According to a Wall Street Journal report, Meta may be looking to some of Hollywood’s top brands to produce exclusive content for its next XR headset, which is expected to feature a completely new thin and light design when it reportedly ships next year.

Citing people familiar with the matter, Meta has recently been in talks with a number of entertainment brands, including Disney, A24, and smaller production companies to create both episodic and standalone immersive video tied to well-known IP.

Additionally, WSJ reports that talks include the possibility of timed exclusivity, allowing producers to later sell on other platforms after a specified period.

It’s said Meta hopes to use the videos to attract users to the company’s next XR headset, which is expected to compete with Apple Vision Pro when it launches next year.

SEE ALSO
'Thief Simulator' VR Sold Over 300,000 Copies on Quest Alone

Codenamed ‘Loma,’ the headset is said to feature a design similar to a pair of eyeglasses that connects to a pocketable compute puck, which is described as more powerful than its Quest 3 series of headsets. WSJ reports Meta is looking to price the device less than $1,000.

Provided the report is true, this would mark a sharp departure from the company’s current line of Quest headsets, which pack all components into a single standalone unit. Outside of Quest Pro, which was largely seen as a commercial failure, the company has also increasingly focused on sub-$650 hardware. Quest 3S, its most recent, is currently priced as low as $300.

Meta Quest 3S side profile | Image courtesy Meta

Speaking to WSJ, Meta says it develops multiple headset prototypes at all times—a non-committal answer if we’ve ever heard one. Whatever the case, shopping around for exclusive content deals suggests something substantial is coming down the line.

A separate report from UploadVR additionally claims Meta’s top Quest 4 contenders, codenamed ‘Pismo Low’ and ‘Pismo High’, have been canceled. Quest 4 was reportedly expected to land next year; rumors echoed by respected VR leaker Luna recently suggested Quest 4 is however now coming in 2027 in favor of the new design mentioned above.

Notably, Meta CTO and Reality Labs chief Andrew Bosworth said last December that wireless puck units for mixed reality headsets like Quest aren’t “a magic bullet,” suggesting the separate compute unit may be tethered to the headset in question.

Orion Puck Computer | Image courtesy Meta

“We have looked at this a bunch of times. Wireless compute pucks just really don’t solve the problem. If you’re wireless, they still have a battery on the headset, which is a major driver of weight. And, sure, you’re gaining some thermal space so your performance could potentially be better, although you’re somewhat limited now by bandwidth because you’re using a radio,” Bosworth said.

In the meantime, the XR landscape is invariably moving towards thin and light hardware of all types, encompassing everything from PC VR headsets like Bigscreen Beyond 2, to smart glasses that offer built-in heads-up displays, such as the upcoming Android XR-powered glasses from Google—set to be released by Warby Parker and Gentle Monster. Widely reported rumors of Meta’s next-gen smart glasses and Apple’s upcoming smart glasses also persist.

At least in the case of bulky XR headsets though, the hope is that removing weight will also reduce user friction, and drastically increase long-term engagement.

Newsletter graphic

This article may contain affiliate links. If you click an affiliate link and buy a product we may receive a small commission which helps support the publication. More information.

Scott Hayden
Well before the first modern XR products hit the market, Scott recognized the potential of the technology and set out to understand and document its growth. He has been professionally reporting on the space for nearly a decade as Editor at Road to VR, authoring more than 4,000 articles on the topic. Scott brings that seasoned insight to his reporting from major industry events across the globe.
  • Storymode Chronicles

    Best design would be a wireless puck with its own integrated battery, and a separate swappable battery either on the headset strap or a dockable tether that could also charge the puck if connected.

    Still think the real reason they're so afraid of a wireless capable puck design is it then becomes possible to use the puck as an all purpose compute unit powering multiple other devices, threatening multiple hardware partner markets.

    • Interesting. I hadn't thought of that…

      • Storymode Chronicles

        Yeah, a "puck" capable of running a next gen headset should be easily capable of acting as an all-purpose computing device that could just as well power a laptop, smartphone, tablet, AR glasses, smart TV, gaming handheld etc. If it was small enough it could even just insert inside devices such as a smartphone "shell" to bring them to life.

