A new rumor claims that some iPhone 18 Pro models will use slower QLC NAND storage, mimicking a similar 2024 iPhone 16 Pro report. It makes more sense now than it did then, but doesn’t matter much in practical usage.
This latest report suggests that Apple will use the faster TLC storage for the iPhones that people are most likely to buy. But those choosing the larger 1TB and 2TB capacities may be left with a slower QLC alternative from SK Hynix.
Companies like Apple continue to struggle to source the storage components required for new products. With that in mind, it may not be surprising to see Apple go this route. Sourcing 1TB and 2TB TLC components may be difficult, if not impossible.
And, certainly, it will be spendy given the current economic environment surrounding flash media.
Advertisement
However, we’ve heard this story before. And it doesn’t seem to have been accurate that time around. And as we discussed back then, it’s unclear whether the use of QLC storage would be a real issue for iPhone owners.
QLC or TLC for iPhone 18 Pro
This latest report centers around the iPhone 18 Pro and iPhone 18 Pro Max. WCCFTechshared details of a post by the leaker “Reptalica” which claims Apple will use different storage types for different models.
According to the X post, Apple will use TLC NAND provided by SK Hynix, Kioxia, and SanDisk when building 256GB and 512GB iPhone Pro/Pro Max models. The 1TB model will use a mixture of SK Hynix QLC storage and Samsung TLC chips.
It’s then argued that Apple will solely use SK Hynix’s QLC storage for the 2TB model.
Advertisement
A rumor, repeated
If this all sounds familiar, it’s because we saw very similar claims in January 2024, prior to the iPhone 16 Pro’s unveiling in September of that year. We were told then that Apple would use QLC storage for iPhones with 1TB of storage or more.
Getting concrete information on whether that actually happened isn’t easy. That being said, we’ve only seen reports of high-capacity iPhone 16 Pro models with the fast TLC storage. That doesn’t mean there aren’t some QLC NAND chips floating around.
If there are, we’ve yet to see one.
The differences between QLC and TLC
Triple-Level Cell (TLC) NAND flash and Quad-Level Cell (QLC) NAND flash are both types of storage. But they aren’t the same.
Advertisement
The iPhone 18 Pro storage may be a hot topic this hear.
One difference is the way QLC can store four bits of data per cell of memory, rather than the three of TLC. This then allows QLC NAND to store more data, which is why it’s sometimes used in larger-capacity storage. It’s also cheaper to produce.
Unfortunately, QLC is also thought to be less reliable than TLC and, importantly, it’s also slower as it is rewriting all four bits instead of the three.
How much slower in the real world, on mobile, is a matter for debate. The report notes that QLC storage is particularly slow when reading random data. But it’s unclear how that would impact the way people use iPhones.
Advertisement
Smartphone loads on flash storage are generally in bursts, instead of sustained transfers. As such, the difference in performance is likely to be imperceptible to users who don’t resort to benchmarking tools.
It’s also important to remember that this rumor did the rounds two years ago and, as far as we can see, turned out to be incorrect. Only time will tell if this latest report is more accurate.
For most Singaporeans, Raffles Medical is a familiar name. The healthcare group has built a reputation as one of Singapore’s most established private medical providers.
But behind the scenes, the company has spent the last decade chasing a much bigger ambition.
Rather than remaining a Singapore-focused healthcare operator, Raffles Medical wanted to become a regional healthcare brand—one with hospitals and clinics stretching across Asia.
Today, however, that investment still hasn’t translated into equally impressive financial returns. While its Singapore operations are well-established and consistently profitable, its sizeable investment in China continues to lag behind.
Why China looked like an obvious market
Image Credit: Raffles Medical Group
Back in the mid-2010s, expanding into China seemed like the logical move.
The country’s population was ageing, disposable incomes were rising, and healthcare reforms were gradually opening the door to private healthcare providers.
Rather than stopping at outpatient clinics, Raffles Medical doubled down on its China ambitions by investing in full-service hospitals.
The Raffles Hospital in Beijing./ Image Credit: Raffles Medical Group
Together, the projects required years of planning, construction, regulatory approvals, specialist recruitment and investment in medical equipment before they could even begin seeing patients.
Advertisement
Unlike retail stores or restaurants, hospitals can’t simply open their doors and expect customers to flood in.
Patients need to trust the brand. Doctors need to establish referral networks. Insurance partnerships have to be secured. Operating theatres, diagnostic equipment and inpatient wards all have to be utilised before a hospital starts generating meaningful profits.
In other words, healthcare is a long game.
The investment is huge, but so is the gap
The Raffles Hospital in Chongqing./ Image Credit: Raffles Medical Group
That long game is becoming increasingly visible in Raffles Medical’s financials.
Ahead of its 2026 AGM, shareholders questioned why China’s business had grown so slowly despite years of investment. Between FY2018 and FY2025, revenue from China increased by only S$25.4 million, reaching S$65.4 million.
Advertisement
The disparity becomes even more striking when compared with the group’s asset base. China accounts for around 30% of Raffles Medical’s total assets, yet contributes only 10% of group revenue.
By comparison, Singapore’s asset base is only about 2.2 times larger than China’s, but generates more than 10 times the revenue.
The figures suggest that while Raffles Medical has built a sizeable presence in China, its overseas assets have yet to achieve the same level of utilisation and productivity as its mature Singapore operations.
