TechRadar News

• For now you need to build the feature yourself on a Mac
• Native Player delivers similar audio but isn't open source
• Proposed as an extension to the Nuvio media player

A clever developer has come up with a way to deliver lossless Dolby TrueHD movie soundtracks, complete with Dolby Atmos effects, to home theater speakers from the third-gen Apple TV 4K.

The Apple TV only officially supports Dolby Digital Plus, which is the more compressed Dolby audio codec used by streaming services to reduce the amount of data used, while Dolby TrueHD is used on Blu-rays where bandwidth isn't a problem. So it makes sense for the Apple TV 4K to only use Dolby Digital Plus, since it's a streaming device — but that doesn't stop the enthusiasts from wanting the flexibility of playing files with TrueHD audio on their devices.

The system that Reddit user Agathorpt used to deliver Dolby TrueHD through the Apple TV 4K requires not actually delivering it in TrueHD at all, but converting it to other formats, converting the Dolby Atmos positional data into other formats too, then recombining them in a format that the Apple TV 4K can deliver, but in a higher quality.

How they got TrueHD Atmos on the Apple TV 4K

As mentioned above, the audio takes a bit of a journey here. The path is:

TrueHD Atmos → decoded audio + timed object metadata → fixed 9.1.6 speaker render → APAC → native tvOS playback.

APAC is Apple's Positional Audio Codec, and while it's not truly lossless (the source audio, TrueHD, is a lossless format but APAC is not) the consensus is that the results here are really impressive.

Basically, the system decodes the audio and the 3D Atmos positioning data separately, creates a new audio system where the Atmos effects are mapped to a 9.1.6-channel speaker configuration, then converts that to Apple's own proprietary spatial audio codec, which can then be played by the Apple TV 4K.

The goal was to replicate the audio functionality of a closed-source app called Native Player with an open-source alternative because "keeping the implementation closed… leaves everyone dependent on that vendor for fixes and future support," Agathorpt says. "Other players can't reuse or improve that code, and the community can't independently inspect how it handles the audio."

Apps such as Native Player support TrueHD Atmos but they aren't open source (Image credit: Native Development)

"The spatial audio has been working amazingly well for me," Agathorpt says. "It sounds excellent on my setup." Other redditors have tried the same software and are reporting great results too.

In order to copy what Agathorpt has created, you'd need several things: a Mac; Apple's Xcode development software; and your own Apple development account to sign an app: the full steps they went through are detailed here. The Apple TV 4K needs to be running tvOS 26 or later, and of course you'll also need movie files with 48kHz TrueHD Atmos audio.

There is hope from home theater enthusiasts that Agathorpt's project will be implemented in the Nuvio player for everyone and in other tvOS apps too, but for now you'll need to build it yourself. It doesn't seem too difficult to do, but if you're not the technical type you might want to wait for it to roll out as part of the app. It's been proposed for addition so hopefully that won't take too long.

Agathorpt is at pains to point out that they didn't write the software projects used in this process. It builds on the work of others, especially Superuser404notfound's work on the AetherEngine media engine and LibTrueHD object audio decoder.

One assumes that this will work with the new Apple TV 4K (2026) that Apple just announced too, though its price hike and lack of big new AV features means fans aren't particularly enamored.

Thinking of buying a new TV?

Try our TV size and model finder! You tell it how far you sit from your TV, we'll tell you what size to buy based on viewing angle advice from image quality experts, and we'll recommend our three top TVs at that size for different prices.

• WindBorne's weather balloon fleet secures a potential $49 million Air Force contract
• The military has ordered $1.7 million in data so far
• WindBorne operates about 700 balloons collecting atmospheric measurements worldwide

WindBorne Systems has secured a potential $49 million US Air Force agreement, allowing the military to purchase atmospheric measurements gathered by autonomous balloons worldwide.

The California startup operates about 700 balloons, supplying readings to paying customers that use the data to develop AI systems capable of forecasting weather conditions.

Unlike conventional aircraft, WindBorne's balloons change altitude to encounter winds carrying them toward particular locations, adjusting buoyancy through air release and sand ballast.

Military spending supports WindBorne's expanding balloon network

The company uses atmospheric models to guide these movements, combining predictive software with an extensive fleet to place balloons where useful measurements can be collected.

Back in August 2026, WindBorne became the first company to guide an unpowered flight system into the typhoon's eye, collecting data from the typhoon.

“That’s a pretty novel step, and the two ingredients for that are one, having good wind and weather models, and knowing how to navigate them; and two, having a large enough constellation that you have balloons in the right places,” said Andrey Sushko, co-founder and CTO of WindBorne Systems.

“It’s one of the first times that a deep-learning model was used for actual operational decisions.”

The military agreement allows the Air Force to request measurements for particular locations and periods instead of purchasing a fixed quantity of information upfront.

Although the deal carries a maximum value of $49 million, actual spending depends on military needs and so far, the military has placed orders worth about $1.7 million.

“This is largely a vehicle through which continued growth of the constellation can be supported, as well as some work on research and development to continue to improve the hardware and performance of the balloons,” Sushko told TechCrunch.

The contract was awarded without competing bids, and Sushko said the government determined that “for this capability of collecting global data, we were the only company able to provide it in the way that met their needs for these operations.”

AI forecasting gives balloon measurements additional value

WindBorne also incorporates balloon measurements into WeatherMesh 6, its forecasting system, which the company claims performs better than the European Centre for Medium-Range Weather Forecasts' traditional approach.

The European Centre relies on atmospheric physics simulations, while machine learning systems derive predictive relationships from historical observations and generate forecasts using learned patterns.

Neither approach eliminates the need for reliable measurements, because forecasting systems depend on atmospheric information to estimate developing conditions and assess possible future changes.

AI forecasting has advanced alongside transformer architectures and access to large historical weather datasets, enabling researchers to develop models without repeatedly solving conventional atmospheric equations.

The company's claimed forecasting advantage remains its own assessment, and independent comparisons are necessary to establish how consistently WeatherMesh 6 outperforms established systems across different weather conditions.

It's been a theory for quite some time now that Nvidia is — what's the polite way to put this? — let's say deprioritizing PC gamers, perhaps, or more specifically, its gaming graphics cards. But this week has added more evidence to the collective weight that GeForce GPUs are increasingly at risk of being sidelined further by Team Green.

I'm talking first about the story that emerged a few days ago, where a leaker claimed Nvidia is diverting all GB202 (Blackwell) chips to make AI (RTX Pro) graphics cards. That's the graphics processor used in the RTX 5090, but apparently not anymore; it's for AI boards only now, we're told.

That makes some sense as the RTX 5090 is already disappearing from shelves, certainly in the US, where you can't buy one for less than the best part of 10 grand (from gouging third-party sellers, with no major retailers actually having stock anymore). So arguably this effective discontinuation of the RTX 5090 is already happening.

On top of that, the latest developments are that Micro Center is reportedly listing the Blackwell flagship as 'not carried', and Nvidia itself has removed all mention of the RTX 5090 from its own GPU marketplace (though it is still mentioned elsewhere on the site). Search for the RTX 5090 on that marketplace, and you won't find it any longer — the highest-end graphics card listed is now the RTX 5080 (hat tip to Wccftech for spotting this being flagged up on Reddit).

Aside from an actual announcement from Nvidia (which probably wouldn't ever happen anyway), that seems like about as clear an indication you can get that the RTX 5090 has been shoved out of the exit door, never to return, because Nvidia needs those GB202 chips where they are most profitable. RTX Pro GPUs offer a much heftier premium that Team Green can heap on top.

A new Blackwell flagship?

(Image credit: Future)

So, what does this mean? For starters, if it is true that the RTX 5090 has been scrapped, the RTX 5080 becomes the flagship Nvidia GeForce GPU. Once the existing stock of RTX 5090s has been depleted, this could mean the price of the RTX 5080 is inflated. When I say inflated, I mean jacked up even more, because it has already seen some nasty price hikes. Currently, the RTX 5080 is at around $1,800 in the US on Newegg, or £1,450 on Overclockers in the UK – and those are the lowest prices; it goes up from there. It's a very similar story in many other regions.

This is clearly a concern for gamers as you can see from the Reddit thread I pointed out above, with comments such as: "If you want a 5080 better buy one while you can still afford one." Really, it's debatable if the card is anything like 'affordable' relative to its graphical frame rate muscle even now.

Here's where another rumor comes into play. Elsewhere on the grapevine, there's talk that Nvidia may actually release the RTX 5080 Super variant which has long been speculated about. This would be a 24GB graphics card, and it'd essentially replace the discontinued RTX 5090 – with the theory being that Nvidia would also ditch the vanilla RTX 5080 (16GB). This would leave the rumored RTX 5080 Super as the sole high-end GeForce graphics card, quite some distance above the RTX 5070 Ti.

If that happens, imagine the pricing concerns here – what MSRP would Nvidia attach to the RTX 5080 Super? And then consider the cost inflation that will be added when you take into account that this card will also be sought after for AI usage (as seen with the RTX 5090).

If an RTX 5080 Super is indeed inbound for early 2027, that would also suggest that the next-gen RTX 6000 family is further away than some rumors believe – as this Super version would be a stopgap GPU to keep the Blackwell range ticking over. The RAM crisis may have a fair bit to do with this strategy also, given that most predictions foresee it getting worse next year (and the misery continuing into 2028, and maybe beyond). Of course, launching a new GeForce range into such memory headwinds wouldn't be easy.

Software focus and streaming dreams

(Image credit: Nvidia)

All in all, it feels to me like there's a stream of subtle – and not so subtle – hints and rumors that Nvidia is very much slowing down the development of gaming GPUs (while not wanting to give that impression, naturally). The other part of the equation here is the increased focus on the software side, of course, rather than the hardware. DLSS 5 is out and Nvidia is already pushing hard with honing and refining that neural rendering tech, and something else that happened earlier this week was that Team Green highlighted some big leaps in performance here. Pretty huge strides forward, in fact.

Of course, you could argue that all the development resources being ploughed into DLSS – including frame generation, which got a hefty boost alongside DLSS 5 – shows Nvidia isn't abandoning gaming GPUs for PC gamers. And I concede that's true to an extent, but there is very much a vibe of a major slowdown leading towards the feared deprioritization of GeForce GPUs – and there's another angle to consider here too: cloud gaming.