        I had actually hoped the Switch 2 would be something like this, not just to make a real Nintendo VR headset possible after the Virtual Boy debacle and Labo VR concept, but so they could release different "shells" like a GBA or mini Switch for different use-cases. Sure, why not a Nintendo phone too. In my headcanon they called it the Super Switch lol.

        The US military and companies like Boeing and Raytheon already use devices like this stemming from on an old 2009 patent US7516484 from ADDC. So, part of it could also be prohibitive licensing, but that's expired in 2029 anyway around the time this form factor should really feasibly reach maturing anyway, and I feel like Meta, Apple, Microsoft, etc. are more than big enough to jump that hurdle now.

        I think the more immediate reason they're opposed to it is definitely the fact that right now they can charge full price for each of these devices with their own discrete CPU/GPU/memory, instead of just charging for one "puck" that you can easily upgrade from without replacing all your existing devices and transfer to different headsets and other devices that can follow a separate upgrade path.

        For the record @arnovanwingerde:disqus this is also much different than a desktop PC, but I get where you're coming from. It's closer to having a gaming laptop if anything, but one that's smaller than a smartphone. The one's produced under patent US7516484 are actually closer to credit card sized, a PCMCIA card chassis to be exact, designed to be small enough to insert into a smartphone indeed, or for that matter into a wireless hub paired with a VR headset.

        I also had an interesting conversation with @christianschildwaechter:disqus about this a few months ago the last time Meta loudly proclaimed a wireless puck is not the answer, but didn't go back and grab the patent number at the time. He's a bit more skeptical than me if you're interested in the pitfalls.

        • Christian Schildwaechter

          What you hoped for with the Switch 2 sounds similar to the Pimax Portal from early 2024, a 5.5" 4K 144Hz Android tablet with magnetically attached Switch-like controllers, HDMI-in, built-in tracking camera and a headstrap that looked like a mix of Daydream with an Elite strap with extra swappable battery in the back that you'd simply slid the Portal into to use it as a VR HMD.

          The slim controllers could be plugged into special Pimax VR controllers. It by allowed streaming flat games using the controllers or the touch screen via Steam Link, PS remote play or services like Geforce Now or Amazon's Luna, could play Android games and emulate old consoles, all using the same XR2 Gen 1 as the Quest 2. It could connect extra external tracking cameras for both VR and Android games. And as it had USB-C and could run all Android apps, you could have used it as a laptop/desktop replacement, and connect mouse/keyboard and an external display. Or use it as a separate display with the HDMI-in port. It basically tried to do everything at the same time for USD 200-400 depending on model.

          pimax_com/collections/pimax-portals

          It is still listed on the Pimax website, but all models are "temporarily out of stock". They actually sold those, though the headstrap seems to have been dropped quickly, and the reviews weren't great. The hardware itself was mostly solid, but somewhat outdated, ran rather hot and there was basically no VR software available, and no support for PCVR streaming. Still an interesting concept, but with a very lacking, unfinished implementation.

          • Storymode Chronicles

            You know it’s funny, when I first heard of Pimax Portal when it was announced my heart actually skipped a beat. Unfortunately, its form factor is essentially just Gear VR. It’s missing most of the functionality described in patent US7516484.

            There’s really a quantum leap in usability when the compute resources are completely offloaded to something like a PCMCIA card that can insert into these devices to provide their primary compute resources and memory, especially when coupled with the ability to cast wirelessly to other devices at the same time.

            The use cases explode. You can insert it into your phone and cast to your AR glasses and smart watch, and then move it to a tablet or laptop to seamlessly continue your work in different form factors. If you were just headed out for the night and wanted to unplug, you could put it in a dumbphone shell. If you were headed to the gym, you could put it in a streamlined mini phone shell that’s comfortable in an arm strap.

            It completely opens up the space for form factor experimentation by offloading the main cost of these devices to a separate unit that can be shared among them. You’re removing a built-in cost of something like $500 – $1000 depending on specs from each of these devices, and allowing an easy upgrade path at the same time. Upgrade a single compute card and you upgrade every other device’s compute resources concurrently, for less than the traditional cost of upgrading any of them since you don’t need to upgrade all those displays, batteries, I/O and specialized chips like eGPU or environmental processing that are in the shell devices.