The company says patience is part of the plan
Raffles Medical doesn’t dispute that its overseas operations are taking time. Instead, management argues that’s simply how hospital investments work.
Advertisement
Building a hospital isn’t the hardest part—building patient volumes is.
Image Credit: Getty Images
The group says overseas operations typically require years to develop clinical capabilities, improve utilisation and reach sufficient scale before becoming meaningfully profitable.
China has also become a tougher operating environment than many expected. The company cited geopolitical tensions, technological restrictions and broader economic challenges as factors weighing on its performance.
Even so, management continues to view China as a strategic market, pointing out that around 30% of the country’s population can already afford higher-quality healthcare, giving it a sizeable addressable market.
More importantly, Raffles Medical has gradually secured access to China’s public insurance system, allowing it to treat more local patients instead of relying primarily on expatriates—a key milestone that could improve patient volumes over time.
Advertisement
China was the exception, not the rule
Image Credit: Raffles Medical Group
Despite more than a decade of overseas expansion, Singapore still remains Raffles Medical’s financial backbone. In FY2025, the group’s local operations generated nearly 90% of its revenue, effectively funding its regional ambitions while newer markets continue to mature.
Not all of its overseas markets, however, have followed the same playbook.
While China saw Raffles Medical invest heavily in building full-fledged tertiary hospitals, its expansion elsewhere has been far more measured.
In markets such as Vietnam, Cambodia and Japan, the group has focused on outpatient clinics, specialist centres and partnerships with local healthcare providers instead of embarking on similarly capital-intensive hospital projects.
That more cautious approach is reflected in its balance sheet. As at FY2025, Raffles Medical’s non-current assets in Greater China stood at about S$304 million, compared with just S$13.4 million across the rest of Asia.
Advertisement
This makes China the group’s biggest regional bet and the market that will likely determine whether its international expansion ultimately pays off.
So, was the gamble worth it?
The Raffles Hospital in Shanghai./ Image Credit: Raffles Medical Group
Hospital investments are unlike most businesses. They take years to generate sustainable returns, but there are signs that Raffles Medical’s China operations are beginning to gain traction.
In FY2025, both its Shanghai and Chongqing hospitals reported higher patient volumes, while Shanghai also recorded revenue and profit growth. The group has also expanded partnerships with leading public hospitals and secured access to China’s National Health Insurance Programme for its Shanghai hospital, moves aimed at broadening its local patient base.
Still, there’s no denying that its financials are still catching up.
If Raffles Medical succeeds in improving utilisation and profitability, years of investment could prove worthwhile. If not, its China expansion could become a costly reminder that succeeding overseas is much harder than replicating a proven business model.
Advertisement
For now, Raffles Medical appears committed to seeing the strategy through.
After spending a decade—and hundreds of millions of dollars—building its regional footprint, turning back is no longer really an option.
Read other articles we’ve written on Singaporean businesses here.
Micron and Anthropic announce four-part strategic agreement
Micron will adopt Claude models as both a daily driver and an assistant to oversee parts of its infrastructure stack
Despite billing itself as a full-stack collaboration, the agreement is silent on computational storage and processing-in-memory
Anthropic and Micron Technology have announced a new strategic agreement which will see the latter use Claude AI models to better oversee parts of its infrastructure stack.
However the move does have a curious aspect to it versus most other deals: generally, buyers tend to invest in their suppliers to support them financially while also benefiting in turn from the business they bring in.
We often see capital flowing the other way here, with Micron essentially investing in one of its largest customers for the foreseeable future.
Latest Videos From
AI to optimize memory and storage for AI consumption needs?
Anthropic runs some of the largest and most memory-hungry inference fleets in existence, and its telemetry on how HBM bandwidth, DRAM capacity, and SSD latency actually bottleneck real frontier-model serving is data Micron cannot generate internally, but it could learn how to work around these limitations while leveraging Claude to process said data to generate actionable optimizations across its organization.
Advertisement
Anthropic painted this as a solution to its scaling needs, noting that the agreement allowed it to work closer with Micron across two major segments: memory and storage.
“Our compute strategy depends on getting every layer of the stack right, and memory and storage are central to how efficiently we can train and serve Claude. Partnering with Micron means we collaborate closely on optimizing these systems for our workloads and secure the supply we need. As demand for Claude grows, this is how we scale our compute for the long term,” noted Tom Brown, co-founder and chief compute officer at Anthropic.
(Image credit: Micron)
The arguably more interesting part of this agreement is not what Micron already mentions, but what it chooses to gloss over. Not only do both companies fail to elaborate on the financial terms of their multifaceted agreement, but they also choose to skip mentioning what is increasingly becoming a core theme in AI inference workloads: Computational storage.
Advertisement
Sign up to the TechRadar Pro newsletter to get all the top news, opinion, features and guidance your business needs to succeed!
A growing share of Anthropic’s needs is inference-based, and that share is increasingly bound by memory bandwidth rather than computing power. Nvidia is already a few steps ahead in this department: at CES 2026, it announced the Inference Context Memory Storage Platform, which uses BlueField-4 DPUs to extend GPU KV cache into NVMe SSDs, a solution it calls CMX.