Don't forget that GeForce Now also benefits from DLSS. Just this week, Nvidia added DLSS 4.5 override controls to the streaming service to help push up frame rates – those on the RTX 5080 (Ultimate) plan can use Dynamic Multi Frame Generation of up to 4x now. Ray Reconstruction for bolstering ray-traced visuals is available, too, among other options.

Would Nvidia rather have you subscribed to its cloud gaming service, making monthly payments, than buying a GeForce GPU for a one-off sum? It seems an obvious direction for the company to be heading in, and with increasing warning signs around Team Green's gaming GPUs, I'm becoming more convinced that this is where Nvidia's priorities are shifting for the future of PC gaming.

This isn't to say GeForce graphics cards are definitely on their way out, or that an RTX 6000 line-up won't happen – it's already well underway after all – but beyond the next generation, I'm increasingly dubious about where GeForce GPUs might fit into the big PC gaming picture. Between AI demand and cloud gaming (plus increasingly faster internet connections), there's a squeeze underway that could prove painful for the GeForce line-up – and the ongoing RAM crisis is hardly helping, either.

This is a laptop that's designed for creators who are on the go. The 15-inch screen is big enough to give you enough workspace to do what you need on the go, without being so cumbersome that it won't fit in a standard laptop backpack. 15 inches is usually the sweet spot that makes or breaks whether it fits in a standard bag.

While this may not have full 4K, it has a crisp, color-accurate display, making it fantastic for those who need absolute perfection in color profiles. It is also powerful enough to handle almost anything you throw at it. While it's not designed for the average business user, those who dabble in creative work but still need to get business done can do so with ease on this laptop.

It has enough dedicated GPU to carve through video, enough RAM to run several lines at once, a beautiful enough screen to edit photos without a problem, and a solid enough machine to handle everything else, all in one device.

In short, if you want to get this kind of work done while still on the go, this is the machine for you. If you plan to park it at a desk or workstation most of the time, the 9i may be a better fit. Either way, this is for the creator.

Lenovo Yoga Pro 7i: Price and Availability

The Lenovo Pro 7i Gen 11 Aura Edition is available direct from Lenovo in the US, in the UK, and other territories. Prices start at $2,599.99 / £2,430 for the base configuration.

My review unit, however, had a slight upgrade and was sitting around $2,700. The Yoga Pro 7 could be configured up to a Core Ultra 9 with an RTX 570, 64 GB of RAM, and 2 TB of SSD.

Compared to the Lenovo Yoga Pro 9i Gen 11 Aura Edition I just reviewed, which sits around $3,000 for the same configuration, the Pro 7i saves about $250 for the same CPU, GPU, RAM, and SSD storage.

That means it is up to you whether it's worth saving $250 for giving up an inch of screen, or perhaps gaining an inch of screen for portability, depending on which camp you're in

(Image credit: Collin Probst // Future )Lenovo Yoga Pro 7i: Unboxing and First Impressions

I'll admit it, life has gotten crazy. I've had this laptop in my queue, sitting on my testing desk or in testing, for 178 days.

I've used it for: numerous meetings, a few days in my life to see how it operates, several articles have been written on it, a lot of testing, some benchmarking, thrown in the backpack a couple of times, used it as my main machine while in my home office for a couple of days

All around, this machine runs fantastic. The Luna Gray color is fantastic. It boots quickly, and it runs fast. The 15-inch OLED touchscreen is spectacular and accurate. The colors are incredible, and the ports are great. Overall, it's a laptop most people would dream of having.

I even prefer the 15-inch over the larger 16-inch 9i I just finished testing, though it is still a tad large for my liking. 14-inch seems to be my sweet spot. The 15-inch is still manageable and fits in most full-size laptop backpacks. For example, it fits well in my Nomatic Navigator and my Nomatic backpack.

(Image credit: Collin Probst // Future )Lenovo Yoga Pro 7i: Design and Build Quality

As expected, the keyboard and trackpad are spectacular. Like the 9i variant, the trackpad also doubles as a drawing surface for the Yoga Pen. Since I adjusted on the 9i, it was much easier for me to get used to it. However, as I said in the 9i review, if this is your first time using a drawing surface like this, it will take a second to get used to. It's not quite as seamless as drawing on the screen, but it becomes fairly intuitive after a little use.

The 15-inch 2.5K display is hard to recognize as only 2.5K. It looks significantly nicer than that. It could be because of the colors and the 165 Hz refresh rate, but this display (HDR True Black 1000, 100% sRGB, P3, Adobe RGB, 1,100 nits peak brightness, and PureSight) makes this OLED panel look absolutely spectacular. Unless you know for a fact you need a 4K display, you can use this for creative content pretty easily.

The last thing I'll mention is that the port layout is also what I would hope for. I've never been a fan of laptops that shove ports on the back and make it seem like they're meant to stay stationary on a desk. Even though I love the thinner laptops, I've never been a fan of the laptops that only have USB-C. I get it, and I get what they're going for, but in the professional space, that's really hard to justify.

For ultrabooks, that makes sense, but for professional business laptops or even creative laptops like this one, having only a USB-C means that you are guaranteed to need a dongle. This laptop, though, has a good range of ports for its intended use, and it's designed for that use case, which is ultimately what a laptop should be tested against anyway.

(Image credit: Collin Probst // Future )Lenovo Yoga Pro 7i: In use

I've been able to do some light photo editing and some video editing, and more, on this machine. While creating, I actually did a lot of content approval on this laptop during my testing. My roles in other companies have shifted. For now, I review content my team makes daily, so I appreciate this screen more and more.

The OLED is absolutely spectacular and has made me want an OLED panel for pretty much everything that I have. 165 Hz is also buttery smooth. The RTX 5060 is absurdly powerful. I tested it by running 22 lines of a security system I was setting up for a client without any issues. It barely taxed the memory, and I ran all 22 lines at full speed without any issues.

Of course, because even creatives still need to write some things out, I tested the keyboard thoroughly as well. It's a comfortable keyboard, just as I expected from Lenovo. It's comfortable to use for days on end, hours each day, and you won't need a different laptop keyboard unless you want one purely for aesthetic or ergonomic purposes.

(Image credit: Collin Probst // Future )Lenovo Yoga Pro 7i: Final verdict

The question that we land on is: should you buy this laptop? If you want a more portable but powerful creator workstation that isn't difficult to carry around in a backpack, then yes, you probably should buy this. It does everything mobile photo and video editors will need it to do.

However, if you need the extra screen space afforded by the 9i, I can see why that'd make sense for many deskbound creatives.

Either way, either laptop is a spectacular buy if you're in the creative space and don't need a true 4K display, because the screen itself, with its color accuracy, visuals, and refresh rate, is absolutely gorgeous. You get the reliability and dependability of the Lenovo family, all in a machine designed for creators to do exactly that: create.

Collin Probst // Future Collin Probst // Future Collin Probst // Future Collin Probst // Future Collin Probst // Future Collin Probst // Future

For more creator machines, we've tested the best photo editing laptops and the best laptops for video editing.

Kiwibit Bird Feeder 2 Pro: two-minute review

Smart bird feeders have quietly become one of the busiest corners of the camera world. Hang one in your yard, garden or on a balcony railing and it’ll snap and film every feathered visitor automatically, often identifing the species for you, too. We’ve tested plenty of them here at TechRadar — you’ll find the best in our guide to the best bird feeder cameras — and the Kiwibit Bird Feeder 2 Pro is the latest to land in my garden.

If it looks familiar, that’s because it’s clearly a descendant of the Kiwibit Beako, which we tested last year. The headline upgrade is in the camera: Kiwibit says this is the first bird feeder camera to combine 4K HDR with phase-detection autofocus (PDAF), the kind of focusing tech you’ll find in a decent smartphone. The other big change is a roof with solar panels built into it, so there’s no separate panel to mount and wire up.

(Image credit: Future | Sam Kieldsen)

That roof turns out to be one of the feeder’s best features. After an initial charge I never needed to plug it in again during testing, and the battery hovered near full. The rest of the hardware impresses too: the plastic shell is tough and well finished, mounting it took minutes, and lifting out the two-chamber seed hopper for a refill or clean is about as easy as it gets.

The camera is a game of two halves, though. Photos and 4K clips are crisp, and the autofocus generally locks on to birds reliably — though ones that hop right into the seed tray can get too close for it focus. The biggest frustration is motion: 4K is limited to 15 frames per second, which makes footage look juddery, and the 120fps slow-motion mode currently produces clips that are little better than slideshows. The 1080p 60fps mode is the pick of the bunch.

Then there’s the app, which is cluttered and hard to navigate, and keeps its most compelling feature — AI species identification — behind a Kiwibit Plus subscription. That ID system was accurate in my testing, even if it can only name one species per clip. Taken as a whole, the Bird Feeder 2 Pro’s hardware is impressive, but its software and video modes need to catch up.

Kiwibit Bird Feeder 2 Pro: price and availability

• $279.99 in the US with 'lifetime AI' included, or $219.99 without
• £289.99 in the UK; launching Q1 2027
• Kiwibit Plus subscription priced from $7.99 per month

The Kiwibit Bird Feeder 2 Pro is currently on sale in the US for $219.99, or $279.99 with a lifetime subscription to the Kiwibit Plus service. It’s due to arrive in the UK in the first quarter of 2027, priced at £289.99.

That price puts it in premium territory, alongside the likes of the Bird Buddy 2. And there’s a running cost to factor in, because AI species identification and several other app features require a Kiwibit Plus subscription, which costs from $89.99 (a UK price is TBC) per year or $7.99 a month.

