The tech world was once wowed by virtual reality, that the technology became a hype.
The industry expected it to evolve the way people interact with computers, opening opportunities for businesses to showcase their products right in front of consumers' eyes, but in a much more engaging manner.
Facebook CEO and founder Mark Zuckerberg even renamed Facebook the company to Meta to showcase the company's focus towards the metaverse.
With big players and big brands on board, experts expected the industry to be worth billions of dollars, if not trillions.
But the hype of the alternate world the industry was trying to create, faded as fast as it came.
Thanks to generative AI hype that came after it, notably thanks to OpenAI's ChatGPT that disrupted everything, the metaverse is no longer the industry's hope for evolving the foreseeable future.
As a tech titan with only a handful of competitors, Mark Zuckerberg knows how the metaverse hype it helped create with Horizon Worlds and its Oculus headsets, is no longer the hype.
Because of this, Zuckerberg is redirecting Meta-wide efforts to develop smarter and more powerful AI instead.
To be precise, Zuckerberg wants to dedicate his company's resources to build an Artificial General Intelligence.
Or also referred to as AGI, the term is to describe a highly autonomous system that is expected to outperform humans at most economically valuable work. Unlike narrow AI, or ANI, which is designed for specific tasks, AGI possesses the ability to understand, learn, and apply knowledge across a wide range of domains.
Essentially, it aims to emulate human-like cognitive abilities.
Developing AGI is a complex challenge. It involves creating machines with advanced reasoning, problem-solving, perception, and natural language understanding capabilities. Researchers are exploring various approaches, including machine learning, neural networks, and reinforcement learning.
To pursue this goal, Zuckerberg wants to secure a whopping 350,000 or more Nvidia H100 GPUs by the end of the 2024.
According to Nvidia in its website, its H100 taps into the "unprecedented performance, scalability, and security for every workload," and that it's also includes a dedicated Transformer Engine to solve trillion-parameter language models.
"The H100’s combined technology innovations can speed up large language models (LLMs) by an incredible 30X over the previous generation to deliver industry-leading conversational AI," explained Nvidia.
The GPU, which consists of 80 billion transistors, allows bringing LLM to mainstream with its ability to accelerate AI workflows such as AI chatbots, recommendation engines, vision AI, and more.
Read: Paving The Roads To Artificial Intelligence: It's Either Us, Or Them
That's not all, because Meta also plans to use other types of chips to achieve a total computing power equivalent to 600,000 H100s using a variety of GPUs.
It's even working on its own AI silicon.
Under Zuckerberg's vision, Meta's Fundamental AI Research and Generative AI teams is working more closely together to develop core capabilities needed to achieve AGI.
"It's become clear that the next generation of services requires building full general intelligence, building the best AI assistants, AIs for creators, AIs for businesses and more needs advances in every area, from reasoning to planning to coding to memory and other cognitive abilities," he said.
The news came as OpenAI reportedly signaled it wants to raise billions of dollars in funding to build its own network of chip to supply enough AI accelerators to meet its needs.
With the trends of generative AIs and LLMs, Meta's plan to vacuum up more than a third of a million Nvidia GPUs alone, shows where the industry is moving at this moment.
At this time, Meta is training LLaMA 3, the successor of LLaMA 2, with plans to release it openly, the billionaire confirmed via Instagram.
But this plan has its own issues the company has to address.
In 2023, Nvidia shipped 150,000 of its H100 chips to Meta, and according to Zuckerberg, Meta's arsenal of Nvidia AI chips may spell trouble for other companies who want to cash in on the hype.
"We have built up the capacity to do this at a scale that may be larger than any other individual company," Zuckerberg said.
"I think a lot of people may not appreciate that."
After all, as more and more tech companies scramble to get their hands on Nvidia's limited supply of AI chips in an effort to build the best products, Nvidia's stock has risen by over 200% in the last 12 months. Overall, global demand for semiconductors has been outstripping production.
It's worth noting that despite Meta's plan to pursue AGI and openly share its efforts, and release its models for other developers to use, Zuckerberg hasn't quite turned his back on that whole metaverse and VR headset ambition.
Despite Zuckerberg has quietly dumped his metaverse ambitions since early 2023, he suggests that AGI is able to help build and fill out the project's interconnected worlds of 3D virtual reality.
"People are also going to need new devices for AI and this brings together AI and the Metaverse over time. I think a lot of us are going to talk to AIs frequently throughout the day. And I think a lot of us are going to do that using glasses. Glasses are the ideal form factor for letting an AI see what you see and hear what you hear," he predicted.