This audio was created using Microsoft Azure Speech Services
Building a virtual world is not going to be easy, fast, or cheap, with cost estimates soaring into the trillions. But many mega-sized companies are diverting ample resources to develop their portion of the metaverse in a bid to become like the Oracle in the Matrix movies. Much of the media attention revolves around VR headsets, AR glasses, haptic gloves, and other wearable hardware required for an immersive virtual reality experience. While this otherworld experience is the end result, layer upon layer of behind-the-scenes technology will form the foundation of the metaverse, including data centers and network infrastructure.
Of course, we have data centers and networks now, but they lack the speed and capacity for an immersive experience of this magnitude. Additionally, due to the almost zero latency requirement, metaverse data centers will need to be in very close proximity to the user, and network speeds must be blazing fast. For users all over the world to interact, a fleet of decentralized local edge data centers will be essential.
Facebook founder and Meta CEO Mark Zuckerberg emphasized this point in a statement before the opening of the Mobile World Congress: “Creating a true sense of presence in virtual worlds delivered to smart glasses and VR headsets will require massive advances in connectivity. Bigger than any of the step changes we’ve seen before.”
Host data centers will contain every environment, object and avatar in the metaverse
In the metaverse, a new virtual world will be digitally created, replete with all aspects of a physical environment. While a simple space like an office can be modeled, more complex environments, such as a city, are also modeled. Within the modeled physical space, the number of objects and the number of virtual people (avatars) may be very limited in that small office. However, a city street may have buildings, vehicles, dining areas and many people. Software designers create these modeled environments, objects and avatars, which are stored on servers and hosted in a central data center.
These host data centers will be mega-scale and contain every environment, object and avatar available in the metaverse. Housing all the metaverse data will drive significant power usage. For example, a planned data center (that has since been put on pause) from Meta in the Netherlands to host a portion of the metaverse in Europe has an expected energy consumption of 1,380 gigawatt-hours per year. This single data center would consume nearly half as much energy as all the data centers in the Netherlands combined. For a data center of this magnitude to gain approval, it must be built in the most environmentally friendly and sustainable way.
One metaverse challenge is physics
To make the metaverse experience immersive for the participant, the virtual environment and the participant interact in real-time to give the sense of actually “being there.” These simulated graphical elements must update rapidly in response to the interactions of the participants. The roundtrip latency required to support live, single-participant interactions has to be less than 10 milliseconds, which is much faster than today’s latency-sensitive applications, such as video calling and cloud games, that have a roundtrip efficiency of about 100 ms to operate seamlessly.
One main challenge is physics: Nothing can travel faster than the speed of light—300 million meters per second for massless particles. But data isn’t massless. It doesn’t travel in a straight line when going through the fastest transport medium—fiber optic cable. Instead, the real speed is 30% to 40% slower than the speed of light. For example, at 40% slower than light, it takes around 50 ms for data to travel roundtrip from New York to Los Angeles (4,500 kilometers [2,800 miles] times two), and that is five times too slow for a single-participant metaverse.
Realistically, the farthest the data center can be from the participant is 900 km (559 miles), and mega data centers won’t be within major city boundaries. A participant who lives in a large city could interact with all of the environments and avatars hosted in the data center that is just outside the city. This model pushes the entire rendered experience to a user’s console as a video stream with which the user can interact.
This model would also be advantageous for multiparticipant engagements in which participants live in the same city, and their avatars are interacting and evolving in the same digital world. But they all must be located physically near the host data center to deliver the video in 90 to 120 frames per second, ideally with a 2K to 4K definition with less than 10 ms of latency.
Edge computing provides a seamless experience
Delivering this experience reliably to everyone who wants to participate in a single-hosted environment at the same time and with low latency is challenging, but edge computing holds the key to powering the metaverse. Edge computing is an IT deployment that puts applications and data as close as possible to users—exactly what’s required for a seamless experience, giving users the local computing power necessary while minimizing network-based latency and network-congestion risk.
Whether you live in Paris, France, or Paris, Texas, the environment you wish to be immersed in must be downloaded into a local edge data center in close proximity to where you are. You would choose a specific environment—a small subset of the entire metaverse. If you wish your avatar to interact with other avatars, the people owning those avatars must also download that environment into their local edge data center, and so on for other avatars. The environments must then be synchronized with each other so the avatars can interact in real-time.
Building the metaverse will take a global effort that no single company or industry is capable of sustaining. Instead, the evolution of a reliable internet deployed to billions of people shows how powerful the connectivity industry can be when it works together. Deploying the mesh of local edge data centers globally at the network edge is a key building block.
This article was previously published in Forbes.