Kiwibit Bird Feeder 2 Pro: key specsCamouflage EZ BirdFeed specs

Photo

Up to 12MP

Video

4K at 15fps; 1080p at 60fps or 120fps

Lens

130-degree field of view

Memory

Up to 128GB microSD card (not included)

Dimensions

9.8 x 8.7 x 8.9 inches / 25 x 22 x 22.7cm

Weight

4.08lbs / 1.85kg

Connectivity

USB-C, Dual-band Wi-Fi 6

Battery

5,200mAh battery, 5W solar panel

Kiwibit Bird Feeder 2 Pro: design

• Solar panels built into the hinged roof
• Removable 1.7l dual-compartment hopper
• IP65 rated and works in -4F to 122F (-20C to 50C) temperature range

Side by side with the Kiwibit Beako, the Bird Feeder 2 Pro is almost a dead ringer. The one key difference is up on top, with the solar panels now built into the roof. It’s a smart move that cuts down on clutter, as you don’t need to find somewhere else to mount a panel or route a cable.

The same roof also hides one of the feeder’s best design touches. Flip it up (it just requires you to press a button to release a latch) and you can lift the seed hopper straight out to clean or refill it. That hopper has two compartments, so you can offer two different seed mixes if you like, and each one has its own sliding door for shutting off the supply.

The body is almost entirely plastic, which won’t appeal to everyone. There are wooden and metal bird feeder cameras out there, and personally I’d prefer either material on something that lives in my garden. I also wonder whether the green plastic will fade in the sun over time, although I didn’t notice any change in my few weeks of testing.

Future | Sam KieldsenFuture | Sam KieldsenFuture | Sam KieldsenFuture | Sam Kieldsen

On the plus side, this is solid, hard-wearing plastic, and the feeder feels sturdier than others I’ve reviewed — the hinged roof, for instance, opens and closes without a hint of rattle. It’s IP65 rated and Kiwibit says it’ll work in temperatures from -4F to 122F (-20C to 50C), so bad weather shouldn’t be a worry for the electronics inside.

My review sample came with three mounting options: a fabric strap for wrapping around a tree or lamppost, a pole mount for stick it on the top of poles up to one inch (25.4mm) in diameter, and a wall bracket. I used the bracket to fix the feeder to a trellis against the back wall of my courtyard garden, and it took no time at all. A positioning sticker in the box shows you where to drill two guide holes, then two screws (also supplied) hold the bracket in place. There are wall plugs in the box too, should you need them.

Future | Sam KieldsenFuture | Sam KieldsenFuture | Sam Kieldsen

The feeder itself attaches to the bracket with four small hex screws, and the same screws secure the perch. An Allen key is included in the box, but make sure you keep it somewhere safe or have a spare, as you’ll be in a pickle without one. My sample also came with a water beaker that clips on to the perch.

Power comes from a rechargeable, removable battery that’s topped up by the solar roof panels, and the combination works very well. I fully charged the battery via USB-C before installation, as the manual instructs, and haven’t charged it manually since — weeks later, it’s still sitting at 99% capacity. To be fair, my testing took place in a sunny late summer and early autumn, and I positioned the feeder to catch a few hours of direct sun each day. I’m curious to see how it copes with the gloomier winter months here in southeast England.

Kiwibit Bird Feeder 2 Pro: performance

• 4K video capped at 15fps
• AI species ID requires a Kiwibit Plus subscription
• Phase-detection autofocus with a 130-degree view

Some bird feeder cameras I’ve tested have struggled to stay connected to my home Wi-Fi, but the Bird Feeder 2 Pro’s dual-band connection was stable throughout, even though the signal was rarely at full strength. The live feed kicked in from my phone without much delay, whenever I requested it.

Everything runs through Kiwibit’s app, and frankly it isn’t very well designed. There are too many sections with vague names, I regularly forgot where things lived, and getting to my photos and videos took more taps than it should. A simpler, pared-back layout would make a big difference.

There are some nice touches, mind you. Each species gets a written description, the Nature Expert section answers your bird questions, and the Collection page gives you a quick rundown of every species your camera has spotted. The catch is that many of the best features — especially the AI species identification, which feels like a core reason to buy this feeder — sit behind a Kiwibit Plus subscription. Kiwibit gave me a code for a year’s access, but for everyone else it’s an extra $89.99 on top of an already premium price, and the feeder feels noticeably less capable without it.

Future | Sam KieldsenFuture | Sam KieldsenFuture | Sam KieldsenFuture | Sam KieldsenFuture | Sam KieldsenFuture | Sam Kieldsen

The identification itself was largely accurate. Only five species visited during my testing period, and the feeder correctly named four of them. That one miss was interesting, though: in a clip featuring a robin, a sparrow and a blue tit, only the robin was identified. That was the only clip containing a blue tit, so it appears the system tags just the first species it spots in each video. I’ve no doubt it would have recognized the blue tit on its own, but it’s a shame it can’t log multiple species in one clip.

Image quality is a more complicated story. 4K videos and photos generally look good, with plenty of sharpness and clarity, and the autofocus is accurate. There's even a microphone to capture sound, and it's nice to hear the birds chomping away on seeds or tweeting at each other. But 4K video is capped at just 15fps, which I found baffling. It makes clips look jerky, and it put me off using the 4K mode altogether. Why couldn't it just be 24fps or 30fps?

Future | Sam KieldsenFuture | Sam KieldsenFuture | Sam KieldsenFuture | Sam KieldsenFuture | Sam KieldsenFuture | Sam KieldsenFuture | Sam KieldsenFuture | Sam Kieldsen

Shooting in 4K does have one perk: the software may automatically pull a 12MP Ultra HD “Keyshot” still from your footage. These are sharpened and upscaled for extra detail, but I found them a little over-sharpened, and they’re oddly large files at around 10MB each, given the modest resolution. You can’t pick the frame yourself, either — the software chooses for you, and it doesn’t always choose well. There’s a second type of auto photo, called SwiftShot, which is generated in Ultra Slow Motion mode and is meant to freeze birds in flight. That never happened for me; every SwiftShot I got showed a bird sitting on the feeder, and in noticeable less detail than with the Ultra HD Keyshots.

Video clips are 10 seconds long by default, but extendable to three minutes. I’d have liked to catch birds flying in more often, but my videos generally started with the visitor already perched.

There are two slow-motion modes, both 1080p: 60fps and 120fps. The 120fps mode doesn’t appear to work properly at the time of writing. Whatever settings I tried, the footage seemed to be captured at 15fps, so it plays back like a slideshow when slowed down. I was expecting something akin to the iPhone's buttery-smooth slow-motion mode, but this is in the single digits frames-per-second wise. I’ve reported this to Kiwibit and I hope a firmware update fixes it, because for now it’s useless as a video mode. It may be intended mainly for grabbing crisp stills of birds in motion, but since you can’t choose the frame yourself and have to rely on Kiwibit’s cloud processing to pick it, that’s not much consolation.

The 60fps mode is far better. Played back at half speed you get smooth 30fps slow motion, or at quarter speed you get the same 15fps as the 4K clips. It lacks the fine detail of 4K, but its smoothness at normal and half speed makes it the sweet spot of the lineup for me. One quirk to note: neither slow-motion mode records sound, even when you play clips back at normal speed.

Autofocus is generally reliable, but birds that came right into the seed tray, rather than staying on the perch, were sometimes out of focus. They seem to be inside the minimum focus distance, which was also an issue with the Beako. It’s not a huge deal, but it’s annoying when a bird’s body is sharp and its eyes and beak are soft.

Sharing and saving photos and videos is easy – you just hit the share button when viewing in the app and you get the option to download the file to your phone, or share via a variety of means. Subscribers also get cloud storage for all their files.

Should you buy the Kiwibit Bird Feeder 2 Pro?Buy it if...

You want detailed 4K stills

Its 4K photos are sharp and clear, and the phase-detection autofocus is accurate.

You never want to charge a battery

The solar roof kept the battery near full for weeks without me plugging it in once.

You want easy upkeep

The removable two-chamber hopper makes refilling and cleaning quick and lets you offer two seed mixes.View Deal

Don't buy it if...

You want smooth video

4K is capped at 15fps, and the 120fps slow-motion mode currently plays back like a slideshow.

You don’t want a subscription

AI species identification, a key selling point, is locked behind Kiwibit Plus.

You’d prefer a natural-looking feeder

It’s almost entirely plastic, so look elsewhere if you’d rather have wood or metal in your garden.

Kiwibit Bird Feeder 2 Pro: also consider

Bird Buddy 2

Priced at around $299 / £300, the Bird Buddy 2 also has a solar roof, and its app is far more polished and enjoyable to use, with free AI species ID. Its standard video is only at 1080p resolution, though, with 2K reserved for subscribers.

Read our Bird Buddy 2 review

Camouflage EZ BirdFeed

If you’re in Europe and value matters more than image quality, this £139 feeder is hard to beat. It offers AI bird ID with no subscription and twin roof-mounted solar panels, but its 1080p images can’t match the Kiwibit’s crisp detail.

Read our Camouflage EZ BirdFeed review

How I tested the Kiwibit Bird Feeder 2 Pro

(Image credit: Future | Sam Kieldsen)

• Mounted on a garden trellis for several weeks in late summer/early fall
• Tested all video modes, including both slow-motion settings
• Used the app with a Kiwibit Plus subscription

I installed the Kiwibit Bird Feeder 2 Pro in my garden in southeast England using the supplied wall bracket, fixing it to a trellis in a spot that got a few hours of direct sunlight a day. After fully charging the battery via USB, I left it to run on solar power for the rest of the testing period, which ran through late summer into early autumn.

I tested the feeder with the Kiwibit app on my phone, using a year’s Kiwibit Plus subscription supplied by Kiwibit to access the AI species ID and other premium features. I tried every video mode, including 4K and both 1080p slow-motion settings, compared the automatically generated KeyShot and SwiftShot stills, and checked the AI identifications against the birds I could see in each clip.

• First reviewed October 2026

• Britain's nuclear submarine crews are training to fight drones using water
• Fire hoses provide emergency protection when drones breach naval security
• Each Vanguard submarine costs approximately £4 billion despite its basic defences

British sailors are receiving training to repel hostile drones using powerful water jets aboard nuclear submarines stationed at a naval installation in Scotland.

The unconventional approach provides a relatively inexpensive and environmentally friendly defensive option for vessels facing potential aerial attacks while docked or operating above water.

A report from The Sun, the equipment serves as an emergency measure when incoming drones penetrate existing security arrangements around Britain's nuclear deterrent.

