But even if we increase the computing power of consumer devices, move more enterprise computing power closer to consumers, and build more centralized infrastructure, we may still fall short of our goals.
Here’s an example that blew my mind earlier this year. From December 2020 to March 2021, Genvid Technologies ran its first “MILE” or Massively Interactive Live Event on Facebook Watch. This MILE, Rival Peak, was a virtualized version of LOST Big Brother, a 13-week, 247 simulation of 13 AI contestants. While no character was individually controlled, and no audience was a single character, tens of thousands of concurrent viewers were able to influence the simulation in real time—solving puzzles to help uganda mobile database contestants, choosing what they could do, and even influencing who survived and was booted up. Rival Peak could never run on a consumer device its high CCU worked despite latency because it was designed for low-latency interactions. In fact, it barely ran on AWS. Rival Peak had eight environments production, backup, staging, QA, and development, each powered by more than a dozen GPUs and hundreds of other CPUs. The event used to run out of GPU servers on AWS, and often exhausted available spot servers during testing.
Because there are no specific players let alone "Player One", Rival Peak does not meet the definition of a metaverse. But it supports the operation of infinitely interactive and persistent virtual worlds, each with lasting consequences, and is as close to a final state as any other metaverse. Even in its nascent form, with no meaningful consumer-side processing, it has already run out of computational power.