            So all of these form factors become cheap shell devices, centered around the compute card. Peripheral manufacturers take over a burgeoning form factor market. Testing new form factors becomes a low cost, low risk endeavor, allowing consumers to try out different device forms at impulse prices. It completely changes the market.

            Which, again, is also why I think Meta is actually against moving to a disconnected “puck”, because it basically begs the question: why does each device need its own discrete compute resources? You can only really use any one device at a time after all. So, why not allow that “puck” to cast or connect to a family of shell devices? Then you’re pointing back to patent US7516484.

          • Christian Schildwaechter

            TL;DR: the PCMCIA form factor doesn't allow for proper cooling and therefore only works for low power devices, so pretty much everything has switched to PCIe based connections like Thunderbolt instead, which allow for a lot of flexibility when plugging different peripherals together for the particular use-case dependent performance and size requirements.

            Using a single form factor that allows swapping compute unit may sound neat, but it comes with some severe limitations. If you are going for something small enough to work in a phone, so like PCMCIA, you run into space and especially thermal problems with anything moderately fast. Devices like the Quest 3 or Steam Deck with rather powerful SoC rely on heat pipes and fans to remove all the generated heat without constantly throttling.

            So you either have to pick a module size so large that it can incorporate at least passive cooling with heat pipes and more than a single RAM chip, but then end up with something like a typical mobile mainboard in a case of at least 120*60*15mm³, approaching compute puck size when a battery is added, certainly not usable in a phone or slim HMD. Or you keep it very small like PCMCIA at 86*54*5mm³, but can't go beyond the performance of a mid range phone from a few years ago, largely due to lacking cooling.

            PCMCIA first used the ISA-bus, then PCI, the successor ExpressCard USB-2/3 and PCIe 1/2, but both basically died because the fixed format was too restricting, and modern busses allowed to connect peripherals just as fast. Most similar systems have all been replaced by USB-C/Thunderbolt. Newer versions of these simply tunnel PCIe, and other interface types like NVMe or SCSI Express nowadays are also all just direct PCIe connections to the processor.

            So you can basically plug several compute components together via PCIe over USB-C, these can very in size and performance, and Thunderbolt 5 goes up to 80Gbit/s symmetric, 120Gbit/s symmetric over up to 1m cables, can drive several high resolution displays, carry up to 240W (for devices with a lot of cooling), and tunnel lots of protocols like DP, HDMI, SCSI, ethernet, USB and more. It basically does everything via a single, standardized connection.

            A couple of devices connected by Thunderbolt is of course less elegant than a single cartridge size for everything, but also a lot less restricted and more flexible, and can still lead to very tight integration. Which is why pretty much everything has been moving in that direction. I liked the concept of the Pimax Portal, because they at least tried to build a modular multipurpose device that could connect different peripherals (with USB-C and others) depending on use case, and they started with a small but still powerful compute/display unit.

            The chance that it would work was rather slim though, mostly because the main problem with any mobile device is power and heat, which leads to a lot of specialized devices optimized for a particular use case, so only a few people will accept a generic solution that can be used for more things, but is usually (a lot) worse at them than dedicated devices.

    • Arno van Wingerde

      Somebody already thought of that wireless puck idea … they called it a PC! And you were right, it can be used for other stuff besides driving your VR set.
      I guess you are referring to a console a la Steam Desk with somewhat limited hardware expressly dedicated to VR gaming, like a PS5(Pro) with a wireless connection: the PS5 has WiFi6(AX) which is good, but not top-of-the-line. Then again, neither is the GPU/CPU but instead it is pretty capable.

  • XRC

    Using Magic Leap 1 in 2019, had compute/power puck on a minimal harness I slung over my shoulder whilst crawling about in Attenborough's very impressive Immersive dinosaur application. Thought it would be an issue but didn't notice the puck at all.

  • Christian Schildwaechter

    TL;DR: Moving Quest 4 to 2027 may be a good thing for Meta, Qualcomm and also gamers; a Horizon OS based Loma XR HMD would NOT target gaming, instead be positioned against heavier offerings from Samsung and Apple that still keep all the compute on the HMD, with only the battery in a puck; the improved ergonomics could help Meta counter the much larger flat software library of its competitors.