High-pressure water jets provide final protection

The Royal Navy has equipped its Vanguard-class submarines with firefighting hoses capable of directing concentrated streams of water toward approaching unmanned aircraft.

These devices feature handheld nozzles with pistol-style controls, allowing sailors to aim pressurised water at airborne threats approaching their vessels.

During an attack, sailors would direct powerful water jets at small and medium-sized FPV drones, attempting to knock them down or divert their flight paths.

The hoses are deployed aboard submarines stationed at HMNB Clyde; the Scottish naval installation responsible for accommodating Britain's nuclear deterrent fleet.

Unlike conventional air defence systems, the hoses cannot detect or automatically track incoming drones.

A naval source acknowledged the limitations of relying on firefighting equipment against airborne threats.

But when in a dire situation, “you have to use what you’ve got,” said an unnamed Royal Navy source.

The Vanguard-class vessels reportedly cost approximately £4 billion each, making their reliance on water-based protection particularly striking.

These submarines carry Trident II D5 ballistic missiles equipped with nuclear warheads, forming the maritime component of Britain's strategic deterrence capability.

Although the hoses provide no conventional anti-aircraft weaponry, they offer crews an immediate means of responding when other defensive measures fail.

Their practical value will depend on whether sailors can deploy them quickly enough against threats that evade existing defences.

Naval base faces potential close-range drone attacks

This arrangement addresses particular vulnerabilities where submarines become more exposed to aerial surveillance and attacks during port operations.

HMNB Clyde occupies a location beside Gare Loch, a relatively confined Scottish waterway measuring approximately one kilometre across at certain points.

Nearby hills, accessible roads and surrounding terrain could allow hostile operators to launch drones without providing defenders sufficient warning.

Such geographical constraints create additional security challenges because relatively short flight distances can significantly reduce the time available for interception.

The Ministry of Defence says military installations rely on several protective measures, including physical barriers, security patrols, surveillance equipment and counter-drone systems.

Officials have also announced £470 million in additional expenditure over four years to improve security across Britain's military facilities and installations.

These security investments become more important as underwater sensors were discovered near routes used by British submarines departing their Scottish naval headquarters.

Those devices were suspected of having possible connections to Russian efforts to monitor submarine movements, although their precise origins remain unconfirmed.

• "Wooferpads" transmit pulses through your skin
• Coming to the US in November 2026
• $249 (about £187 / AU$357)

As much as I love hearing music on some of the best headphones, there's something missing from every mix: the physical sensation you get from a set of speakers pushing lots of air. Sonic Lamb has a solution. Its latest headphones model features unique drivers to deliver sound that you can actually feel.

The new Beyond headphones look like any other premium pair, but in addition to their dynamic drivers, they have what Sonic Lamb calls an Impulse Driver in each ear. That driver transmits low-frequency vibration as low as 2Hz through the headphones' "wooferpads" so that you can feel the vibration through your skin.

The vibrations are synchronized with the audio to follow not just their timing but the dynamics of the song. It sounds gimmicky, I know, but I promise you it isn't: we've tested the first generation of this technology and it really works.

We were very impressed with Sonic Lamb's first generation in 2022, bar a few niggles over the design. (Image credit: TechRadar)We're getting those good vibrations

We've seen lots of attempts to bring the physicality of bass to headphone listening, with various degrees of success (and various levels of practicality, if you recall Ol' Thumpy, which had its wearer "looking like a human dumbbell).

So far, Sonic Lamb's version is one of the most compelling: the first generation was subtle but effective and really did add an extra dimension to music. This newer, more advanced version offers 10 intensity levels and personalized profiles for different use cases: music, movies, gaming or just adjusting the audio to suit your preferences.

Even without the impulse drivers the Beyond headphones offer an impressive specification, although there's no active noise cancellation. The Beyond headphones have Bluetooth 5.4, LDAC, aptX Lossless, Bluetooth LE Audio, LC3, USB-C digital audio, 3.5 mm analogue audio and multipoint connectivity. There's also Beyond Link, a dongle that gives you low-latency connectivity for gaming, and you can expect up to 40 hours of battery life.

Other than the somewhat tacky "eargasm" marketing slogan, which Sonic Lamb is still using, our main gripe we had with the initial model was the design, which we said was "a little unrefined".

But the key features — sound, comfort, weight and padding — were all very good. As our audio editor Becky put it: "they recreate the experience of being at a live gig much better than I could've imagined."

The Sonic Lamb Beyond headphones go on sale in the US in November 2026 with a price tag of $249 (about £187 / AU$357).

• Biomemory has unveiled DNA Shield, designed to provide offline, airgapped, secure storage of critical digital assets in synthetic DNA
• The company foresees challenges to the resilience those assets (also known as “cold data”) and has hinted at the introduction of what it calls Molecular IT
• Biomemory is a co-chair of the DNA Data Storage Alliance

Commercial DNA data storage company Biomemory has unveiled DNA Shield, a purpose-built synthetic DNA repository service that ensures critical digital assets remain untouched by cyber criminals and emerging technologies like artificial intelligence.

So-called cold data comprises 60% of all data in the world, and refers to information that cannot be altered. This might be passwords and biometrics or it could be incontrovertible data, facts, finances, and other material that serves as a record.

DNA Shield stores data in DNA, using an engineered encoding scheme that ensures it is biosecure – it’s a physical molecule that stores information long-term without power. Pilot slots have opened, ahead of a full service expected in 2027.

A shield of DNA

Biomemory has launched DNA Shield off the back of its acquisition of Boston-based DNA storage pioneer Catalog Technologies in March 2026, and aims to provide organizations with DNA storage of data.

Data could be biometric, it could be cryptographic keys, or it could be accounts. It might even be intellectual property, blueprints, unique or proprietary knowledge. Encrypting the data into synthetic DNA and storing as a medium that a million times denser than magnetic storage, offline and airgapped, is arguably the most secure storage available, potentially with a 150-year lifespan.

"DNA Shield is the moment Biomemory moves from proving a science to delivering a commercial product," notes Laurent Guengant, CEO of Biomemory. "Organizations need a way to hold their most critical digital assets, cryptographic keys, core intellectual property, regulated records, and long-term institutional memory, somewhere no single vendor, cloud, or system can reach, and to prove it still works over time.”

The molecular IT platform

Availability of the DNA Shield service is currently limited to a few pilots, running from October 15 2026 through to mid 2027. Biomemory tells us that a full service will proceed after that point, and its launch makes them the first company to launch a molecular IT product.

Vincent Franceschini, Vice President of Product Management at Biomemory and co-chair of the DNA Data Storage Alliance, said, "DNA Shield is built on the same molecular platform we have spent years industrializing, but it is the first time we are asking a customer to rely on it operationally rather than as a research demonstration. Getting that right, end to end, from encoding to authorized recovery, is what makes us the first company bringing Molecular IT to market."

Organizations interested in DNA Shield will be paying a subscription that starts at €10,000/month, and early access applications are open directly via Biomemory.

Nostalgia in tech is all the rage right now. Whether it's film cameras making a comeback, CDs seeing a resurgence or flip phones dominating headlines — old tech has somehow become cool again.

I'm not immune to the nostalgia bug either but, rather than buy something new, I decided to set myself the task of giving a 2013 Apple iMac a new lease on life.

This particular 27-inch model had been gathering dust, but with its generous display and otherwise functional hardware, throwing it away seemed like a terrible waste. Especially when a new 24-inch M4-powered iMac would set me back upwards of AU$2,000. With a little research and DIY knowledge, I could get the vintage 27-inch model humming again for under AU$500 instead.

My ambitions were relatively modest — I wanted something capable of web browsing, a little light creative work and other basic daily tasks. I wasn't expecting it to compete with my M2 MacBook Air.

After a little research, I settled on two relatively straightforward upgrades: more RAM and a new 1TB solid-state drive (SSD). And for a total outlay of approximately AU$310, I was hopeful I could give this old Mac a fighting chance.

Two upgrades, one aging iMac

The first thing I needed to address was the storage. My iMac originally came with a mechanical hard drive, which is hardly a recipe for speedy performance in 2026.

Swapping an old hard drive for an SSD is often one of the most effective ways to breathe new life into an aging computer. With no moving parts and significantly faster data access, an SSD can improve everything from boot times to opening applications.

My original plan was to replace the internal drive entirely, but that presented a slight problem. Unlike some earlier iMac models, my late-2013 version has a display that's secured to the chassis with adhesive. Getting inside requires cutting through that adhesive, carefully lifting the display and disconnecting cables before you can even think about accessing the drive.

I'd never attempted anything remotely like this before, and the thought of potentially damaging a perfectly good 27-inch screen wasn't especially appealing. Fortunately, there was an alternative — an external SSD.

Rather than dismantling the computer, I could connect a TeamGroup T-Force Vulcan Z 1TB SSD via USB, install a later macOS onto it and use it as the startup drive. It's a considerably less intimidating approach, even if the iMac's aging USB 3.0 port meant I wouldn't be getting the full performance potential of a modern SSD.

The 1TB SSD, enclosure and DDR3 RAM I bought for the project. (Image credit: Future / Max Langridge)

The second upgrade — the RAM — was comparatively easier as Apple made upgrading the RAM on this generation of iMac remarkably straightforward. A small access panel on the rear of the computer provides access to four memory slots, meaning there's no need to dismantle anything to install additional modules.

My iMac had 8GB of RAM originally, and adding two 8GB Timetec DDR3 sticks brought the total to a rather generous 24GB. I don't claim to be a computer hardware expert, but even I knew that it would provide considerably more breathing room for multitasking than the original configuration.

I'll admit, I wouldn't have known where to start without a little help from ChatGPT. It guided me through choosing compatible components and, more importantly, the process of installing macOS Sonoma on a computer that doesn't officially support it.

Rather than spending hours trawling through forums, I could explain exactly what I was seeing on screen and get advice on what to try next. It wasn't entirely plain sailing, but having something to turn to whenever I hit a stumbling block made the whole process far less daunting. Although, as my colleague discovered when asking ChatGPT and Gemini to help build a gaming PC, AI assistance doesn't always guarantee a stress-free experience.

Upgrading the memory requires only removing an access panel on the back of the iMac. (Image credit: Future / Max Langridge)How much did it all cost?

