Quest Pro Had Twice the RAM and Still Died First

On November 2, Meta published the developer note that ends two headsets at once. Horizon OS 2.13, which reaches every Quest 2 and Quest Pro by March 2027, is the final operating system either device will receive. One day later, Sony quietly removed PSVR 2 from the PlayStation Store in the United States and Japan, a month after it started pulling the listing from its own direct store.

Two platforms, one week, the same decision made twice. XR has begun retiring its installed base faster than it is replacing it, and the criterion it is using is not capability. It is unit count.

The evidence sits inside Meta’s own announcement. Quest Pro shipped in October 2022 with 12GB of RAM, double Quest 2’s 6GB, plus pancake lenses, QD-LCD panels with mini-LED local dimming, eye tracking, face tracking, and controller-free tracked Touch Pro controllers. It was, on paper, the most capable standalone Meta ever built. It dies on exactly the same schedule as the $300 headset it was meant to sit above, and the OS version that kills both is the same one: 2.13.

That should change how you think about every platform roadmap you are planning against, because it means the floor of what an XR API can assume is set by the median device in the field, not the best one. When those two things diverge, capability loses.

What v213 actually freezes

The mechanics matter more than the headline. Horizon OS 2.13 begins rolling out in January 2027 and reaches all Quest 2 and Quest Pro devices by March. Meta VR OS SDK v213 becomes the last supported SDK for that hardware. Quarterly security patches and critical bug fixes continue through the end of 2027, and Meta has committed to testing new OS and SDK releases only against Quest 3, Quest 3S, and newer devices.

Developers get a longer runway than users do. New app submissions that support Quest 2 or Quest Pro are accepted through December 2027, existing apps can keep receiving updates and content drops, and cross-play multiplayer has no announced end date. Meta also requires 180 days’ notice before a developer ends support in a way that would cost users access to content or essential services, and it offers device targeting so a studio can freeze a final build for the older hardware.

Read that as a fork rather than a fade. Under OpenXR, optional functionality arrives as extensions: a runtime advertises what it supports, an application enables the extensions it needs when it creates its instance, and a missing one fails with XR_ERROR_EXTENSION_NOT_PRESENT rather than silently downgrading. The design is clean, and it converts an OS freeze into a source-level split. After v213, any new Meta extension is by definition absent on Quest 2 and Quest Pro, so a studio that wants both audiences has three options: gate every new feature behind a runtime check, maintain a second build with a frozen manifest and separate QA, or stop supporting the older devices. Only one of those is free. The guidance in what OpenXR actually standardises explains why this is the normal cost of a cross-vendor API layer, and why the cost lands hardest on the smallest teams.

The floor is unit count wearing a spec sheet

Quest 2 is the reason this decision is awkward. Launched in October 2020 at $300, it outsold all of its predecessors combined within six months, and an internal Meta presentation indicated roughly 20 million Quest headsets sold by March 2023. It was the principal driver of Quest Store revenue, and Meta said in 2023 that the ecosystem had passed $2 billion. That is the base the platform was built on, and it is now frozen in place.

Quest Pro is the reason the decision is revealing. At $1,500 it was a commercial failure, cut to $1,000 within four months, and its planned successor, a device reported under the codename La Jolla, was cancelled on cost grounds. Meta replaced it with something much smaller and much more expensive: Meta VR Glasses, shipping spring 2027 at $1,300, with 2.4K micro-OLED displays, a 100 gram frame, and a tethered compute puck running Qualcomm’s Snapdragon Reality Elite.

The device that survived the transition and the device that did not tell the whole story in one comparison. Quest 3, at $500, was described by Road to VR at launch as a better and cheaper Quest Pro minus eye tracking and face tracking. Its spec table reads: 2,064 x 2,208 per eye, Snapdragon XR2 Gen 2, 8GB of RAM, eye tracking none, expression tracking none. Quest 2 managed 1,832 x 1,920 per eye, about 30 percent fewer pixels.