    Moving Quest 4 to 2027 is interesting, as we somewhat expected it to coincide with Qualcomm's next XR2. 20203 Quest 3 got the XR2 Gen 2 with six months exclusivity, following the 2020 XR2 Gen 1 in Quest 2, so a 2026 Quest 4 with an XR2 Gen 3 again three years later seemed plausible. We don't know if this was actually the plan, and with several high end HMDs still expected to release soon with an XR2+ Gen 2, Qualcomm may be reluctant to introduce a newer and faster generation just one year later.

    And as Meta will still be their largest customer for XR2 SoCs by far, and the XR2 mostly a core reconfiguration of whatever top of the line Snapdragon is available at that time, postponing both XR2 Gen 3 and Quest 4 shouldn't even waste a lot of investments already made. It would allow Qualcomm to sell more XR2+ Gen 2, and hopefully would allow Meta to use microOLEDs in Quest 4, which are up to this point prohibitively expensive for use in a relatively low price consumer device. This would make their use in a 2026 Quest 4 very unlikely, even if they use only 2.5K ones. Which might be their best option, esp. if a Quest 4S targeting the giftable USD 300 price would still have to use LCD panels with Fresnel lenses that don't really allow for more than 2K.

    Like the Quest Pro was never intended for gamers, a slim XR ‘Loma' HMD with external compute wouldn't be either, that would still be the role of a delayed Quest 4. Hopefully Meta will communicate this better this time, and also have more non-gaming use cases to show for Loma than the too-little, too-late Horizon Workrooms on Pro.

    The comparisons to a Bigscreen Beyond 2 are always dangerous, as this HMD achieves its low weight and size by extreme minimalism, basically containing only microOLEDs, pancakes and a small SoC for video decoding and connecting the very light IMU and Lighthouse sensors. Loma will probably include audio and lots of cameras. 4x bw room/hand tracking, 2x hires passthrough, 2x bw eye tracking cameras and 1x depth sensors are already nine cameras more than on Beyond 2 (seven more than on Beyond 2e), adding not only a little bit of weight and size, but also the need for local processing on the HMD to handle all the raw data you cannot send through a thin cable to an external compute unit without adding lots of delay and image degradation from compression, and even this would require a SoC on the HMD.. Loma will very likely require a separate DSP/SoC on the HMD for signal preprocessing, similar to the R1 on AVP.

    • Nevets

      I like the TL;DR. Keep using those please.

  • Bram

    And so it's the compute puck replacing the smartphone here, going in tour pocket and not the actual hmd, if it connects to 5g network as well. I absolutely see the potential for wider acceptance in such a light and small formfactor then, easily to be used for productivity apps, communication, media consumption, gaming and a lot more, while traveling, at home or in the office. But Samsung and apple aren't sleeping and could turn their next flagship smartphones / iphones into hybrid vr pucks and so bring serious competion to the same targeted consumer groups.
    Historically, consumers allmost choose for 'practical use above all' as it comes to accepting new technology, so accepting a lightweight hmd as addon to their smartphone/ iphone is more likely than accepting an extra compute puck in their pocket while also still needing to bring a smartphone. So here lies a split in the forseeable xr future market: a smartphone as hybrid vr puck for light mobile use while traveling, allowing for light productivity, communications and media consumption and a more heavy separate compute puck, replacing laptop/tablets/pc's for more heavy productivity work and gaming. A bigger compute puck in combination with an onboard small physical keyboard could have potential as well as laptop/pc replacement.
    Plenty of options there.

    • Christian Schildwaechter

      We won't see smartphones used as compute pucks for VR anytime soon. Partly because VR is a constant high load application while phones permanently switch between burst and low power modes to increase battery life, so Snapdragon and XR2 SoCs use very different core configurations, making smartphones a bad choice for running VR.

      But mostly because VR is an energy hog, and people nowadays are basically glued to their smartphones. So an accessory that drains the phone battery that usually lasts for a day in less than two hours, leaving people without their internet lifeline, will be rejected. It may be more acceptable just for use as virtual screens similar to the Xreal glasses that draw a lot less energy with applications that still allow the phone to run on its energy saving low power cores most of the time.