I'm well aware of the current RAM crisis affecting the tech industry, so my immediate thought was that buying additional memory could make this project rather expensive.

Thankfully, it turns out nobody running an AI data centre is particularly interested in 13-year-old DDR3.

A quick search on Amazon revealed a compatible pair of 8GB RAM modules for around AU$70, while a 1TB SSD came in just under AU$230. That’s the components taken care of for AU$300. Throw in a Ugreen external SSD enclosure (I paid AU$17) and, factoring in some Amazon promotions at the time of ordering, my total spend was approximately AU$310.

That's considerably less than buying a brand-new iMac, of course, but it's still a reasonable amount of money to invest in a computer that's more than a decade old. And while I was quietly optimistic, I was also aware that throwing RAM and an SSD at an aging Intel processor wouldn't magically turn it into a modern Mac.

There was only one way to find out whether I'd made a sensible investment.

So, how does it perform?

My Late 2013 iMac now has 24GB of RAM and runs macOS Sonoma. (Image credit: Future / Max Langridge)

With the upgrades completed and macOS Sonoma installed, I finally had a functioning iMac sitting on my desk. But getting an old computer to boot is one thing. Making it genuinely useful in 2026 is another.

My first test was to see how long it took to start up. From pressing the power button to reaching the desktop, I recorded a time of 1 minute and 52 seconds.

I'll admit, I was hoping for better. I'm accustomed to the near-instant responsiveness of my M2 MacBook Air, so waiting almost two minutes for a computer to boot feels like a small eternity.

There are some mitigating factors, of course. This is still a 2013 machine powered by a 3.2GHz quad-core Intel Core i5 processor, and it's booting macOS from an external SSD over USB 3.0 rather than a faster internal connection.

It's also worth pointing out that I didn't record performance benchmarks before upgrading the hardware, so I can't offer any definitive figures showing how much faster the machine has become.

Fortunately, things improved once I actually started using it. Basic web browsing proved perfectly fine, with pages loading successfully and the overall experience remaining usable, albeit with the occasional lag. I was particularly interested to see how it would handle Canva as I occasionally use the browser-based design platform for light creative work.

macOS Sonoma is installed on an external 1TB SSD connected via USB. (Image credit: Future / Max Langridge)

And, to my surprise, it worked. I opened Canva on one half of the 27-inch display, with the TechRadar YouTube channel running in another browser window alongside it. I was able to work on a simple design while keeping the website open, although there was some noticeable sluggishness.

It's certainly not an experience I'd choose over a modern Mac, but it demonstrates that this old machine still has something to offer.

The display is another highlight. With its 2560 x 1440 resolution, the 27-inch panel still provides plenty of screen real estate, and being able to arrange multiple windows side by side is something I particularly appreciate.

Sure, it doesn't have the sharpness or brightness of Apple's newer displays, but that doesn't automatically make it unpleasant to use.

And there's something rather satisfying about seeing an old computer handling tasks that, just a short while ago, I wasn't entirely convinced it would be capable of.

There are still some big compromises

Of course, there's only so much new storage and additional memory is able to achieve. The processor and graphics hardware are still more than a decade old, and while upgrading them isn't entirely impossible, doing so would require a far more involved disassembly process. Given the relatively modest tasks I had in mind, it was a level of effort and expense I wasn't prepared to entertain.

Then there's the issue of software compatibility. My late-2013 iMac officially supports macOS Catalina at most, but I've managed to get macOS Sonoma running through an unofficial installation method.

While this gives me access to a much newer operating system, it's not an officially supported configuration, and that brings potential complications around security updates, drivers and long-term reliability.

Canva and TechRadar running side by side — although the experience wasn't entirely lag-free. (Image credit: Future / Max Langridge)Was it worth the money?

So, would I recommend spending a few hundred dollars upgrading a 13-year-old iMac?

Well, that depends. If you're hoping to turn an old Intel Mac into something that feels as quick and responsive as a modern Apple silicon machine, you're probably going to be disappointed, unless you’re working with a relatively newer model than my machine.

My nearly two-minute boot time alone is evidence that upgrading the storage and RAM hasn't miraculously solved every performance issue. And while web browsing and Canva are perfectly usable, there's enough occasional lag to remind me that this isn't a computer I'd want to rely on for demanding work.

But that was never really the point of the exercise. I already have an M2 MacBook Air that I turn to for my everyday computing needs. What I wanted was to see whether I could rescue an old iMac and give it a useful second life, rather than let it become a part of our growing e-waste problem.

And on that front, I'd consider the experiment a success. I now have a working 27-inch desktop that can browse the web, handle simple creative tasks and provide a decent amount of screen space, all without having to buy an entirely new computer.

Honestly, if you've got an old computer gathering dust, it might be worth investigating whether it still has some life left in it. Just don't expect it to be as quick as you remember.

• AI now handles website creation for 53% of surveyed entrepreneurs
• Business owners saved $850 and 32 hours using AI tools
• Website design job postings have fallen 59% since early 2024

Freelance website development opportunities have fallen sharply since 2024, while an increasing number of American entrepreneurs are using artificial intelligence to create websites independently.

Research commissioned by Bluehost found that 53% of surveyed business owners had adopted AI website builders rather than relying on professional developers.

The findings coincide with substantial declines in freelance programming advertisements, particularly in website design, PHP, WordPress, and HTML development.

Website development vacancies fall by nearly 60%

The Bluehost survey questioned 201 American entrepreneurs involved in decisions concerning their companies' online operations.

Among respondents adopting AI website generators, 85% indicated that their websites would have been developed differently without access to these technologies.

Participants reported average savings of $850 and approximately 32 working hours compared with commissioning developers or agencies to complete equivalent projects.

However, website hosting and domain setup remained tasks that respondents were reluctant to delegate to AI, with 21% identifying these among the least likely to be outsourced.

Previously, 34% identified insufficient time as an obstacle to launching websites, while 28% cited limitations in their technical knowledge.

Even among respondents who had avoided AI website generators, 71% expressed willingness to consider these technologies for future website development projects.

Historical recruitment figures on Freelancer.com indicate that website design advertisements declined 59% between the second quarter of 2024 and the third quarter of 2026.

PHP opportunities experienced an identical 59% reduction, while WordPress assignments decreased 58%, followed by HTML programming vacancies with a 55% decline.

These reductions exceeded those recorded across certain other freelance categories, suggesting that website development assignments have experienced particularly substantial pressure.

Graphic design advertisements decreased 43% during the corresponding period, whereas data entry opportunities increased 16%, despite their exposure to automation technologies.

However, these figures cover one freelance marketplace and cannot confirm that the adoption of AI tools directly caused the reduction in advertised programming jobs.

Business owners increasingly manage websites without developers

Business owners reported considerable satisfaction with AI websites, with 72% expressing strong approval and 97% reporting at least moderate satisfaction.

Nevertheless, just 9% published their websites without modifying the initial AI output, indicating continued involvement in reviewing and adjusting generated material.

Content revisions were undertaken by 61% of respondents, while 60% modified visual elements before making their websites accessible to customers online.

Following publication, 94% continued employing AI technologies for business activities, extending their use beyond the initial website development process itself.

Among these respondents, 34% reported increased visitor numbers, while 31% experienced improvements in sales or revenue after adopting AI assistance.

However, 54% preferred reviewing every proposed website modification, compared with 30% who were comfortable allowing AI to operate independently without approval.

Only 50% expressed willingness to disclose their websites' AI origins to customers, indicating reservations despite generally favourable assessments of the technology.

The survey's relatively small sample also limits generalisation, with adoption estimates carrying approximately 6.9 percentage points of uncertainty at 95% confidence.

Although freelance advertisements have declined considerably, the evidence does not establish an equivalent collapse in employment across the entire website development industry.

Looking for a different day?

A new Quordle puzzle appears at midnight each day for your time zone – which means that some people are always playing 'today's game' while others are playing 'yesterday's'. If you're looking for Friday's puzzle instead then click here: Quordle hints and answers for Friday, October 16 (game #1726).

Quordle was one of the original Wordle alternatives and is still going strong now more than 1,500 games later. It offers a genuine challenge, though, so read on if you need some Quordle hints today — or scroll down further for the answers.

Enjoy playing word games? You can also check out my NYT Connections today and NYT Strands today pages for hints and answers for those puzzles, while Marc's Wordle today column covers the original viral word game.

SPOILER WARNING: Information about Quordle today is below, so don't read on if you don't want to know the answers.

Quordle today (game #1727) — hint #1 — VowelsHow many different vowels are in Quordle today?

• The number of different vowels in Quordle today is 4*.

* Note that by vowel we mean the five standard vowels (A, E, I, O, U), not Y (which is sometimes counted as a vowel too).

Quordle today (game #1727) — hint #2 — repeated lettersDo any of today's Quordle answers contain repeated letters?

• The number of Quordle answers containing a repeated letter today is 2.

Quordle today (game #1727) — hint #3 — uncommon lettersDo the letters Q, Z, X or J appear in Quordle today?

• Yes. One of Q, Z, X or J appears among today's Quordle answers.

Quordle today (game #1727) — hint #4 — starting letters (1)Do any of today's Quordle puzzles start with the same letter?

• The number of today's Quordle answers starting with the same letter is 0.

If you just want to know the answers at this stage, simply scroll down. If you're not ready yet then here's one more clue to make things a lot easier:

Quordle today (game #1727) — hint #5 — starting letters (2)What letters do today's Quordle answers start with?

• Q

• E

• S

• B

Right, the answers are below, so DO NOT SCROLL ANY FURTHER IF YOU DON'T WANT TO SEE THEM.

Quordle today (game #1727) — the answers

(Image credit: Merriam-Webster)

The answers to today's Quordle, game #1727, are…

• QUARK
• ENEMY
• SPASM
• BLOCK

I hope I wasn’t the only person who found today’s game difficult.

The rare letter Q in QUARK — and the relative obscurity of it as a word itself — and the odd S-M letter combination in SPASM made it quite a challenge and one I only just scraped through to complete my week-long win streak.

