I had the opportunity to try out the new, Facebook-owned, Oculus Rift DK2 virtual reality kit over the weekend. Words fail me. The experience knocked me sideways. My mind was totally blown. The virtual reality nirvana of creating presence, that feeling and sense of being “somewhere else” was fully experienced. By no means is the technology finalised but the type of VR as envisioned by many an imaginative mind appears to be very near. This got me thinking about how the whole VR circuit could pan out and some of its implications.
As the Oculus Rift headset improves, its core components will be pushed to their limits. For example, in such areas like increased frame rates, refresh rates and higher resolution. This will allow higher graphics throughput so you can render the image to be displayed and get awesomely low persistence with best possible pixel switching times. In order to experience this improvement in technology within the headset, you’ll need increasingly better discrete graphics cards (GPUs). Gaming console upgrades are at least 8-10 years. Consoles will not be able to keep up with the quick evolving refresh cycle demands you’re going to get from VR headsets like the Oculus Rift. The consoles won’t be able to output renderings of these images. VR technology’s refresh and improvement cycle will be short and rapid as implied by Palmer Luckey, the creator of the Rift.
As VR content creators design immersive worlds to maximise the potential of the Oculus Rift hardware, graphics card manufacturers will likely get back to their glory days, just like the heyday of the PC era when there was strong competition for even more powerful cards. However, I question the ability of the consumer to fully utilise the latest and greatest of these graphics cards on shorter upgrade cycles in the present day, in which wages are not rising in real terms and jobs are tougher to find. Can consumers spend, say, 1000 USD on the latest graphics card so they can play the latest VR experience, every year? Most likely not. The only reason Apple and Co were able to sell millions of 600 USD smartphones was because they were subsidised, and cost amortised over 24 month contracts by the telcos. Could graphic cards go the same route, so you that you could rent your graphics card (including the computer rig), and costs get amortised? In this model, you would pay a monthly fee like Netflix to get access to the newest graphics card.
A quick side note: Eventually mobile GPUs will most likely be integrated into the headmount display of the VR kit itself, which will get increasingly lighter over time, even lighter than it is in its current incarnation (already very light!). This means a separate high cost PC with expensive discrete GPUs may not be required. This mobile GPU based Oculus Rift is not my own idea but suggested by the founder of the Rift himself at a talk earlier this year. However, even the Rift team acknowledge that, at the moment, due to factors such as heat constraints, packing everything into a mobile headset will be tricky. For now and a few upgrade cycles, discrete heavy duty GPUs connected to PCs will most likely deliver the best virtual experience.
Consumers are not the only target market for low cost VR. Businesses (including SMEs) and professional clients may potentially offer a significantly large target market. A device like the Oculus Rift could generate cost savings for these corporate and professional users. Imagine how many hours in the virtual operating theatre future medical surgeons could log, before ever entering a real one. Firemen could be placed and tested in complex and hazardous situations before going out in the field. The military could put their ground troops to train against pre-programmed counterparts in the virtual world. The number of professional application scenarios are multitude. The consumer may struggle to keep up with the upgrade cycle of GPUs during the multi-year (and potentially multi-decade) duration in which the VR technology matures, plateaus and generates experiences that converge close to real life (including the embedding of haptics and other experience enhancing peripherals). On the other hand, the professional services and corporations will be able to stomach these graphics hardware upgrade cycles easier. It is cheaper than exposing employees to real life events in order to train them, where there would be potentially huge safety risks.
Let’s say, hypothetically, that the low cost VR market in terms of graphics card upgrade cycle is like 80% corporates/professionals, and 20% consumers. If the graphics card manufacturers make a similar percentage of their revenues in the same split, they could potentially afford to subsidise private consumer use of their graphics cards. In fact this will have a positive effect on their GPU sales. Because the more consumers who have experience with VR in the home, the more likely they will be receptive and want VR training at their place of work. This is the kind of scenario that led to the dominance of Microsoft in the workplace. For example, many households had Windows and Office suites in the 80s/90s. I’m sure many of them were pirate copies; in fact I believe there was some very high count of piracy of Microsoft software in the developing world. Even though Microsoft lost out on revenue in the home market in this sense, this only meant when people went to work, they would want to use Microsoft products, and corporations would pay up for its licenses.
The Oculus Rift itself is the culmination of nearly a decade of R&D and mass production into smartphone technology including their low cost sensors. The device uses OLED smartphone screens to display the VR image. So you can easily foresee that Oculus Rift kits will be sold cheaply. Perhaps even below cost by the likes of Facebook, as the Rift team has previously implied. Revenue generation considerations will most likely come later, perhaps from selling digital content, where there are plenty of models to borrow from. As VR improves to mimic reality better over the next two decades, the bottleneck for the mass acceptance for VR may lay in the graphics cards. Ultimately, high-end GPUs determine how well the VR can be displayed and piped to your eyes, in order to replicate the emerging reality. Let’s see. Literally.
There is no spoon but you will need powerful GPUs to power the brave new virtual worlds.