesn’t require a hardware recall.
- More independent outlets, likely including English-language reviewers and YouTube teardown channels, will replicate Notebookcheck’s testing over the next two to four weeks, either confirming or narrowing the scope of the finding.
- Expect increased scrutiny of NAND cell type as a disclosed spec across the phone industry through 2026 and into 2027, with enthusiast press pushing for “TLC vs. QLC” to become as commonly cited as RAM or chip generation.
- Android flagship makers using QLC at high storage tiers, a practice that already exists in parts of the market, will likely face comparison pieces referencing this iPhone story as a benchmark reference point.
- Given the ongoing DRAM and NAND price pressure TechInsider has tracked through 2026, expect QLC adoption to expand rather than reverse across premium devices industry-wide, making storage-type transparency a bigger consumer issue over the next several product cycles.
The bigger picture: cost pressure meets premium pricing
Step back from the benchmark numbers and this story is really about a squeeze playing out across consumer electronics in 2026. Memory costs have risen sharply this year, and manufacturers have two basic options: raise prices further, or substitute cheaper components and hold the line on price. Apple chose a hybrid, keeping headline pricing largely stable generation over generation while quietly changing what’s inside the highest tiers. That is a defensible business call in isolation. The friction comes from the gap between marketing language, NVMe storage, seamless performance, and the specific, measurable conditions under which that promise breaks down. Notebookcheck’s testing didn’t find a broken phone. It found a phone that behaves exactly as fast QLC-based NVMe storage should be expected to behave once its cache runs out, which is the part that should have been on the spec sheet from day one.
Timeline: how the story broke
Apple unveiled the iPhone 18 Pro and iPhone 18 Pro Max on September 9, 2026, with pre-orders opening September 12 across all four storage tiers, 256GB, 512GB, 1TB and 2TB, and general availability arriving September 18, according to Apple’s own newsroom announcement. Within days of the phones reaching reviewers and buyers, Notebookcheck published its storage benchmark findings on both its German and Swedish editions, dated September 21, flagging the 1TB configuration’s cache-exhaustion behavior. Chinese outlet Sina published a parallel technical analysis around the same window, tracing the QLC decision back to Apple’s broader storage-tiering strategy and confirming that the 256GB and 512GB models were not affected. MacRumors’ own September 2026 coverage of the launch lineup corroborated the four-tier pricing structure, including the $500 jump the 2TB model took over its iPhone 17 Pro Max predecessor. As of this writing, no additional English-language outlets have independently replicated the specific SD-card comparison, and Apple has issued no public statement addressing the claim.
Frequently Asked Questions
Is the iPhone 18 Pro Max actually slower than a microSD card all the time?
No. Under normal daily use, and during any burst write that fits inside the drive’s SLC cache, the phone performs at roughly the same level as prior TLC-based iPhones, around 3 GB/s according to reports. The slowdown only appears during sustained heavy writes after the cache empties, particularly when the device’s storage is already close to full.
Which iPhone 18 Pro Max storage tiers are affected?
Reports indicate the 1TB and 2TB configurations use QLC NAND flash, while the 256GB and 512GB tiers reportedly retain the same TLC NAND used in the iPhone 17 Pro Max generation.
What’s the difference between QLC and TLC NAND?
TLC (triple-level cell) stores three bits per memory cell; QLC (quad-level cell) stores four. QLC packs more capacity into the same chip area at a lower cost per gigabyte, but each cell takes longer to program and endures fewer write cycles, which is why sustained write speeds and long-term durability both trend lower than TLC.
What triggers the slowdown in practice?
Large, continuous write operations, restoring a full backup, transferring a big offline media library, or recording extended high-resolution video directly to the device, are the workloads most likely to exhaust the SLC cache and expose the underlying QLC write speed.
Has Apple responded to these benchmark reports?
Not as of this writing. Apple’s iPhone 18 Pro Max announcement describes its NVMe storage in general terms and does not address NAND cell type or the specific cache-exhaustion behavior independent testers have measured.
Should I avoid buying the 1TB or 2TB iPhone 18 Pro Max?
For most buyers, no, typical daily use is unlikely to trigger the worst-case scenario. Buyers who plan to shoot extended high-resolution video, restore large backups repeatedly, or run heavy on-device AI workloads that write large scratch files should weigh the reported behavior against their specific use case before paying the premium for the top tiers.
Does this affect the regular iPhone 18 Pro (non-Max)?
The reporting so far centers on the iPhone 18 Pro Max’s 1TB configuration specifically. MacRumors has confirmed that the 2TB tier is in fact offered on both the iPhone 18 Pro and the iPhone 18 Pro Max, not just the larger model, so given that both Pro models share the same chip and storage architecture across matching capacities, the same QLC tiering pattern would reasonably be expected to apply to the standard iPhone 18 Pro’s 1TB and 2TB options as well, though this has not been independently benchmarked in the available reporting.
How does this compare to storage issues in other device categories?
QLC-related slowdowns are a known tradeoff already documented in budget laptops and some external SSDs, where cache-backed burst speed masks a much lower sustained write ceiling. What’s notable here is seeing that same tradeoff appear inside a flagship phone at a $1,899-plus price point rather than in a budget product.
Related Coverage
- Arctis Nova Pro vs Stealth 600 vs BlackShark V3 Pro: $250 Gap [2026]
- A20 Pro Runs 27B AI Model at 2x A19 Pro Speed [2026]
- Eero Max 7 vs Orbi 970 vs Deco BE95: $600 Price Gap [2026]
- AULA F99 vs TH80 Pro vs Kisnt KN85: 2x Battery Gap [2026]
- Razer Pro Type Ultra vs WOBKEY vs 8BitDo: 15dB Gap [2026]