Daily Sequence today (game #1727) — the answers

(Image credit: Merriam-Webster)

The answers to today's Quordle Daily Sequence, game #1727, are…

• FREER
• SCALP
• CRATE
• TOWER

Quordle answers: The past 20

• Quordle #1726, Friday, 16 October: SHORT, WORTH, JOIST, SLUNK
• Quordle #1725, Thursday, 15 October: STALE, EPOCH, PIANO, CORNY
• Quordle #1724, Wednesday, 14 October: TIMID, GROUP, GRILL, EVERY
• Quordle #1723, Tuesday, 13 October: FLUFF, DEBUT, TEETH, STEIN
• Quordle #1722, Monday, 12 October: HELLO, BELOW, RIGHT, SLANT
• Quordle #1721, Sunday, 11 October: LILAC, INLET, TRIAD, SHAPE
• Quordle #1720, Saturday, 10 October: SWEAT, THIRD, RURAL, CLASP
• Quordle #1719, Friday, 9 October: OLDEN, DUNCE, SWATH, HILLY
• Quordle #1718, Thursday, 8 October: MINTY, ROWER, BROOD, RESET
• Quordle #1717, Wednesday, 7 October: PALER, FROTH, DOZEN, ERASE
• Quordle #1716, Tuesday, 6 October: FRESH, SKULK, BATCH, ENSUE
• Quordle #1715, Monday, 5 October: SIXTH, SPEAK, SEDAN, HEIST
• Quordle #1714, Sunday, 4 October: INFER, RAINY, THING, ARENA
• Quordle #1713, Saturday, 3 October: CHORE, FLYER, STICK, PURSE
• Quordle #1712, Friday, 2 October: IVORY, SADLY, ANGEL, LLAMA
• Quordle #1711, Thursday, 1 October: NAVEL, SHEER, BIRTH, TACKY
• Quordle #1710, Wednesday, 30 September: CHOKE, BATCH, MOSSY, EAGER
• Quordle #1709, Tuesday, 29 September: BILGE, MEDAL, LODGE, PORCH
• Quordle #1708, Monday, 28 September: BRUNT, TORSO, STONE, HEATH
• Quordle #1707, Sunday, 27 September: VALET, CHEST, VIGIL, PUREE

Looking for a different day?

A new NYT Connections puzzle appears at midnight each day for your time zone – which means that some people are always playing 'today's game' while others are playing 'yesterday's'. If you're looking for Friday's puzzle instead then click here: NYT Connections hints and answers for Friday, October 16 (game #1223).

Good morning! Let's play Connections, the NYT's clever word game that challenges you to group answers in various categories. It can be tough, so read on if you need Connections hints.

What should you do once you've finished? Why, play some more word games of course. I've also got daily Strands hints and answers and Quordle hints and answers articles if you need help for those too, while Marc's Wordle today page covers the original viral word game.

SPOILER WARNING: Information about NYT Connections today is below, so don't read on if you don't want to know the answers.

NYT Connections today (game #1224) - today's words

(Image credit: New York Times)

Today's NYT Connections words are…

• BREA
• KING
• SNAP
• PING
• STILL
• PAZ
• SNOW
• SHOT
• BOHÈME
• FIDDLER
• ALERT
• PRINT
• CHIME
• SCALA
• HERMIT
• DING

NYT Connections today (game #1224) - hint #1 - group hints

What are some clues for today's NYT Connections groups?

• YELLOW: Your phone wants your attention
• GREEN: Images
• BLUE: Types of marine beastie
• PURPLE: Missing a word that means “the”

Need more clues?

We're firmly in spoiler territory now, but read on if you want to know what the four theme answers are for today's NYT Connections puzzles…

NYT Connections today (game #1224) - hint #2 - group answers

What are the answers for today's NYT Connections groups?

• YELLOW: DIGITAL NOTIFICATIONS
• GREEN: PHOTO
• BLUE: KINDS OF CRABS
• PURPLE: LA_____

Right, the answers are below, so DO NOT SCROLL ANY FURTHER IF YOU DON'T WANT TO SEE THEM.

NYT Connections today (game #1224) - the answers

(Image credit: New York Times)

The answers to today's Connections, game #1224, are…

• YELLOW: DIGITAL NOTIFICATIONS: ALERT, CHIME, DING, PING
• GREEN: PHOTO: PRINT, SHOT, SNAP, STILL
• BLUE: KINDS OF CRABS: FIDDLER, HERMIT, KING, SNOW
• PURPLE: LA_____: BOHÈME, BREA, PAZ, SCALA

• My rating: Easy
• My score: Perfect

My initial thoughts were that there were a whole bunch of tiles about noises off and at least one word that you would associate with the word LA.

I decided to go with my second hunch and was rewarded with a Purple First for connecting BREA, PAZ, BOHÈME, and SCALA. Living in Europe possibly helped.

Having scored a success with this quartet and deciding to ignore the feeling I was falling into a trap, I submitted CHIME, PING, ALERT and DING. Next, now reduced to eight tiles, the four synonyms for PHOTO were easier to spot — a good thing, as my crab knowledge didn’t extend beyond HERMIT.

Yesterday's NYT Connections answers (Friday, October 16, 2026, game #1223)

• YELLOW: SURFACE IMPERFECTIONS: CHIP, MARK, NICK, SCRATCH
• GREEN: LANDLOCKED COUNTRIES: CHAD, HUNGARY, LAOS, PARAGUAY
• BLUE: EUROPEAN COUNTRIES MINUS A LETTER: BULGARI, CYRUS, HUNGRY, SPAN
• PURPLE: STARTING WITH WORDS FOR "STYLISH": CHICKEN, COOLIDGE, HIPPO, SMARTIES

What is NYT Connections?

NYT Connections is one of several increasingly popular word games made by the New York Times. It challenges you to find groups of four items that share something in common, and each group has a different difficulty level: green is easy, yellow a little harder, blue often quite tough and purple usually very difficult.

On the plus side, you don't technically need to solve the final one, as you'll be able to answer that one by a process of elimination. What's more, you can make up to four mistakes, which gives you a little bit of breathing room.

It's a little more involved than something like Wordle, however, and there are plenty of opportunities for the game to trip you up with tricks. For instance, watch out for homophones and other word games that could disguise the answers.

It's playable for free via the NYT Games site on desktop or mobile.

Looking for a different day?

A new NYT Strands puzzle appears at midnight each day for your time zone – which means that some people are always playing 'today's game' while others are playing 'yesterday's'. If you're looking for Friday's puzzle instead then click here: NYT Strands hints and answers for Friday, October 16 (game #957).

Strands is the NYT's latest word game after the likes of Wordle, Spelling Bee and Connections – and it's great fun. It can be difficult, though, so read on for my Strands hints.

Want more word-based fun? Then check out my NYT Connections today and Quordle today pages for hints and answers for those games, and Marc's Wordle today page for the original viral word game.

SPOILER WARNING: Information about NYT Strands today is below, so don't read on if you don't want to know the answers.

NYT Strands today (game #958) - hint #1 - today's themeWhat is the theme of today's NYT Strands?

• Today's NYT Strands theme is… Quite the crush

NYT Strands today (game #958) - hint #2 - clue words

Play any of these words to unlock the in-game hints system.

• MUST
• FLUSH
• MINTS
• FIVE
• SLOG
• LEAF

NYT Strands today (game #958) - hint #3 - spangram lettersHow many letters are in today's spangram?

• Spangram has 10 letters

NYT Strands today (game #958) - hint #4 - spangram positionWhat are two sides of the board that today's spangram touches?

• First side: left, 1st row

• Last side: right, 6th row

Right, the answers are below, so DO NOT SCROLL ANY FURTHER IF YOU DON'T WANT TO SEE THEM.

NYT Strands today (game #958) - the answers

(Image credit: New York Times)

The answers to today's Strands, game #958, are…

• BLUSH
• PERSPIRE
• BABBLE
• FIDGET
• GIGGLE
• STUMBLE
• SPANGRAM: FALLINLOVE

• My rating: Hard
• My score: 1 hint

Annoyingly, even though I suspected that today’s theme was about the squirming embarrassment of falling in love, I struggled to find any game words.

Even more annoying, having found both “plush” and “lush”, the hint gave me BLUSH.

I was obviously having an off day, and patently unattuned to the language of love, but I managed to STUMBLE my way around the board without the need of further assistance.

Yesterday's NYT Strands answers (Friday, October 16, game #957)

• PINPOINT
• MOLECULE
• ATOM
• MITE
• FOLLICLE
• AMOEBA
• SPANGRAM: SMALLWORLD

What is NYT Strands?

Strands is the NYT's not-so-new-any-more word game, following Wordle and Connections. It's now a fully fledged member of the NYT's games stable that has been running for a year and which can be played on the NYT Games site on desktop or mobile.

I've got a full guide to how to play NYT Strands, complete with tips for solving it, so check that out if you're struggling to beat it each day.

• Hebrew University physicists estimate a single superconducting vortex can act as a memory bit at about five times Landauer's limit, at 4.2 kelvin
• The fivefold figure is a simulation-based estimate anchored to continuous-drive measurements, not a measured erasure, however
• The release's million-fold edge over conventional computers disregards refrigeration costs

Every time a computer erases a bit, it pays an energy toll that physics will not waive; IBM physicist Rolf Landauer pinned down the minimum in 1961: kT ln 2, a figure that scales with temperature and works out to about 2.9 zeptojoules at room temperature.

No real chip comes close, making that gap a favorite target as computing's electricity bill keeps climbing.

Physicists at the Hebrew University of Jerusalem now say they have come within about five times of that floor using a single superconducting vortex, a nanoscale whirlpool of electrical current.

Storing a bit in a whirlpool might be a slightly controversial play

The work, published in Nature Communications on September 24 and publicized in a university release on October 11, is a clever measurement, but it has limitations: the device works only at liquid-helium temperatures, and one of the paper's three peer reviewers has opposed the final version.

In a thin superconducting film, a magnetic field can punch through in tiny tubes, each ringed by circulating current and carrying one quantum of flux, and these vortices snag on defects.

The team, led by Yonathan Anahory, used a 55-nanometer niobium film patterned into a strip 10 micrometers wide and cooled to 4.2 kelvin. Sweeping a steady current through the strip nudged individual vortices into spots where the film's natural disorder offers two neighboring resting places only a few nanometers apart. The positioning matters here, acting effectively as a store of value: left or right becomes zero or one.

