Recognizing the magnitude of the opportunity, many businesses and financial institutions are flooding into metaverse investments. Within the first five months of 2022, corporations, venture capitalists, and private equity firms have already invested $120 billion into the metaverse, more than double the $57 billion invested in all of 2021.
But despite the grand vision of a digital reality that runs alongside our analog world, a big question remains. Do we have the computing power to bring this to life?
What is a Supercomputer?
A supercomputer, as the name suggests, is a computer with a significantly higher level of performance power relative to a standard consumer laptop or desktop PC. Used to process vast quantities of data and solve complex computing operations, a supercomputer is generally not a singular device, but rather a collection of parallel processing units.
In response to the increased adoption of artificial intelligence and cloud computing, the supercomputer industry has skyrocketed in recent years. With the supercomputer market size expected to reach 17.2 billion by 2030, tech giants like Meta, Microsoft, and Intel are investing heavily in developing their own AI supercomputers, combining multiple GPUs into computer nodes, and connecting them together via a high-performance network fabric.
The Metaverse and AI Supercomputers
The metaverse is a collection of hyper-realistic, immersive 3D virtual worlds designed to revolutionize the way we work, shop, learn, and socialize with one another over the internet. But while the metaverse of the future is still largely undefined, one thing is certain: to recognize its full potential we need a lot more computing power.
According to Raja Koduri, Intel’s Senior Vice President and Head of Accelerated Computing Systems and Graphics, we’ll need a 1,000-times increase in our current collective computing capacity to crown the metaverse as the future of virtual interaction.
Enter AI supercomputers.
Much of the amazement around AI supercomputers lies not in their ability to process mind-blowingly large sets of data, but rather in the way in which they draw key insights from said data.
Between natural language processing (NLP), computer vision, and machine learning, AI supercomputers will be widely responsible for concurrent global communication within the metaverse between both users and NPCs. But perhaps the most important utility for metaverse supercomputers is world-building. Using transformer neural networks, neural-symbolic AI, and related technologies, metaverse providers would be able to generate entire virtual worlds with little to no human input. As the technology gets fine-tuned over time, virtual environments will be able to scale much more efficiently alongside our physical world.
Examples of AI Supercomputers
Supercomputing has quickly become a core area of competition for big tech companies, with each aiming to outcompete, outpace, and outperform their fellow incumbents.
In late January, social media conglomerate Meta announced the development of their own AI supercomputer, claiming the new AI Research SuperCluster will be the world's fastest supercomputer by the end of 2022. Designed to train AI systems with more than a trillion parameters, the RSC is apt to tackle everything from ultrafast gaming to image moderation, and metaverse world creation. Consisting of over 16,000 GPUs, it will also help Meta build better AI models and develop new augmented reality tools.
“We hope RSC will help us build entirely new AI systems that can, for example, power real-time voice translations to large groups of people, each speaking a different language, so they can seamlessly collaborate on a research project or play an AR game together,” said Meta engineers Kevin Lee and Shubho Sengupta in a blog post outlining the news.
Intel has also announced plans for its own AI supercomputer, Aurora, which will be completed in late 2022. Sponsored by the United States Department of Energy (DOE) and designed for the Argonne National Laboratory, Aurora is expected to provide the unprecedented capability to integrate data analytics, AI, and simulation for advanced 3D modeling.
While Aurora will specifically be used to tackle issues such as neuroscience and cancer research, aerospace modeling and simulation, and computational fluid dynamics, it is likely that Intel will use similar computing power for metaverse development in the future.
Last year, Microsoft developed an AI supercomputer in exclusive collaboration with OpenAI. Hosted on Microsoft’s cloud provider Azure, the single-system computer features more than 285,000 CPU cores and 10,000 GPUs, with each having network connectivity of 400 gigabits per second. According to Microsoft, when compared to other machines listed on the Top500 supercomputers in the world, its performance ranks in the top five.
The Future of AI Supercomputers - Risks to Consider
But with all good, comes the bad. And as with any new emerging technology, it’s important to
discuss the many risks and drawbacks that accompany AI supercomputing.
Given that these supercomputers are trained on massive data sets, much of these risks are drawn not from the AI algorithms, but rather from the data itself. Similar to the way judgments are passed down through human generations, if the AI models are trained with biased data, their results are bound to inherit the same bias.
As supercomputers become more prevalent, it’s important that we as both creators and consumers examine our interaction with AI. Companies will be expected, both by regulators and the general public, to keep AI usage ethical, transparent, bias-free, explainable, and accountable. At Touchcast, we recognize this responsibility and consider it a privilege to offer the next iteration of Metaverse-as-a-Service enterprise solutions to the world, placing data ethics, integrity, and transparency at the forefront of every decision.
We offer adaptive metaverse solutions for a wide range of use cases, including immersive metaverse experiences where enterprise businesses can build digital twins of their offices and retail storefronts, host events, and provide both customers and employees with unforgettable digital experiences. To learn more about the metaverse, check out the .metaverse podcast on Apple, Spotify, or wherever you get your podcasts.