Micron’s (and by proxy, Anthropic’s) silence on the matter feels deliberate: the former benefits considerably from selling HBM to the highest bidder, and such solutions directly undercut or invite unfavorable comparisons to its most lucrative product lineup.
Advertisement
The latter simply has far too many options to tie itself to one particular supplier for all its inference needs; Anthropic currently has deals with AWS, Google, SpaceX, Broadcom, Microsoft, and CoreWeave to guarantee it compute, and by proxy, memory and storage needs, even as it has made strategic commitments with Nvidia to ensure it has access to its solutions.
With Anthropic’s most ambitious consumer-grade AI model, Fable 5, now back on the table, its route seems to be clear-cut: securing as much of Micron’s memory and storage supply as is possible while also making it a stakeholder in its success.
This is even as it turns to a mix of data center companies to address its short-term compute needs for a growing, and increasingly capable suite of AI models it offers to a diverse set of consumers, including governments. Its agreement with Micron is simply one of the strategic stepping stones the AI juggernaut had to take, even as it could look sideways for its computational storage needs.
Advertisement
The agreement, which has multiple facets, has been received well by investors, propped up the stock post-announcement by about 6%, with many factoring in Micron’s stake in one of the world’s most prolific AI companies positively.
Neither of the two companies mentioned the financial nitty-gritty of Micron’s investment or the supply agreement between the two even as they outlined how the planned to co-operate in the future.
This kind of deal, however, is not unique in the AI space, with Microsoft, which provided compute and cash to OpenAI in exchange for a stake in the company, and Nvidia making similar commitments with Anthropic’s rival in addition to a mix of data center and infrastructure companies, many of which are also direct customers of the world’s biggest AI hardware company.
Back in June, the UK government announced that it would ban those under 16 years old from accessing social media platforms.
While details are yet to be officially confirmed, the government has stated that under-16s won’t be able to use Instagram, YouTube, TikTok, Snapchat, Facebook and X. If you’re surprised to see YouTube included there, then visit our guide which explains all you need to know about the video-sharing platform’s social media ban.
However, one platform that’s missing from the government’s initial round-up is Discord. At the time of writing, we don’t know whether the government will eventually add Discord to the ban list or not.
In the meantime, we explain everything you need to know about Discord including whether it is classed as a social media, what safety measures it takes for younger users and more.
We’ll start with a refresher on what Discord actually is. Discord is designed for gamers and allows its users to communicate with others online, using either video or voice calls and instant messaging.
At the heart of Discord are “servers” which are a collection of chat rooms and voice channels that can be accessed either through private invite links or simply by searching. Each server can hold up to a massive 25 million users at once, though you can also create smaller and private servers for chatting with friends.
Advertisement
For more information, our dedicated what is Discord explainer goes into more detail on the platform.
Streaming on Discord via Xbox
Discord is described as being a “communications platform” that enables users to build connections around the “joy of playing games through voice, video and text features”.
So, although it does enable communication and sharing with friends, it isn’t technically classified as social media.
Advertisement
At the time of writing, Discord is not included in the list of platforms that will be banned by the UK governments for under-16s. However, the government hasn’t confirmed whether this list is exhaustive or not, so there’s potential for more platforms to be added.
Advertisement
The government has also disclaimed that it doesn’t intend for “messaging services like Whatsapp and Signal” to be included in the ban. Considering Discord is classed as a communications platform, this could suggest that the government may not see it as a social media platform.
Plus, the UK government has said that it plans to use the “same model for a social media ban as Australia”, who doesn’t include Discord in its own ban. However, many critics have since called for Australia to include Discord in its ban, as the platform allows for video chatting and live streaming.
What is Discord’s minimum age, and does it change under the ban?
The minimum age you need to be to join Discord is 13 years old, however this varies depending on where you are in the world. For example, while UK residents can join when they’re 13, some European countries like Spain and Italy require users to be 14 years old. In fact, countries including Ireland, Germany and Poland have a minimum age requirement of 16 years old.
Discord hasn’t disclosed whether it plans to change the UK’s minimum age in-line with the upcoming social media ban. That means for now, we can assume its minimum age will remain at 13 years old.
Advertisement
What other apps are and aren’t included in the ban?
So far, the apps included in the ban are: X, Snapchat, Facebook, Instagram, YouTube and TikTok. Messaging apps are “not intended” to be included in the ban, with the government explicitly referencing Whatsapp and Signal. At the time of writing, those are all the apps that we know about.
Advertisement
When does the ban come into effect?
The UK’s social media ban for under-16s should be implemented in Spring 2027, after the first set of regulations are laid out by the end of 2026.
Take advantage of holiday savings in the U.S. on AirTags, MacBooks, iPads, and more. Many of these 4th of July deals could sell out after Apple’s recent price hikes.
Holiday deals are in effect in the U.S. and Apple products are now up to $650 off. Highlights can be found below, with even more discounts in our Apple Price Guides.
Apple’s newest AirTags are still at the lowest price seen since release when you opt for the 4-pack that’s on sale for $89.
AirPods from $99
AirPods 4 prices are as low as $99 heading into the weekend.
AirPods 4 without ANC are still available for $99 at Amazon, matching Prime Day pricing.
iPads up to $400 off
iPad Pro and iPad Air models are heavily discounted at Amazon.