  • James

    In summary: Magic Leap had it right all along.

    • Andrew Jakobs

      No they didn't .. cable to a puck on your belt or pocket is just awful.

      • foamreality

        I don't really see a major problem with a puck and wire. Just tuck it under your tshirt. It means the headset can be light (like bigscreen beyond) and the compute can be more powerful, and it could be put in a small back pack, strapped to arm or whatever. It could also be upgraded without upgrading the headset. You could plug it in to a PC instead when you need more power. None of that heavy battery and processor on your face which also limits how powerful it can be. Seems like a win win to me. Better something in your pocket than carrying it on your face. You don't carry your phone on your face.

        • He's right.

          A more powerful AIO requires more battery.
          More battery means more weight & heat.

          Less powerful means less battery, so lesser weight.
          But that means an undazzled consumer ….

          Wierd tradeoff.

        • Andrew Jakobs

          I have a Pico 4 which isn't heavy on the face and can be used instantly. Having to first handle the puck and then the headset is another extra hurdle.
          and the phone analogy doesn't hold up as you wouldn't carry your phone on your face anyway, unless you're using it with 'cardboard'VR. But when you use your phone, you don't use it all day with an external batterypack now do you?
          from my own experience with the HTC Vive Pro and its 'wireless'module, and now using the Pico 4, I never want to go back to something which requires my to have a cable down my face/neck to anywhere on my body as it always hinders at the most critical times, including the extra hurdle of starting to play.
          yes, in regard to comfort the Quest with its standard headstrap sure is not great and due to its design is front heavy. But with the battery in the back of the strap, like the Pico 4 it is very comfortable. But of course it could be better. The Pico 4 doesn't have an easy replaceable headstrap so if it breaks or you want another third party headstrap, you can't just replace it.
          As long as the headset has good pcvr streaming capabilities, you don't need to care about upgrading the internal SOC. But even that could be fixed by using a modular headset.

          • Christian Schildwaechter

            You are probably a prime candidate for something like a Bigscreen Beyond 2e wireless edition that adds inside-out tracking, audio and a small SoC mostly for tracking and stream decoding, in a halo strap with a (hot) swappable battery at the back at maybe 400-500g, but well-balanced.

            Which is feasible today, but currently not a lot cheaper to produce than for example a Dream for Reality or other hires HMD that also comes with a rather powerful SoC for standalone use. So it is still unlikely that someone will produce such a "streaming only" HMD that a lot of enthusiasts might love, because most people will expect a much lower price without the standalone functionality.

            It may become feasible in a couple of years, once the costs for microOLEDs have fallen significantly, as the rest of the tech is already rather cheap, and there is probably a market for a USD 400-500 light, hires, streaming only HMD. Just not for one above USD 1000.

          • Andrew Jakobs

            I don't need 4k displays, the current resolution is good enough for me, but I would like better lenses so no glaring or at least much less as the Pico 4(which already is a day and night difference compared to the awful godray fresnellenses). And at this time I'm still mainly using the headset standalone.

    • xyzs

      Scamming investors with fake claims and fake tech demos will never be what I call “having it right all along”.

  • MasterElwood

    Nobody wants a "glasses" style VR HMD when the tradeoff is a stupid puck. That's for smart glasses.

    Quest like HMD for VR is the best option hand down! Why would i care that my HMD is smaller when it's just for use at home? Make it more comfortable, maybe lighter. But for VR I don't care about the size…

    • Andrey

      You will be very surprised if you will go on Reddit and read the comments under posts with those articles. 99.9% of users there (I assume they can't be really considered "VR enthusiasts" if they only get news second-handed from Reddit anyway, lol) are in awe and can't wait for this device, because their biggest problem with the current Quest headsets is "comfort" and they don't mind having a wire going down from their head to their pocket/belt where this puck (that, again, will also weight something and most probably won't fit in every pocket) will be placed.

      Can't wait to witness the moment when all those people will buy this very comfortable to wear ~1000$ device just to realize that there is still not proper (constant flow of) content for it delivered by Meta.

      • Andrew Jakobs

        But that's more because those 'morons' stick with the default strap the headset comes with, most never invested some money in a better headstrap, which makes the Quest so much more comfortable and easier to use, especially with hot-swappable batteries.

