Thread by @liron

Meta is spending $30 billion on NVIDIA GPUs just to train their AI that will surpass human intelligence.

This is more than the Apollo moon mission cost.

Here's everything you need to know: 🧵

Since ChatGPT was unveiled, OpenAI and Microsoft have dominated the headlines in artificial intelligence.

Meta, however, has been steadily increasing its presence with frequent releases of new AI products, never content to be left behind.

Meta recently announced that they are spending billions to buy 350,000 Nvidia H100 GPUs to develop its version of artificial general intelligence (AGI)

AGI is an AI that matches or even surpasses human intelligence in nearly all areas.

For those unfamiliar with the term AGI, here is a brief explanation.

Sam Altman's bold statement at Stanford shows the current climate:

"I don't care if we burn $50 billion a year, we're building AGI and it's going to be worth it."

With such high stakes, the 𝙧𝙖𝙘𝙚 for 𝑨𝑮𝑰 is well and truly on.

Some renowned competitors of Meta in this AGI race are,

- Microsoft
- Google
- OpenAI
- Anthropic
- Deepmind
- Google Brain

According to Meta, the key ingredient in their strategy to build AGI is the vast amount of data available on its platforms.

Using this information, Meta wants to build advanced AI products and services for the users of its products such as Instagram and Whatsapp.

However, achieving true AGI remains a distant goal.

Despite this hurdle, Meta has made significant strides with recent developments that showcase their progress.

Some of these developments include:

1. Meta also released a new AI model V-JEPA that learns by watching videos.

Unlike generative models, V-JEPA forms an internal understanding of the world.

This could be the first step toward Artificial General Intelligence (AGI).

2. Llama 3 is presented as a superior model to its predecessors.

It can generate realistic text, translate between languages, and offer thoughtful answers to complex questions.

It is integrated into all Meta Platforms.

Source: theaivalley.com/p/metaai

3. Meta is also testing an AI-powered neural wristband that could let you type simply by thinking.

4. Furthermore, Meta recently revealed their multimodal glasses that can:

- Play Music
- Ask what you're looking at
- Translate what you're looking at
- Send voice messages through WhatsApp
- Live stream
- Take videos and photos
- Ask questions

Meta estimates that it could spend an additional $9 billion on AI development this year.

Over and above the $28 billion it spent on its other expenses last year, taking its tally to well above $30 billion yearly.

Why is Meta spending so much on AI?

1. Building AGI:

Zuck envisions the AGI powering a wave of cutting-edge services and devices, such as more powerful digital assistants and augmented reality glasses.

2. Building the Metaverse:

A core part of Meta's vision is the metaverse, a network of interconnected virtual worlds.

AI is crucial for making these worlds interactive and engaging, with features like AI-powered avatars and virtual assistants.

3. Improving User Engagement:

Meta already uses AI to personalize content feeds and recommend products.

Further investment will make these features even more effective, leading to users spending more time on their platform.

Looking Ahead

Zuckerberg sees a connection between AI and the virtual world of the metaverse.

He predicts that by the end of the decade, many people will be talking to AIs throughout the day using smart glasses, like the ones Meta just released with Ray-Ban.

I hope you enjoyed that! If so:

Retweet the first tweet for others:

https://twitter.com/1552871185431527424/status/1786410914675057012