The Metaverse is about experience. Experience in a digital environment is driven by infrastructure architecture, application design, and environment performance. Once design is optimized, distance latency is what you have to accommodate next. Enter Edge Computing, which came into life to solve latency challenges. There can be little doubt that latency and a few other location-based factors will be critical to the success of the Metaverse.
So that begs the question… Is the Metaverse the “killer app” for Edge?
The Metaverse is being developed by large enterprises and will be delivered to many in ways similar to the Internet today. If Internet performance isn’t usually hampered too much by latency, why would the Metaverse be affected?
My thinking is that using the Internet is more closely related to reading a billboard. You could be 100 meters away and read it or you could be a mile away with binoculars and still read it, the experience doesn’t change too much. The Metaverse on the other hand, is more like being the rover on Mars. Today, when Houston wants the rover to do something new they have to wait about 8 minutes for the commands to arrive. Can you imagine how frustrating it would be to attempt to experience a 3D interactive environment where you almost have to plan ahead for what you want to experience before you can just go there? Like if you want to comment on this piece and begin a conversation with me, then have to wait seconds to find out whether I’ve heard you and then find out I’ve already left the room. Even more to the point, get motion sickness because the VR interface you’re using in some Metaverses isn’t keeping up with your senses.
We also have to consider the idea that whether you’re using the Metaverse to work/play in Paris from your hotel in Singapore or doing something closer to home, your policies, personal information, meta information and data compliance will all need to be accommodated. I believe the ability to have your governance and data closer to where the action is occurring will be critical to your experience.
What makes the Metaverse a killer app
Sheer potential scale and global distribution requirements are, in my opinion, the biggest reasons that the Metaverse qualifies as a killer app for Edge computing. There are estimated to be 50,000 active users of the Metaverse today and yet estimates for growth suggest $800 billion in revenue by 2024 and up to 5 billion users over the next decade. In case you’re wondering about the opportunity for NFTs, my expectation is that interest will explode alongside the growth of users in the Metaverse.
Estimates for Metaverse value and adoption are still all over the place, but for a simple comparison, it took us (technically) 36 years to get to 1 billion users on the Internet, but it’s expected that we could have as many as 5 billion users in the Metaverse by 2030 or about half the time to get to 5X the number of users. The value of this new environment is estimated to be $13 Trillion by Citi. I don’t know about you, but even a small sliver of $13 Trillion can justify a ton of investment. Prior to Metaverse considerations, Edge was already an investment worthy opportunity for most technology companies or enterprises doing digital transformation. Combine that opportunity with the Metaverse and well, sky’s the limit.
Why does the Metaverse drive so much new demand?
I expect that many metaverse applications, by virtue of the experience requirements, will require 4-5x the infrastructure investment of a comparable Web2 application. I also expect that a significant percentage (at least 50 percent) of applications delivered via the Metaverse will be net new. So basically, what I’m trying to say is that the Metaverse will drive a ton of demand, much of it net new, not replacements for existing applications and infrastructure. There will be growth in traditional hyperscale public cloud environments, but there will be exponential growth in novel new Edge delivery models.
What does the volume of demand across all trends look like?
In my 2021 blog post “The Future Demands of Edge Infrastructure Require a Rethink” I estimated there would be demand for 810 million square feet (SF) of new data center space by 2030. That estimate of 810 million SF of Data Center space was done before bringing the metaverse into the equation. If we include metaverse demand, I expect that by 2030 we’ll likely need 3-4X 810 million or 2.4 - 3.6 billion SF of new Data Center space. Each Megawatt of new, filled data center space equates to roughly $16 million in infrastructure investment over the first 10 years ($3.5 million in Data center cost, plus $250K per rack, times 50 racks at 20kW each, with a hardware replacement life cycle of 3.3 years). These aforementioned estimates capture very rough capital expenditure costs only, not operational overhead and power, etc.
2.4 billion SF of data center space would equate to roughly 80 million racks (2.4 billion divided by 30). Each rack has a value of 250K or more in servers, storage and networking gear. Each rack could hold from 10-200 servers depending on the server model.
The assumptions used to make an estimate for future demand are many; lifecycle of hardware vs improvement in capabilities, the amount of power per rack, overall efficiency of power delivery, changes in the demands of software, etc., etc. All the aforementioned assumptions taken into account; my back of the napkin math suggests that we’ll likely need to grow our existing global data center footprint as much as 3X to 2.7 Billion SF by 2030. Growth to 2.7 billion SF will be driven by continued organic expansion in existing use cases, but in large part by net new demand generated through digital transformation, Edge xomputing, Metaverse and Web3.
While I’m obviously fairly positive on the impact Metaverse adoption will have on demand for IT infrastructure, it’s also true that mature Metaverse solutions are still a few years away. There will be steady improvements on today’s examples each year going forward, but the ability to have a virtual world that always looks the same, regardless of where you’re connecting from will take considerable improvements in AI, CPU performance and maybe even quantum computing. An example might be that a virtual pothole gets created in the road for a participant in London and someone logging in from Singapore hours later still sees that pothole, as well as everyone else, until the virtual city services fill it in. The comparison would be what you see in a video game when you shoot at something and create damage, but no one else sees that same damage and if you leave and come back the damage is gone. The ability to make these environments much more real-world-like will require incredible capabilities that we can’t easily provide today.
How we grow is critical
The next eight to 10 years will be an incredible ride, fluctuations in the economy notwithstanding. I for one am very bullish on the growth of our industry, but the bigger questions are, can we grow the industry sustainably, equitably and cost effectively? We must continue to challenge our assumptions about how things are built, deployed, measured and managed, yes, but most importantly we need to improve our ability to generate power and build products without destroying our planet in the process.