Advertisement
Amazon’s iPad discounts of up to $400 off are worth checking out, as Apple’s recent price hikes may result in these deals expiring or selling out soon.
That perfect black foundation is driven by the alpha 9 AI Processor Gen8, which analyses and upscales every frame in real time to sharpen detail without making anything look artificially smoothed over or overworked.
Advertisement
Brightness Booster works alongside that same processor to lift highlights and punch through glare, so the picture holds its impact whether the room is pitch dark or lit by an unforgiving afternoon sun.
That kind of responsiveness carries straight into motion handling, where the 120Hz refresh rate keeps fast-paced sport and gaming sequences fluid rather than letting quick pans smear into a distracting blur.
That same fine texture in skin, foliage and fabric gets matched by sound, with Dolby Atmos and AI Sound Pro spreading dialogue and effects convincingly around the room via a virtual 11.1.2 up-mix.
Advertisement
None of that immersive setup requires extra hardware either, since the AI Magic Remote’s dedicated AI button lets you search, adjust settings or ask questions using nothing more than your own voice.
Connectivity has been built with the same forward thinking, offering four HDMI ports, three USB ports and both Bluetooth and Wi-Fi so every console, soundbar and streaming stick stays connected without a fuss.
The webOS platform ties the whole experience together, pulling in Netflix, Prime Video, Disney Plus and Apple TV without ever needing a separate streaming box cluttering up the cabinet underneath.
If you want to see how this LG model stacks up against rivals from Samsung, Sony and Panasonic, our Best OLED TV 2026 roundup breaks down the strongest option from every major brand.
Of the many impressive aircraft developed for the United States military, Lockheed’s SR-71 Blackbird is probably the most well-known. The high-altitude reconnaissance jet has set numerous world records for speed, and many of the SR-71’s records have yet to be broken. The SR-71 is unlike pretty much every aircraft ever operated by the U.S. Air Force, and it had strict build requirements that enabled its altitude and speed achievements.
Among its many unique requirements, the SR-71 cannot use standard jet fuel, and it was built out of highly expensive titanium, much of which was procured from the Soviet Union by the CIA. Working with titanium was challenging, leading to another interesting quirk of the aircraft’s design: many of the SR-71’s parts were prone to fail, but only when they were manufactured during the summer. At the time of its initial construction, the use of titanium was relatively new, so finding out what caused the problem took some time to figure out.
Advertisement
Winter-built parts functioned seemingly indefinitely, and the unusual reason for this early issue in the SR-71’s development was something few could have imagined: water. It turns out that water mucked up the wing panels that were welded during the summer, and it had everything to do with the chlorine content of the water used during manufacturing. Lockheed solved the problem after spending some time trying to determine its cause, and it’s something Lockheed’s engineers couldn’t have imagined being a problem in the construction of one of the most secretive and advanced aircraft ever built.
Advertisement
The problem chlorinated water posed to the SR-71
Titanium isn’t an easy metal to work with, and it is highly sensitive to contaminants capable of corroding it. Wing panels needed to be welded into place, but the water used to wash the panels after an acid treatment caused the summer problem because of chlorine. The SR-71 was assembled by Skunk Works out of the Burbank Airport in California. The Burbank Water treatment plant added chlorine to the water during the summer to prevent algae blooms. Because algae prospers in warm environments, chlorine wasn’t required during the winter.
As a result, the parts welded during the summer degraded within six to seven weeks. Conversely, the parts that were worked on during the winter didn’t have this problem. The fix was relatively easy, requiring the workers to switch to using distilled water devoid of chlorine or other contaminants for the post-acid treatment. Another contaminant caused similar issues, but it wasn’t found on the parts — it had to do with the tools.
Skunk Works’ personnel working on the problem found that cadmium-plated wrenches left enough residue that weakened the bolts, causing them to fail. Cadmium, like chlorine, degrades titanium, so they replaced the tools. All of these issues resulted in a great deal of titanium and expensive tools tossed in the trash, elevating costs and delaying the production of the SR-71 Blackbird. Ultimately, it all worked out; the aircraft took to the skies, where it was most often used to spy on the very nation that unknowingly supplied much of the metal that went into its construction.
Top Kalshi trader Caleb Davies usually speaks to the press about how prediction markets help him rake in money. The Minneapolis-based IT worker estimates he’s made $1.2 million overall across different prediction platforms, with $414,000 in winnings from Kalshi’s culture markets alone. He especially enjoys wagering on music charts, because he carefully analyzes Spotify data to pick winners. “Every single morning, I’m going in, downloading the data, and updating my projections,” he tells WIRED.
This summer, though, he’s become increasingly agitated about what he claims is an obvious, bot-fueled effort to manipulate Spotify-related markets. He recently began compiling and publishing evidence for his theory, eventually becoming so convinced that he contacted Spotify, Kalshi, and Polymarket with his concerns.
This week, the situation hit a boiling point when the song “Earrings” by Malcolm Todd surged to number one on a Spotify chart. In a series of X posts, Davies outlined his suspected culprit: “botting,” or scammers who purchase bots to juice streaming numbers. Davies argued that prediction market traders were botting the charts to influence the outcome of related events contracts. Todd’s song was such an underdog that it wasn’t even listed as an option on Polymarket: “Looking at the dataset of Sunday to Monday changes, it was a 11.24 sigma event, or a roughly 1 in 77 octillion chance of happening randomly,” Davies wrote.