  • ZarathustraDK

    Using a puck is a great way to take weight off the head. The odd thing I see is how all companies basically go "then you just put it in your pocket and voila". I can't possible be the only one bothered about the prospect of having a wire dangling down my side for my arm to get caught in, can I? Not to mention people have different heights so to accomodate everyone the cable will have to be longer than necessary for most people.

    Seems to me the most sensible place for a puck would be somewhere on the upper chest. The weight gets off the head. The distance for the wire is roughly the same for all people. It enables you to sit down without leaning into the puck. The wire's potential slack/taut length-difference is neglible.

    With stuff like pucks and AI-necklaces I wonder if we're seeing the drunk and clumsy staggering of technology towards a Star Trek-esque communicator-pin-computer.

    • Andrew Jakobs

      From my experience with the battery of the wireless module of the HTC Vive Pro, a cable dangling beside your body is pretty awful. Let's not forget the amount of times it popped of my belt during ducking or other weird moves, or just falling out of my pocket.
      wireless would be the only way to go, but it still needs a good battery in the headset to process that wireless signal, so just adding the whole thing to a headset would be not even increasing the weight that much.
      they should invest, at the current technology state we're in, on improving the headstrap. The default headstrap of the Meta Quest is just awful, but look at something like the Pico 4(ultra), which is one of the better comfortable headsets (but still leaves more than enough room for improvements).
      AR only becomes another matter of course, you want that one to be certainly not larger then thick sunglasses (at current technological state), but for MR/VR you want complete lightblockage anyway, so a larger headset isn't a problem, but no wires to a puck/battery/pc.

  • xyzs

    Smart move.

    Take the weight off the head at all cost.

    A cable down to my waste? I dont care unless I play a game where you are supposed to pull your virtual body hair…

    I think the people against it are just attached to the all in one device form factor and don’t care about comfort as long as it looks unified. That’s very shallow. Unified and comfortable and powerful is not yet possible.

    • mwbrady

      A cable down to your waste would be unsanitary.

  • Alex Hand me the Grapes

    I've been thinking about this. Make a headset as light as possible, which just the tech it needs to function: cameras, sensors, etc. Then, a cable running to the back pocket with the computer power necessary. Why more companies are not doing it this way is preposterous. All-in-one is a nice concept, but the weight for some people is as equally ridiculous. And, it will make the unit more comfortable for longer period of time.

    • Christian Schildwaechter

      TL;DR: external compute pucks sound like a neat idea, but you run into problems with data transfer from lots of camera data that needs to be processed and limited performance benefit from current ARM SoCs, so in reality the benefits are limited.

      Aside from a lot of people not really liking a form factor that requires a external compute unit and a cable dangling down from the head, there are some very practical problems with moving the compute away from the HMD itself. The Rift S with inside out tracking used five low resolution nIR tracking cameras at low frequency that were mostly used to correct for sensor drift from the IMU that reported translation and rotation at a much higher frequency, and these cameras might have been connected to a single USB bus running through the rather thick wire, requiring only four wires.

      Today's HMDs run a plethora of cameras, including hires color cameras used for time critical passthrough, so way more bandwidth is needed. Mobile SoCs typically have several direct camera ports with three or four lines each that connect them directly to the ISP/DSP/CPU processing the data, circumventing any slow bus. The Quest Pro already exceeded the seven CSI ports on XR2 Gen 1 with two extra eye tracking and one face tracking camera, using a daughter board to combine the streams.

      If you wanted to do do this directly in an external compute box, you'd need a lot of lines. A future Quest adding eye tracking would use nine cameras, for a minimum of 27 lines. And as you need differential signaling with two wires for longer distances/lots of cables, we are looking more at 45 lines just to get the camera data to the puck. Which is completely unpractical.

      So you need to do at least some preprocessing on the HMD itself, merging feeds or doing room/hand/controller tracking locally. Which is what the PSVR2 does with a custom SoC (only using four room/hand and two eye tracking cameras, producing a lot less sensor data than a Quest 3), or the AVP with the R1 chip featuring mind boggling bandwidth to handle 14 camera sensors, doing all tracking and passthrough even when the M2 SoC is rebooting. But this means you now need two SoC, each with RAM, with the one on the HMD just handling the sensor data and sending the results to the second, more powerful one in the external compute unit, which then renders the image and sends it back via the cable. Doable, and exactly what PS5 and PSVR2 do, but a lot more complex than just moving everything to an external box connected by a thin wire, which also makes it more expensive.