So the platform’s new baseline is sharper, faster, and poorer in one specific way that matters: it has no gaze data. Eye-tracked foveated rendering needs a gaze vector, and on Meta standalone the only device that ever supplied one was Quest Pro. When v213 freezes it, the incentive to expose and tune gaze-driven rendering paths on Meta hardware drops to zero until new hardware ships, and fixed foveation stays the default: a static high-resolution window in the middle of the frame with the periphery rendered cheaply. That is a real trade, and the mechanics of eye-tracked foveation explain why the difference between fixed and dynamic is the difference between saving 20 percent of your shading cost and saving half of it.

There is a second inversion buried in the same table. The supported memory floor is about to drop from 12GB to 8GB on Meta standalone, while the supported feature surface rises, because new capabilities after v213 are Quest 3 and newer only. Tighter memory budgets and a larger API surface at the same time is the worst possible pairing for anyone shipping a renderer. The pressure lands exactly where framebuffer and memory budgets already hurt: resolution targets, MSAA, and the number of render passes you can afford per eye.

Sony ran the same calculation, one region at a time

PSVR 2 is the cleaner version of the same story because Sony never had to write a developer note. Road to VR tracked the withdrawal: the headset disappeared from direct.playstation.com starting around mid-October, is gone from the PlayStation Store in the US and Japan, still shows stock in the EU and UK, and in Canada, Australia and New Zealand the store simply redirects to third-party retailers. Amazon availability, which had been wavering all year, has flatlined.

The detail that gives it away is the price. Inventory at the original $550 MSRP is gone, and the $400 price Sony set in March 2025 is gone with it. What remains at Best Buy and Walmart is third-party reseller stock at markups, with PSVR 2 listed at $750. Three weeks before Black Friday, there is no first-party stock to discount. Companies do not enter the holiday season empty handed on a product they intend to keep selling.

Sony reportedly never resumed production after a pause in March 2024. It closed London Studio, the team behind Blood & Truth, a year after the headset launched, reduced headcount at Firesprite, and added PC VR support in August 2024, which read at the time like a courtesy rather than a strategy. None of that is a judgement on the hardware. It is a judgement on the base.

The growth showed up at the top

The counterweight arrived two days later, and it points the other way. Steam’s October Hardware Survey was the first to include Steam Frame, and Valve’s new standalone landed in fifth place with 5.49 percent of all headsets in use on Steam, while still heavily backordered with shipments running into 2027. Valve Index managed 1.46 percent in its first month on the survey.

Two caveats keep that number honest. The survey has measured headsets connected during the preceding month since March 2020, so this is usage share rather than sales share, which makes it a stronger signal than a shipment count: people are actually wearing the thing. And some of Frame’s share may be migration rather than growth, since Quest 2 lost 3.57 percentage points between the September and October surveys.

But the direction is hard to argue with. The device adding share fastest this month is a premium standalone that costs roughly what an inflation-adjusted Index kit cost, and its most expensive 1TB configuration is priced near $1,300 in today’s money. The device holding the bottom of the market is Quest 3S at $300, which pairs Quest 3’s chipset with Quest 2’s optics. What is thinning out is the middle, and if you are choosing hardware today, the Quest 3 versus Steam Frame comparison is now a question about which end of that barbell you want to be on.

The bill arrives at the studio

Schell Games confirmed 40 percent layoffs this week, roughly 50 people from a studio that stood at about 125 after a 10 percent cut in August. It was the first layoff round in the studio’s 20 plus year history, followed four months later by a much larger one. The affected roles included Unity and Unreal Engine engineers, artists, designers, sound designers, producers, QA, and educational game researchers.

Look at what the studio is shipping next. Dragon Grove is a mixed reality gardening sim for Quest 3, Quest 3S, and Meta VR Glasses, dated for spring 2027. That is a title built for the device mix that will exist after the fork, not the one that existed before it, and it spans two hardware generations with different memory, different optics, and a completely different form factor in the glasses. A studio maintaining two engine pipelines and a title that has to run on both a 515 gram headset and a 100 gram pair of glasses needs more engineering per title, not less. When the addressable base is flat and the target matrix triples, the headcount arithmetic stops working. That is what an API fork costs, denominated in people.