It turns out that he was on to something. Spotify confirmed to WIRED that it investigated suspected manipulation incidents Davies flagged and found evidence of artificial streaming. “All streaming services face ever-changing stream manipulation. Spotify has best-in-class detection and mitigation practices for manipulated streams, and we don’t pay out associated royalties,” spokesperson Laura Batey says. (The company didn’t offer any explanation for the manipulation, however, so Davies’ theory that it was directly tied to a scheme to manipulate prediction markets remains just that.)
Advertisement
Spotify ultimately adjusted its charts to account for the discrepancy, culling over 500,000 artificial streams, which bumped Todd’s song from first to fourth. The process was not immediate, though, and Kalshi had already resolved the market to award traders who selected Todd’s song.
“We’re in touch with Spotify and are actively investigating this matter,” Kalshi spokesperson Elisabeth Diana tells WIRED. Those conversations did prompt a more immediate change: At the Swedish streaming giant’s request, Kalshi removed Spotify’s logo from its markets that relate to the company, and adjusted language that initially suggested Spotify had verified chart results.
When Davies first reached out to Kalshi with concerns, the company’s head of enforcement Robert DeNault told the trader that only Spotify would be able to definitively confirm whether it had been botted, and noted that there could be non-suspicious reasons for the uptick. DeNault also floated a theory that Kalshi traders could be merely copying what peers were doing on Polymarket.
“Nobody from Polymarket profited from the fraud. That’s what undermines Kalshi’s argument, because they didn’t have a Malcom Todd bracket,” Davies tells WIRED.
Advertisement
Polymarket refutes this theory as well. “It’s actually not plausible since we didn’t even have Malcolm Todd as an option on this Spotify market,” said spokesperson Annabel Walsh. The company confirmed it’s reviewing the broader streaming manipulation situation, but hasn’t identified any immediate manipulation thus far.
No one has spoken with the people or group of people behind the streaming manipulation, so their motivations remain unclear. (Todd did not respond to requests for comment, but there’s nothing to suggest he’s anything more than an innocent bystander.)
We spend hours testing every product or service we review, so you can be sure you’re buying the best. Find out more about how we test.
ASUS ProArt PZ14: 30-second review
The Asus ProArt PZ14 arrives as one of the most impressive pieces of hardware to come through the test setup this year. At 9mm thick and weighing in at an impressive 0.79kg, it’s closer in scale to an iPad Pro than a conventional laptop, but with a 14-inch 3K OLED display, and the optional removable keyboard and Asus Pen 3.0, all powered by a Snapdragon X2 Elite processor, it proved to handle every Adobe Creative Suite application without issue.
Lightroom Classic, Photoshop, Premiere Pro, After Effects, and DaVinci Resolve all ran well, with no compatibility or performance issues that I have experienced when running ARM machines in the past. Microsoft Office ran as well as ever, and it was only when testing some of the games at the end of the test that those compatibility issues did appear. Some games, such as Indiana Jones and the Great Circle, were a no-go, while Hogwarts Legacy, suffered some graphics breakup, but beyond that, the performance was generally balanced.
Advertisement
Plugged into AC power you get the option to push the tablet into Performance mode, this boost the processing speed, and while it makes editing and game play smooth, the fans tended to kick in to quite a level.
The touchscreen display is the best 3K resolution display I’ve seen, and for creative work, the 144Hz, Pantone-validated panel with an anti-reflective coating looked great and handled bright outdoor conditions better than most OLED panels, with limited reflections.
One of the weak points was the detachable Bluetooth keyboard, which flexes at the joint, as is common with this type of design, making any use away from a solid desk difficult. Early on in the test, I realised that at a desk, the keyboard was great, away, it was best to use the Asus Pen 3.0 or a finger as the primary input method. Not only did this make sense, but it also transformed the machine’s use.
Advertisement
Overall, however, the power for the size and price is exceptional, and in the creative workflow, like the iPad Pro, it can integrate with the project from beginning to end, from being used to tether to your camera through to the editing, uploading and distribution of your work.
ASUS ProArt PZ14: Price and availability
(Image credit: Alastair Jennings)
How much does it cost? $2000 TBC
When is it out? Now
Where can you get it? From retailers such as Best Buy
Sign up to the TechRadar Pro newsletter to get all the top news, opinion, features and guidance your business needs to succeed!
Outside of the US, I can’t see much availability. UK pricing for the 32GB, 1TB configuration with the ASUS Pen 3.0 and Bluetooth keyboard has not been confirmed at the time of writing. Going on what has come before, expect the price to be around the £2,000 mark in the UK.
Advertisement
ASUS ProArt PZ14: Specs
CPU: Snapdragon X2 Elite X2E-88-100, 18 cores, up to 4.7GHz GPU: Qualcomm Adreno X2-90, 128MB dedicated GPU memory NPU: 80 TOPS RAM: 32GB LPDDR5X Storage: 1TB PCIe 4.0 SSD Display: 14-inch ASUS Lumina Pro OLED, 3K Connectivity: 2 × USB4 (40Gbps), SD card reader, WiFi 7, Bluetooth Battery: 75Wh Cameras: 8MP IR front, 13MP rear Durability: IP52, MIL-STD 810H OS: Windows 11 Home Weight: 0.79kg, 9mm thin Accessories optional: ASUS Pen 3.0, Bluetooth keyboard, stand cover
ASUS ProArt PZ14: Design
The ProArt PZ14 is an exceptionally well-built tablet, and the hardware is very minimal. On the left side of the screen are the two USB4 ports that let you plug into the mains and attach accessories, and these are joined by an SD card slot, which is covered. Around the edges are plenty of ventilation slots that help expel the heat away from the Snapdragon processor and electronics inside.