They then rocked each vortex with an alternating current and tracked it with a SQUID-on-tip microscope, a magnetic sensor built onto a tapered glass probe. The part of the motion that lags behind the push reveals how much energy is lost each cycle.

Anahory said the result puts the bit at roughly five times Landauer's limit, “bringing it remarkably close to the minimum permitted by physics.” The release covers the energy as measured for the actual test, not the cooling aspect, which could significantly reduce the efficiency story here.

The paper adds another caveat here: Its abstract concedes that “our experimental method requires a steady sinusoidal drive incompatible with Landauer’s erasure protocol” and attributes the near-limit figure to numerical analysis.

The lab measured dissipation while vortices were wiggled back and forth continuously; the cost of a true erasure, forcing a bit of unknown value into a known state, was estimated with simulations anchored to that data. The fivefold number is the authors' estimate for a working bit driven at 150 kHz, spelled out in their replies to reviewers.

This has, however, spurred a peer review fight, also published by Nature Communications, which lists the unusually blunt referee reports alongside the paper.

Reviewer #2 rejected the first draft, said the second could run only with extra experiments, and recommended against the final version, opening with “Authors write that they measure work. It is not true.” The reviewer argued that force, displacement, and work were estimated or inferred rather than observed, and asked for two sanity checks: show the vortex being pushed into one well and staying there, and show its spontaneous hopping changing with temperature, as thermal activation predicts.

The authors declined both, citing scanning drift that limits static imaging to about 20 nanometers and noting that changing temperature also reshapes the pinning landscape.

Reviewer #3 backed publication, calling the work original and sound and warning that the extra experiments could delay it indefinitely. Reviewer #1 judged the technique ill-suited to erasure at useful speeds and encouraged a more modest title.

Speed is the tradeoff

Not only is the cost of cooling unaccounted for, with real-world refrigeration costs not being factored in, but the bigger problem is also speed. Landauer's limit is only reachable if you take your time, and work by Proesmans, Ehrich, and Bechhoefer shows that the extra cost of erasure rises as operations speed up.

The Hebrew University estimate applies at 150 kHz, or essentially one operation every 6.7 microseconds, while superconducting logic chips were clocked at 6 GHz more than a decade ago.

Going slower does not help either. By the authors' own noise data, the two wells central to the near-limit result let the vortex hop on its own roughly 3,000 to 7,000 times a second at 4.2 K.

The authors argue that this is fine as long as operations run much faster than the hopping. But it means the most efficient wells hold information for well under a millisecond, and the paper does not show a single well that is both near the limit and able to keep data for long.

The tradeoff, therefore, is one between speed, cost, and durability, and given that the use case is data storage, it is a tough call to make. Vortex memory is not entirely new; the Hebrew University contribution is energy accounting, and its claim to a first is narrow, though it uses a single vortex.

The Hebrew University team has produced a careful, somewhat contested estimate of how cheaply a single vortex can be shuffled between two spots, and a reminder that sitting close to the physical limit and being efficient at the wall socket are very different things.

• Withings users must now pay a hefty premium to sync data to Apple Health
• The change requires a $100 annual subscription to Withings+
• Customers are incredibly unhappy with the move

The whole point of using a health and fitness tracker, one would think, is to be able to analyze your data in order to understand your own well-being. New users of Withings’ smart scales, however, have just been stung with an extortionate annual subscription if they want to enjoy that privilege.

As noted on the Withings forum, the company has decided to lock the ability to sync your wellness metrics to the Apple Health app behind a paywall. And this isn’t just some token fee either — you’ll be charged a whopping $99.99 a year. All to do what has always been free (and what many rivals still offer at no cost).

Withings apparently justified the move to the forum poster by citing unspecified “privacy measures” and an “AI connector.” Yet syncing data to Apple Health using the HealthKit framework doesn’t require any server-side processing or hefty AI lifting that might justify such an expensive fee. Instead, the paywall simply gives the impression of being an opportunistic revenue booster.

The post on the Withings forum described the move as a “silent and incredibly anti-consumer change” and pointed out that anyone who signed up on or after October 13, 2026, would be subject to it. Existing users won’t be forced to pay, but that’ll come as little consolation to everyone else.

A controversial move

(Image credit: Getty Images)

It’s safe to say that the optics don’t look good for Withings, and users have been understandably upset at the move. The original poster on the Withings forum labeled the change “an obvious cash grab” that “ruins the hardware experience and destroys the trust in the brand,” adding that “We shouldn't have to pay a subscription to access our own health data.”

On Reddit, users echoed that dissatisfaction, saying “$99 a year is a real joke” and, about Withings’ decision-makers, “They are digging their own grave.” Another Redditor asked, “Why would I ever buy another Withings product, let alone recommend the products to other people, if you have to pay a subscription to access your own data?”

A Withings representative tried to explain that the $99.99 Withings+ subscription includes other features — including “An improved Health Assistant” and “Better-priced access to medical services” — but no user seems to care about any of that. The main concern is the decision to lock a fundamental feature behind a paywall when said feature appears to require no additional costs on Withings’ part.

Given the strength of the customer reaction already, it wouldn’t be surprising to see Withings backtrack and unlock Apple Health syncing in the near future. With many users already saying they’ll either ditch their Withings devices or abstain from buying one in the first place, the ball is firmly in Withings’ court.

AI has changed many aspects of the professional world, especially the practice of software development and engineering. A swathe of tools and platforms now enable programmers to use AI assistants and copilots – as well as swarms of agents – to work on building systems autonomously, based on instructions. And for one key figure in the AI space, this was something of an inevitability.

Going viral

In a viral post on the social media platform X, one of the key figures behind the conception of OpenAI, Andrej Karpathy, remarked on the future of programming, provoking a reaction from many in the industry at the time.

Quote of the day

This article is part of TechRadar Pro's QOTD project to provide an insight into the minds of the brightest and most recognized figures in the technology industry today and in years gone by. Read the full series here.

ChatGPT had launched a few months prior to that, and developers were already familiar with the basic level of coding that the chatbot could perform, especially in popular languages such as Python and HTML. Much of this was also based on simple text-based instructions – meaning that the AI models could understand and process natural language and churn out complex programming in a way that hadn't been seen before.

Karpathy, who in 2024 co-founded Eureka Labs, an AI education platform, perfectly encapsulated the feeling, with the tech industry in the process of understanding the full implications of this technology.

Good vibes only

Karpathy added in a follow-up post that many examples out there in scientific literature showed that natural language was capable of wielding AI to engage in complex programming. These findings served as a collective precursor to many enterprise features that we see in widespread use today, such as GitHub Copilot, Claude Code, and others.

However, the rise of vibe coding and instructions-based programming also means that the nature of the work is changing for many software developers. Many now simply review code instead of writing it themselves, engaging in validation and bug-squashing.

Thus, the role of software development is becoming something that requires fewer and fewer juniors. This, however, may lead to other unintended consequences – including the fact that the traditional career pipeline is permanently changed.

L-shaped standing desks remain quite a niche pursuit, in my experience. They're ideal if you need additional desktop space. They're more than capable of lifting more than your standard desk, which is ideal for multi-monitor setups. But they also need a lot of room.

Still, for me, the Uplift V3 L-shaped standing desk is one of the best models around. It remained rock-solid in my real-world tests, with a massive 535lb lifting capacity thanks to the surprisingly quiet triple motors.

Given the absolute size of this unit, it's unlikely to be the best standing desk I've tested for most office professionals. But for those that need a stable desk with plenty of space to work on, there's a lot on offer here that I like.

Uplift V3 L-Shaped Standing Desk: Price and Availability

The Uplift V3 L-shaped desk is available direct from the manufacturer's website.

Prices start at $1,000, it comes in a range of styles, and you can expand it to make it a little fancier if you like and still get exactly what you want for a little cheaper.

What you choose, whether that's the barreled-shaped frame that goes all the way down to $1,719 if you just want the frame, or if you do end up choosing a desktop to go with it, you can absolutely make sure that you check Uplift's website to make sure you're getting the best deals.

All Uplift desks come with a standard 15-year warranty, which shows just how much they believe in the product. They also sell pretty much everything you need accessory-wise, and I've had great experiences with these, too.

(Image credit: Collin Probst // Future )Uplift V3 L-Shaped Standing Desk: Unboxing and First Impressions

Let this be a warning to you - learn from my mistakes. I should have followed the instructions I sent, built a standard desk, built the accessories piece, sanded it, and then attached the waterslide.

However, I got slightly distracted and rushed, so I ended up building the entire thing on the floor. I then had to flip it very, very, very carefully and nearly broke my foot, broke my toe, and ended up cutting off the top of the desk pretty significantly (if you look at the pictures) due to messing up a screw because I was once again rushing and trying to solve my issues.

I can't blame Uplift for this. The instructions were clear. I just did not listen. Thankfully, we also got a spectacular desktop on Uplift, so I can use it to cover the gouge from the screw I misplaced during assembly.

Once assembled, I was thrilled with how clean and sharp the desk looked. The power bar attached just as easily as the others. The little desk shelf added a nice elevation change for gear, and the desktop material itself was just as good as I hoped.

(Image credit: Collin Probst // Future )Uplift V3 L-Shaped Standing Desk: Design and Build Quality

L-shaped standing desks have been hit or miss for me. I have used some that were super sturdy and strong, and others that feel like they're not even worth it.

Uplift, on the other hand, does it right, and they make sure the added leg and extension actually add stability, not just complexity. At max height, which I seem to be putting it at often since I'm fairly tall and like to adjust cables a lot, this desk is still just as stable as when it's at its lowest.

I haven't had any issues adjusting the height over these 90+ days, and there have been zero responsiveness issues. The desk responds quickly when I press the buttons to move it, and it jumps directly to the designated preset height, just as I would expect.

Over the last several months, my only qualm has been the gash in the desktop that I made myself. Once again, once we moved the desk mat over it, we now barely notice that, and actually don't notice it at all.

(Image credit: Collin Probst // Future )Uplift V3 L-Shaped Standing Desk: In use

I've already gone into this a bit, but over the last 3 months or so, this desk has done excellently across my team.