The case that this is fine, and where it breaks

The strongest objection is that this is just platform lifecycle, and the precedent supports it. Quest 1 stopped receiving OS updates at v50 in February 2023, new app submissions closed in May 2024, security updates ended in August 2024, and developers could no longer include Quest 1 support in app updates from January 2025. Roughly four years from launch to complete retirement. Quest 2 got more than six years of OS updates, security patches into 2028, a 180-day notice rule, and an entry-level successor at the same $300 price. Pruning the OS test matrix is also how a platform funds its next device, and Meta is funding a $1,300 pair of glasses. The upgrade cycle among existing owners had already stalled before any of this, which the October read on Quest demand laid out at the time.

The objection is right about the schedule and wrong about the shape. The 20 million devices in the field do not disappear in March 2027. They freeze, and the store stops accepting new apps for them in December 2027. That produces a long, quiet decay rather than a clean cliff, which is worse for planning: revenue from the old base tapers without ever sending a clear signal that it is time to stop supporting it, and every studio will read the taper differently. Meanwhile the replacement at the top costs more than four times what the base device cost. A barbell is not a ladder. There is no rung between a $300 headset with 2020 optics and a $1,300 pair of glasses, and the middle of the market is exactly where the studios that just cut 40 percent of their staff used to live.

The one place old hardware got better

On November 4, Valve shipped SteamVR 2.19.1 beta, and it improved image quality for every Steam Link compatible headset, not just Frame. Sharper text, clearer colour, better detail across streaming apps, the SteamVR interface, desktop streaming, overlays, and the theater screen. The mechanism is a higher default render target for streamed PC VR, now aimed at 1.4 times the headset’s panel resolution, and a higher-quality filter for downsampling colour before encoding.

That 1.4 is per axis, so it is roughly 1.96 times the pixels, which is precisely the headroom lens distortion correction and resampling need before the encoder sees the frame. The interesting part is where that work happens: on the PC, not on the headset’s SoC. The headset contributes a decoder, a display, and a Wi-Fi link, and everything expensive happens somewhere else. This is the puck argument arriving through the back door, and it is the honest future for a frozen Quest 2: not a standalone console, but a very good display for a machine that is not bound by v213.

That is also the architecture both platform holders picked for 2027. Meta VR Glasses is a 100 gram frame tethered to a compute puck. Steam Frame runs SteamOS and streams from a PC. Neither company is betting that the next mass market device renders everything on your face.

What to do with this

Treat v213 as a hard fork with a date, not a deprecation warning. If Quest 2 revenue is material, freeze a build now, keep it on the older SDK, and set an explicit sunset date that you communicate to users before the 180-day clock matters. If it is not material, drop support deliberately and say so, rather than shipping a build that quietly breaks on hardware you no longer test.

Then plan against the new floor: 8GB of RAM, no gaze vector, pancake optics, and a feature surface that will keep expanding on Quest 3 and newer only. Budget your render passes for 8GB, not 12GB. Stop assuming eye-tracked foveation is available on Meta standalone. And watch spring 2027, because the real question is not whether the $1,300 glasses are good. It is whether they expand the base or just split it, and the answer decides whether the next round of studios is hiring or cutting.

---

Sources

• Road to VR: Meta’s Quest 2 Era is Coming to a Close as Support Winds Down Next Year
• UploadVR: Meta Is Winding Down Quest 2 & Quest Pro Support Next Year
• Road to VR: Sony Removes PSVR 2 From PlayStation’s Web Store as Stock Dries Up at Online Retailers
• Road to VR: Steam Frame Shows Strong Start as the 5th Most-used Headset on Steam
• Road to VR: Veteran VR Studio Schell Games Lays Off 40% of Staff in Fight for Survival
• Road to VR: SteamVR Update Improves Image Quality for All Steam Link Compatible Headsets
• Road to VR: Meta Quest 3 Review (specification table)