      This still might be an option for Apple with laptop/desktop class SoCs that could benefit from sitting in an external compute unit with more battery and better cooling, or when using AMD APUs. But most ARM SoC are designed for use in mobile phones, not offering a massive performance boost in a compute puck, so it makes more sense to keep everything on the HMD instead of going with a split SoC option that doesn't really improve performance. The main benefit would be reducing the weight of the HMD itself, but it would often be better and much simpler to instead improve the balance by moving the battery to the back of the head.

      • Arno van Wingerde

        But in that scenario, you would use an Apple-class ARM chip, not the mobile version for AVP quality graphics power. For lower graphics quality the advantage of offloading a lot of CPU power to the puck is more than offset by the extra connection bandwidth needed.

        • Christian Schildwaechter

          TL;DR: The M2 used in AVP is faster than the one in the MacBook Air; the bandwidth issue isn't just an inconvenience, but a fundamental problem; moving only the CPU and GPU out of the HMD doesn't significantly reduce the weight, the main benefit of an external compute unit would be allowing for faster SoCs.

          The Vision Pro uses the largest M2 variant with 8 CPU and 10 GPU cores, while the basic MacBook Air and iPad Pro use one with only 8 GPU cores.The (physically) larger M2 Pro/Max/Ultra versions offer more CPU and GPU cores, but AVP already packs the fastest variant used in devices not requiring active SoC cooling.

          Apple could create a much faster, external compute unit for AVP by pairing a tiny Mac Mini M4 mainboard, maxed out to a 14 CPU cores/20 GPU CUs M4 Pro, with a larger battery and active cooling, thanks to the R1 handling all the sensors and passthrough, largely solving the bandwidth problem. But you wouldn't lose that much weight or size, as the AVP mainboard itself is tiny. The red framed chip below is the M2/M4, the yellow one on AVP the R1, which I assume shares the RAM with the M2. Externalizing the SoC might save 50-100g at best.

          A Quest 3 mainboard is even smaller, and you still need not only signal (pre-)processing like the R1 on the HMD itself, but also active cooling due the inefficient pancakes losing 90% of the light, requiring very bright/hot backlights. Quest Pro's two fans therefore sat directly behind the displays instead of on the SoC.

          And the bandwidth problem isn't an inconvenience requiring a slightly thicker cable that is outweigh by added performance, it is a show stopper. You either massively reduce the number of sensors like the Beyond 2/PCVR, or at least merge all the sensor feeds into one or two streams you can then send over a few lines, with a lot of disadvantages. PSVR2 does a mix, using only six b/w tracking cameras and doing basic tracking and re-projection on the onboard SoC, probably sending only results plus a b/w stream for passthrough to PS5 over USB-C.

          […] a cable running to the back pocket with the computer power necessary. Why more companies are not doing it this way is preposterous.

          This is what I was reacting to, because it isn't preposterous, there are valid reasons. It's simply a lot more complicated than moving the SoC from the Quest 3 to an external case with battery and connect the two via USB-C. It is of course technically possible, and you could have hacked together a rather powerful compute puck for Quest 2 with the tiny ~10*6cm² Steam Deck mainboard plus a power bank in 2022. Commercial compute pucks doing this are used for industrial wearables with Google Glass like visors, but usually for making it ruggedized with a battery lasting all day, in use cases nowhere as data and energy intense as VR.
          https://uploads.disquscdn.com/images/b3bbcd2ce32c112335190a9cdd7834379374fbf9e9b1a29adb278c9299fa45b9.jpg

    • Arno van Wingerde

      Depends on how you play: for seated games your approach is OK.
      But for room scale VR games the cable presents more of a hindrance than the weight of the current design, at least for me. I have a Quest3 and a PSVR2 and regularly stumble over the cable.