We've changed the setup a few times, but the guy on my team who normally uses this desk prefers to keep it pretty simple. He uses a laptop, lots of physical books for research, and some photos. It never missed a beat.

(Image credit: Collin Probst // Future )

(Image credit: Collin Probst // Future )Uplift V3 L-Shaped Standing Desk: Final verdict

If you're looking for an L-shaped standing desk and you're willing to spend a little extra to get something quality, long-lasting, and customizable, then the Uplift V3 L-shaped standing desk is definitely one you should consider.

It's got a ton of options on both desktop and materials. It's got great accessories. It's easy to build if you follow the instructions, unlike me. Now, nearly 100 days in, I can confidently say it doesn't just look good at first; it actually runs well after I've pushed it hard for several months.

I have no doubt this desk will last for years with basic care and maintenance, and it will keep serving well for a long time.

Collin Probst // Future Collin Probst // Future Collin Probst // Future Collin Probst // Future Collin Probst // Future Collin Probst // Future Collin Probst // Future Collin Probst // Future Collin Probst // Future Collin Probst // Future Collin Probst // Future Collin Probst // Future Collin Probst // Future Collin Probst // Future

• LG's FLiPP tech could deliver Tandem OLED performance for much less
• Will arrive on smaller displays first, don't expect it on TVs for several years
• Production for laptop and monitor screens could start in 2027

If you're a fan of OLED, there's an exciting new acronym you need to know: FLiPP. It stands for FMM-Less innovative Pixel Patterning, and the FMM bit refers to fine metal masks. Those are normally used in OLED manufacturing to precisely place the organic emissive materials.

With FLiPP, there's no mask. Instead, the materials are deposited across the glass, given a protective layer and then refined using photolithography and etching to get rid of the unwanted sections.

The biggest single advantage to FLiPP is efficiency. According to trade site The Elec, LG Display found that the tech increased the utilization rate – the efficiency of the production – from 75% in a conventional FMM process to over 93% in a FLiPP process.

That's good news for us, because more efficient production means lower production costs. And that ultimately means we could see much better OLEDs without higher prices.

FLiPP also improved utilization of the organic emissive materials from 40% to 74%, boosted the pixel aperture ratio (how much of the pixel area emits light) to 65%, and enabled the light-emitting area to boost luminance by 1.6x, longevity by 2.4x or reduce power consumption by 13%.

So that means these OLED screens can brighter, or they can be more reliable, or they can use less power, depending on the preference of the device. But in the near future, those devices will almost certainly be laptops and monitors, and not TVs — sadly for those desperate for the price of the best OLED TVs to drop.

LG Display announced the technology at the International Meeting on Information Display (IMID) 2026 (Image credit: LG Display)What the FLiPP?

LG Display's Chief Technology Officer (CTO) Choi Young-seok says that the technology is still in the first development stage. The next stage would be developing actual products, and the third stage is mass production.

That third stage will initially concentrate on the monitor market where displays are smaller — the improved efficiency figures mentioned above were for 23.5-inch panels — but laptops are apparently on the cards as well. LG Display has said it expects to reach this point by late 2027 — so the first products might appear at CES 2028.

According to Cho Heung-ryeol, who is the head of the Next-Generation Display Division, LG Display is currently evaluating its existing LCD and WOLED production lines to see if they can be reused without major modifications, or if new plant would need to be purchased. The former option is of course the preferred one. Either way, LG Display still plans for production to start in around a year.

If you'd like to know more about the technical side of FLiPP and how it compares to other OLED manufacturing processes, The Elec has published a transcript of LG Display's online seminar and you can read it here.

• AMD's proposed rack combines 32,768 CPU cores with 16 AI accelerators
• Traditional data centers may need new power and cooling infrastructure
• AMD divides agent workloads into personal, enterprise, and technical computing categories

AMD is developing a rack-scale computing system for artificial intelligence agents, with configurations designed for personal, enterprise, and technical workloads.

Dubbed Helion Agentic AI Rack, it supports up to 32,768 Zen 6 x86 CPU cores, 16 Instinct MI350P accelerators, and 74 petaflops of FP4 computing performance.

The proposed platform also supports 102 Tbps of networking bandwidth and uses modular hardware designed for different agent computing requirements.

A rack built around three kinds of AI agents

AMD groups agent workloads into three categories: personal assistants, enterprise applications, and technical computing, each with different processing requirements.

Personal agents need dense, energy-efficient servers; enterprise applications require a balance between agent capacity and performance; while technical workloads prioritize processing speed.

AMD allows operators to customize CPU core density, processing performance, and power consumption according to their workloads.

The hyperscale configuration uses CPU-only infrastructure, while the enterprise version combines processors and graphics accelerators within an integrated rack.

For personal agents, the system uses 256-core Venice processors operating at 400W, prioritizing processing density and power efficiency.

Enterprise workloads use the 256-core Venice processors rated at 600W, while technical computing uses 96-core, high-frequency Venice processors.

The enterprise configuration separates agent execution, inference, and general-purpose computing into distinct server nodes.

CPU sandbox nodes handle agent execution, Instinct MI350P accelerators support inference, and general-purpose servers accommodate databases, storage, and technical computing.

AMD also includes an integrated token router for directing agent model requests across frontier and open-source models.

The rack supports programmable networking, an open-source software ecosystem, and ORv3 44OU hardware compliant with Open Compute Project specifications.

Why traditional data centers may need substantial changes

Traditional data centers are built for standard cloud or legacy enterprise computing, not racks packing up to 32,768 Zen 6 CPU cores alongside heavy GPU configurations.

Therefore, facilities built around existing server arrangements may lack the appropriate combination of power delivery, cooling capacity and network infrastructure for these configurations.

Workload diversity creates another challenge because personal agents, enterprise applications and technical computing require different processing characteristics.

AMD's proposed rack mixes dense 256-core processors with 96-core high-frequency systems, requiring infrastructure that can accommodate different server types and operating requirements.

This creates a planning challenge for data centers, which must allocate rack space, power, and cooling to different server configurations instead of deploying identical systems.

Operators may need to modify existing infrastructure if it cannot support the selected combination of high-core-count and high-frequency servers.

Networking introduces an additional consideration because the rack supports up to 102 Tbps of bandwidth and integrates a programmable token router.

Existing facilities may need compatible networking equipment and software to support this throughput and route model requests across different AI systems.

The Helion Agentic AI Rack is scheduled for availability in November 2026, but AMD has not disclosed complete facility upgrade costs or cooling requirements.

• Oxylabs, Decodo, Webshare, ScrapingBee, and Honeygain now formally sit under one Vilnius-based umbrella called Source
• Source claims AI agents drove a fourfold increase in demand for services across its organizations
• The brands, previously reviewed as rivals, including three of TechRadar's nine best-proxy picks, now officially share common oversight

Oxylabs, Decodo, Webshare, ScrapingBee, and Honeygain have formally come together under a single name, Source, and the Lithuanian group behind them says AI agents are why it needed one.

The Vilnius-based ecosystem claims demand across its companies has risen fourfold in a year as AI agents, next-generation search tools, and frontier AI products pull live data from the web.

Its systems now process 1.5 million requests per second, and that it is expanding into web search APIs built specifically for AI agents.

One umbrella with five different, competing shopfronts?

Source presents itself as a backbone rather than a business. Its launch post says it has no customers or products of its own; instead, it handles patents, procurement, hiring, and compliance, while each brand keeps its own team, product strategy, and customer base.

The brands are split by customer type. Oxylabs serves enterprises and AI labs; Decodo (formerly Smartproxy) targets the mid-market; ScrapingBee sells developer APIs; Webshare serves professional consumers and solo builders; and Honeygain pays ordinary people for their spare bandwidth.

For anyone shopping for a proxy service, this is the most practical takeaway. Oxylabs bought Webshare in 2022 and ScrapingBee in 2025, and Honeygain became its main residential proxy sourcing partner in 2021.

Decodo's ties were less clear-cut until now, and trade site Proxyway notes that the launch officially links Oxylabs and Decodo under one organization. Source CEO Vytautas Savickas ran Decodo until March 2026, when he moved across to lead the Oxylabs group and founder Julius Černiauskas became chairman.

All of this points to a marriage of convenience rather than a merger by organizations that are already very closely interlinked. This is interesting because some of these brands have been sold and reviewed as alternatives to one another.

The best example of this is that three of the nine picks in TechRadar's own best proxy sites guide now belong to Source: Decodo as best overall, Oxylabs as best for business, and Webshare as best for privacy.

For context on the overlap, our residential proxy roundup described Oxylabs as one of Decodo's toughest competitors. Source says the companies remain fully independent in strategy, customers, and market presence, and that may well hold commercially, but common oversight is now a matter of record.

The release provided to TechRadar says Source gives the group "a unified voice in defending responsible access to public web data," and its legal history explains the appetite.

Reddit sued Oxylabs alongside Perplexity, SerpApi and AWMProxy in October 2025, alleging they evaded its anti-scraping measures by lifting Reddit content from Google search results. Oxylabs said it was "shocked and disappointed" and would defend itself.

It was still listed as a defendant in filings this spring, and in late July a federal judge allowed Reddit's core claims against SerpApi and Perplexity to proceed. Further back, a Texas jury awarded rival Bright Data nearly $7.5 million against Oxylabs in a 2021 patent case.

Source's own essay on AI agents and the open web argues that if access is negotiated one licensing deal at a time, only the biggest companies will be able to afford it. This makes a convenient argument for a large web scraping group, but the gatekeeping fears are real.

The most concrete product promised here is AI search, and it is older than the umbrella itself: Oxylabs already offers a Web API that pairs a search endpoint with a scraping endpoint for agents, so the "dedicated AI Search APIs" promised look more like an expansion of existing work than a new product line.

For customers, the questions to watch are whether the brands keep competing on price, how clearly each one discloses their newfound relationship, and how the Reddit case plays out for Oxylabs.

For everyone else, Source is a bet that AI agents will keep needing the live web, and that the companies supplying the pipes deserve a seat at the table when the rules for that access get written, aiming to negotiate from a clustered position of strength.