Advertisement
Lifting the tablet from the packaging, the Nano Black CNC-machined aluminium chassis gives an instant premium feel, and the finish proved to be pretty much smudge-resistant. Over the month of testing, the build quality proved exceptional as it was moved between the studio, office, and location work, and used at home in the evenings while catching up on TV.
The aluminium chassis gives the table a tough feel, and this is paired with the IP52 rating, which means it can withstand a light rain shower for a short time without issue, but doesn’t go as far as being left out in it.
During the test, I found that the detachable keyboard supplied with the PZ14 was fine for use on a desk, where it offers a decent enough typing surface with good key travel, and likewise, the touchpad. Away from a desk, however, the flex between the keyboard and tablet can make it difficult to use; this is a tablet.
Think of it this way: in the office/studio, the keyboard is a traditional input device; in the field, it’s just part of the protective shell, with the Asus Pen 3.0 and on-screen keyboard handling input.
Advertisement
(Image credit: Alastair Jennings)
I’m not used to using a pen to interact with a screen; however, the Asus Pen 3.0 with MPP 2.6 support and haptic feedback is excellent. It paired instantly, and, in Lightroom, Photoshop, and Asus’ own StoryCube and MuseTree apps, offers a more natural and intuitive way to navigate than the trackpad.
On location with the tablet held in hand, I started using the pen as the primary input method, and it worked far better than the keyboard and I expected.
The magnetic stand cover, of which the keyboard is part, enables you to fold out a small stand, which is handy when everything is supported on a desk, and there’s plenty of flexibility over the angle. In the studio, I tend to use a wired mouse and an external keyboard connected via USB4 to provide a fast input for video editing, which is standard.
Advertisement
ASUS ProArt PZ14: Features
(Image credit: Alastair Jennings)
The PZ14 has been designed with the Creator in mind and features a decent array of creative-focused apps that will help you organise files and generate ideas. The ProArt range has in the past been closely tied to Adobe products, and here the Adobe Creative Suite performance is in mind; the Snapdragon X2 Elite does seem to have been optimised.
Lightroom Classic, Photoshop, Premiere Pro, After Effects, and DaVinci Resolve all opened and ran without compatibility errors on the Snapdragon X2 Elite X2E-88-100, 18 cores, up to 4.7 GHz, with thumbnails and timelines rendered at speed, supported by the Qualcomm Adreno X2-90, 128 MB of dedicated GPU memory, and graphics.
Inside is a relatively small 1TB SSD, but this provides exceptional transfer rates, exactly what you need when transferring large quantities of files, although it would have been good to see at least 2TB with the file sizes associated with the latest cameras.
Testing the capabilities of this hardware, I checked the Adobe Bridge thumbnail rendering for images from a Sony a7 IV and a Canon EOS R5 C RAW files and was impressed by the speed, though the noise from the cooling system was louder than expected.
Advertisement
Lightroom’s touchscreen workflow seems to be made for the system, scrolling through the library and making adjustments with the slide controls; it felt natural on the 14-inch ASUS Lumina Pro OLED (3 K).
Alongside the CPU and GPU is an 80-TOPS NPU that enables Copilot+ features and boosts AI-powered tools in Premiere Pro and Photoshop. Generative fill and timeline expansion both ran smoothly and quickly. Local LLM inference via LM Studio was downloaded and was functional with a few issues.
However, if you really want to reach your full potential, StoryCube, Asus’ AI media management app, highlights exactly what can be done and works well alongside Adobe Bridge, offering useful automatic organisation of RAW files and video assets.
The hardware features of the tablet design differ slightly from those of most laptops and mini PCs, so there are some compatibility issues. For instance, when gaming, ARM isn’t always as compatible as other systems.
Advertisement
Indiana Jones and the Great Circle would not load at all due to a Vulkan plugin compatibility error. Hogwarts Legacy loaded slowly but ran smoothly once started, aside from some graphics break-up, and always with the background of the fans working hard. PCMark 10 and the Windows Experience Index did not run at all, even with the usual workarounds.
Through the test in the ProArt Creator Hub, I shifted the resources into Performance, although this mode is only accessible when plugged directly into the mains. While in all other modes, cooling and noise are well balanced, everything is cranked up to 11, including the fans. However, in the 0dB whisper mode, fans are kept off during light use, meaning the system runs silently.
In Performance mode, which I used for video editing, the fans activate during rendering and timeline processing, and are noticeably audible. If you’re working on audio-sensitive work, it’s best to switch to one of the quiet modes and then back when that content needs editing or wear headphones.
Advertisement
ASUS ProArt PZ14: Performance
(Image credit: Alastair Jennings)
As with most laptops and mini PCs, the tablet is almost ready to run out of the box. To get started, I plugged in the power supply and gave the tablet a full charge. It’s worth noting that the power supply is 65W, and while it is USB-C, if you plug in a higher-wattage USB-C connection, the machine will prompt you to use the provided power supply or one of equivalent rating.