      • Christian Schildwaechter

        TL;DR: Loma is unlikely to improve this, using a compute unit mostly to remove as much weight and size as possible from the HMD itself; it might end up overall less capable than a (in 2026) three years old Quest 3, with a meager 15% higher GPU performance; like the Quest Pro, Loma is probably more a step to the side, targeting a different user group, and not the future of Quest/Meta's gaming HMDs.

        Quest 3 and PSVR2 are pretty representative for the most valid options for high fidelity gaming VR: a mostly lightweight HMD with few sensors and limited onboard-processing, tethered via cable to a powerful, stationary compute unit (PS5). Or, if you'd added a PC or found a way to stream from the PS5 to Quest 3, a somewhat heavy HMD with more/better sensors, doing sensor analysis and tracking onboard and connecting wirelessly to a powerful, again stationary compute unit (PC) for streaming.

        We ended up with mostly variants of these two, and nobody except Magic Leap released anything that included a mobile compute puck, connected either via a tether or wirelessly. That we are now seeing some concepts with external compute units may be partly due to AVP establishing a wired box connected to an HMD as not too horrible. But mostly because this will be the only short term way to quasi-standalone XR HMDs in a thick-glasses or light-ski-goggles form factor capable of doing anything useful involving graphics. Like the Beyond 2, they simply have to remove everything not absolutely necessary on the HMD itself to drive down size and weight that far. But due to this reductionism, these new HMDs with compute pucks very likely won't be (anywhere near) as powerful as standalone-HMDs, and intended more for media consumption with hand tracking rather than gaming.

        Many VR YouTube channels have now released clickbait "Quest 4 was canceled" videos, which is rubbish. What was canceled were the Quest 4 prototypes ‘Pismo Low’ and ‘Pismo High’ targeted for 2026, but this very likely just means they were replaced by something like "Monterey Low/High" prototypes targeting tech available in 2027. The Loma compute unit being "more powerful than […] Quest 3" sounds nice at first, but the Quest 3 XR2 Gen 2 will be a three year old SoC when Loma releases in 2026.

        Based on the yearly performance increases between XR2 Gen 1 and 2, a 2026 Pismo should again have offered ~2.5x the GPU performance, and a 2027 Monterey should be around 3.4x. If Qualcomm delays the XR2 Gen 3 to 2027, then a 2026 Loma will most likely use the XR2+ Gen 2 like Project Moohan or Play for Dream, and "more powerful than […] Quest 3" means 20% more CPU and 15% more GPU performance, similar to what Quest Pro offered compared to Quest 2.

  • pixxelpusher

    A Quest 4 that's as thin and light as glasses but more powerful than Quest 3, I'm here for it! Hopefully it will have a bit bigger FOV too, considering smaller glasses form factor should be able to get right up into your eyes.

    • Christian Schildwaechter

      You could push the lenses right into the eye even with current standalone HMDs, and people have decreased the eye-lens distance by using slimmer padding etc. The problem is that at some point the greasy eye lashes start touching the lens whenever you blink, leaving a film of smudge that then starts blurring the picture. That's mostly a problem for people with long lashes, but limits how close to the eyes the lenses can be put.

      It would be great if HMDs allowed users to adjust the eye-lens distance themselves, so everyone could find their optimal compromise. And ideally also include adjustable lens-display distance, to compensate for myopia. The problem with this is that getting closer to the lens means you'll see more of the edge of the screen, and HMDs like Quest 3 are sort of optimized for one specific FoV, not rendering/displaying a lot of pixel outside of that, so adjustable FoV would lead to some waste by requiring the largest visible FoV to be rendered. It could be adjusted with eye tracking measuring which parts the user can actually see with the current eye-lens distance/FoV, no longer wasting rendering time, only some pixels for lower FoV settings.

  • Duane Aakre

    I hope some guys on Etsy get early access. What I want is a harness that goes over my shoulders to hold the puck against my upper back. I wear a small backpack when I'm biking and on a two-hour ride, I completely forget it is there unlike current Quest headsets which feel like a brick on my face from the moment I put them on.

    Extra points if they can also come up with a reel mechanism. You will need some slack in the cable to get the headset on, but once it is on, it would be nice to just touch a button to reel in the excess and just leave enough cable to comfortably turn your head.

    Come on you guys on Etsy. My credit card is out and waiting!!!!