Once the final steps of the Windows 11 Home setup were complete, I could install all the usual apps. It was apparent from the outset that the internal SSD was fast. With files downloading from my external storage nice and quickly. Sure enough, the benchmarks highlight results of 6,065 MB/s read and 5,356 MB/s write, really showing that the PCIe 4.0 SSD was close to the upper limit.
As the machine filled with all the test applications, the 1TB internal SSD quickly filled with programs and files, including most of the Adobe Creative Suite, local LLM models and video project files; games had to wait due to space limitations.
Storage is always a consideration, and with the ProArt PZ14, if you’re working with video, you’re going to have to supplement the internal storage with a USB4 external SSD or, in this review, a direct NAS connection. During the test, editing 4K Canon EOS R5 C footage via a Ugreen DXP4800 GT NAS connection delivered approximately 800MB/s transfer speeds over USB4-to-10GbE, which was more than enough for smooth Premiere Pro timeline editing.
Advertisement
Over the test period, using Lightroom Classic was one of the major highlights, tethering a camera to the tablet so I could shoot, adjust, and share files with other members of the team. Touch input via the screen, combined with pen interaction for adjustments, just made things easy and far more natural than using my MacBook Pro M1 Max for on-location work.
After the shoot, editing images in batch in Lightroom and editing footage in Premiere Pro were handled well by the system. Run in performance mode, the fans were on constantly; however, switching down to standard mode in Photoshop and Lightroom still provided a good balance of speed, but did cut out the fan noise.
When it came to video editing, a pair of headphones helped me to focus on the vocals and video audio rather than the noise from the machine. It’s also worth noting that Performance mode is only available when plugged into the mains, Standard is the default.
Benchmark results
Advertisement
CrystalDiskMark Read: 6,065.91 MB/s CrystalDiskMark Write: 5,356.66 MB/s Geekbench 6 CPU Multi-Core: 9,976 Geekbench 6 CPU Single-Core: 2,908 Geekbench 6 GPU: 39,018 Cinebench CPU Multi: 4,345 Cinebench CPU Single: 634 3DMark Fire Strike Overall: 8,174 3DMark Fire Strike Graphics: 8,787 3DMark Fire Strike Physics: 19,321 3DMark Fire Strike Combined: 3,423 3DMark Time Spy Overall: 3,522 3DMark Time Spy Graphics: 3166 3DMark Time Spy CPU: 9737 3DMark Wild Life Overall: 31,823 3DMark Steel Nomad Overall: 3,788 PCMark 10: N/A Windows Experience Index: N/A
Two other big points about the performance are the screen, which is just exceptional, with clean, clear detail and tone that, once calibrated, was absolutely ideal for all areas of creative work.
The other is the battery life, which just seemed unending. Usually, when video editing, image processing, or processing endless documents, this processing draws additional power, especially when accessories are also plugged in and drawing on resources.
However, with the combined use, the battery lasted all day, and a two-hour gaming session in Hogwarts Legacy the battery stood at 70%.
Advertisement
One other point to make is compatibility; while most applications load, some won’t. During the test, the usual benchmarking software, PCMark, refused to run a full test, and games such as Indiana Jones would not run either.
ASUS ProArt PZ14: Final verdict
(Image credit: Alastair Jennings)
The Asus ProArt PZ14 is a great machine for photographers who work in the field and need a slim, light, colour-accurate display for Lightroom editing and camera tethering. It’s also a decent option for lightweight Premiere Pro and DaVinci Resolve editing, especially as a, with a display quality that makes colour grading reliable even without a secondary monitor.
The keyboard is a bit of a compromise, depending on how you work. On a desk, it is adequate. Away from a desk, it’s frustrating, as this design tends to be with the flexible connector.
Advertisement
The Asus Pen is excellent and, in many workflows, becomes the better primary input device. You also have to consider the noise from the fans, which did seem louder than with past models, likely due to the increased performance.
ARM compatibility for creative applications is not an issue with all of the Adobe apps, Microsoft Office and the Asus AI apps running without issue. This is a tablet PC that would suit enthusiasts, photographers, and students, providing performance in a very flexible format and a great alternative to a laptop and actually better suited in many situations.
Should I buy the ASUS ProArt PZ14?
(Image credit: Alastair Jennings)
Swipe to scroll horizontally
Value
Advertisement
Well balanced for the OLED display, ARM performance and all-day battery
4/5
Design
Exceptional tablet chassis and display; although the keyboard flexibility is an issue
Advertisement
4.5/5
Features
Large touch screen, powerful CPU, USB4 and all day battery life
4.5/5
Advertisement
Performance
Snapdragon X2 Elite handles all creative applications and workflows; however, 1TB SSD fills quickly, and fans are audible under load
4/5
Overall
Advertisement
A great option for field photography and Lightroom, especially in the tablet format. Video editors will need external storage from day one.
Don’t count on the LLM to return your data – even if you pay up
They’re not bad; they’re just prompted that way. Sysdig threat hunters documented what they say is the first-ever documented agentic ransomware infection with an LLM – not a human – driving the entire extortion operation, from gaining initial access to compromising a production database server and destroying data.
The security shop’s research team named the agentic intruder JadePuffer and said it gained initial access to an internet-facing Langflow instance by exploiting CVE-2025-3248, and then ran a fully automated attack.
Advertisement
“The most striking characteristic, however, was the LLM’s behavior,” Sysdig director of threat research Michael Clark said in a blog about the agentic ransomware and extortion operation.
JadePuffer’s “self-narrating” payloads “contained natural language reasoning, target prioritization, and the kind of detailed annotations that human operators don’t often write but LLM-generated code produces reflexively,” Clark added. “The operation also adapted in real time, retrying failed steps within refined parameters. In one sequence, it went from a failed login to a working fix in 31 seconds.”
After exploiting CVE-2025-3248, a missing authentication vulnerability in Langflow that allows remote, unauthenticated attackers to execute arbitrary Python on the host, the AI agent began scanning for and collecting secrets, including LLM provider API keys, cloud credentials “with explicit coverage of Chinese providers” including Alibaba, Aliyun, Tencent, and Huawei, while also scanning for AWS, Azure and Google Cloud Platform, cryptocurrency wallets, and database credentials.
The AI also installed a crontab entry on the Langflow server to maintain persistence and call back to the attacker’s infrastructure every 30 minutes.
Advertisement
JadePuffer’s intended target was a separate internet-exposed production server running a MySQL database and an Alibaba Nacos configuration service, we’re told. Nacos is an open-source service-discovery and dynamic configuration platform developed by Alibaba and used in the cloud provider’s microservices applications.
The agent connected to the server’s exposed MySQL port using root credentials, although Sysdig doesn’t know how the attacker obtained them. These credentials weren’t stolen from the victim’s environment.
JadePuffer then attacked Nacos via multiple vectors including an authorization bypass flaw (CVE-2021-29441) and forging a valid JSON web token (JWT) using Nacos’s default signing key. Additionally, using its root database access, the LLM injected a backdoor administrator into the Nacos backing database.
It ultimately encrypted all 1,342 Nacos service configuration items using MySQL’s built-in AES encryption function, and created an extortion demand, ransom note, Bitcoin payment address, and a Proton Mail contact:
Advertisement
“YOUR DATA HAS BEEN ENCRYPTED. All NACOS configurations, REDACTED customer data, and REDACTED PII have been encrypted with AES-256.”, “3J98t1WpEZ73CNmQviecrnyiWrnqRhWNLy”, “e78393397[@]proton[.]me”
However, according to the threat hunters, the victim can’t recover the encrypted data, even if they paid the ransom demand, because the agent escalated “from row-level deletion to dropping entire database schemas, narrating its own targeting rationale,” without backing up any of the encrypted data.
There are a couple of things that security teams and vulnerability managers should do immediately to avoid being ransomed by this AI agent. First up: patch Langflow to a release that fixes CVE-2025-3248, and do not expose code-execution/validation endpoints to the internet.
Also, don’t ever expose Nacos to the open internet, change its default token.secret.key, and upgrade to a release that forces a custom key.
Advertisement
The threat hunters also recommend against running any AI orchestration servers with provider API keys or cloud credentials in their environment.
While the AI agent didn’t use any especially sophisticated or unique techniques in this attack, the fact that an LLM “strung them together into a complete ransomware operation against neglected internet-facing infrastructure,” is notable, according to Clark. “The skill floor for running ransomware has dropped to whatever it costs to run an agent, and if that agent is running on stolen credentials through LLMjacking, the cost to an attacker is close to zero.”®
This is one of those projects that was both inspired and made possible by the absolute embarrassment of dev boards available to the modern hacker. In this case, the dev board was the M5Stack PaperS3, which as the name implies combines an ESP32-S3 with an e-ink panel. [Wenting Zhang] picked one up and was immediately inspired to try and make an e-ink Game Boy.
The M5Stack PaperS3 made this project possible by exposing the display with row/column control — parallel, some would call it, as opposed to the usual serial interface of SPI. That allowed [Wenting] to work some of the same e-ink magic he perfected on his Modos monitors to allow partial refresh at up to 60 Hz. That the ESP32-S3 is capable of emulating a Game Boy while driving the screen should surprise no one, since it can emulate an MSX while outputting VGA or even Windows 95 on a 386. In this case, he’s basing the actual Game Boy emulation on Crank Boy.
Of course the e-ink screen on the M5Stack is far larger and has a much higher resolution than what the Game Boy shipped with, which lets him implement touch controls and scale the image up 3X so he can fake a couple of shades of grayscale while actually outputting black and white. Even better, if he was actually playing this thing on the regular, once the high-refresh portion of the screen starts to wear out, he can flip the orientation and keep gaming on the virtually-unrefreshed control portion of the screen — doubling the lifetime of the system, something many of you raised as a concern when we last looked at a his e-ink monitor project.
Advertisement
The only real shortcoming of this hack is the sound. With one-bit beeps coming out of the M5Stack buzzer, it’s got nothing on Nintendo’s hardware. Of course, that’s partially down to using the hardware as-is. With the addition of an I2S sound chip like the one used in the MOD player project we featured recently, you’d just need to squeeze out enough processor cycles to make this sound as good as it looks.
You must be logged in to post a comment Login