Dear Reader,

On July 10, Elon Musk’s Twitter got some competition…

That’s the date that Meta’s Twitter-like application – Threads – hit 100 million users.

Threads became the fastest app to reach 100 million users. It beat ChatGPT’s record of two months set in January 2023.

Admittedly, Threads had a bit of a head start. Meta integrated Threads with the popular Instagram app, which boasts more than 2 billion monthly active users.

All the same, it’s an impressive launch. And Threads is Meta’s answer to Twitter. But there’s one burning question…

Why now?

After all, Twitter’s been around since 2006.

CEO of Meta, Mark Zuckerberg, had over 17 years to release a competing product. And he’s not shy about ripping off a good idea.

He added the “Stories” feature to Instagram in 2016 after seeing Snapchat’s rise in popularity. IGTV was added in 2018 to compete with YouTube. And in 2020, “Reels” were added to compete with TikTok.

The reality is that Zuckerberg is trying to make his own version of Twitter for the exact same reason Elon Musk wanted to buy Twitter in the first place… for a play on AI.

Let me explain.

Data Farming

On July 1, Musk limited the number of posts Twitter users could see.

This cap was tiered based on your account:

  • New, unverified users could view 300 posts a day.

  • Unverified users could view 600 posts a day.

  • Verified users could view 6,000 posts a day.

That angered avid users that spent hours a day scrolling through hundreds if not thousands of posts.

Meta pulled forward Threads’ launch date to July 5 in an attempt to win over outraged Twitter users.

It was a clever move. Threads was pushed as the new Twitter alternative. Signups are still limited to Instagram users, but it was still able to set a new record for 100 million users.

But Zuckerburg’s plan to take on Twitter started long before that.

Meta started working on Threads in earnest in January 2023. Veiled in secrecy, the project was rushed as Zuckerberg shortened the deadline again and again.

Most analysts and journalists covering the story will point to Musk taking over Twitter as the catalyst for Zuckerberg to create Threads. But that idea makes even less sense when you look at the numbers.

When Twitter was public, its highest reported revenue was between 2021 to 2022, when it made $5.2 billion. 

That’s just 3% of Meta’s overall $116 billion it made last year.

Even today, Twitter has 450 million users. That compares to Facebook’s 2.96 billion users and Instagram’s 2.35 billion users.

Unlike TikTok with a rapidly growing user base, Twitter’s user count has a modest growth rate.

Twitter isn’t a cash cow or wildly popular. And as we’ve seen, Twitter has come under intense scrutiny from politicians over what content is allowed or disallowed on the platform.

Why would Zuckerberg want that additional headache just to emulate a platform that was a fraction of the success of Meta’s other products?

The answer is that Twitter content is great for training AI models.

The Real Motivation

Large language models (LLMs) have to consume mountains of text-based data to become effective. That’s how these AI models learn the nuance of spoken language and grind up enough information to land on the most probable right answer.

That makes a platform like Twitter and Threads wildly valuable.

Posts can include images, but they’re largely text-based social platforms. That’s unlike Instagram and Facebook, where posts are often centered around videos and images.

Zuckerberg has big AI plans. In June, he announced that Facebook, Instagram, and WhatsApp would be rolling out generative AI tools for users.

Having Threads’ text-based posts to feed his AIs would improve results and users’ experiences.

That’s exactly what Elon Musk has been trying to do all along. Musk has long touted his idea for “X,” a platform for everything.

During his buyout talks for Twitter, he stated that it could be the foundation for X. With the rise in generative AI, it’s clear that he wanted Twitter for its ability to train AI.

His recent limitations on the number of posts users can view were an attempt to stop competitors from training their AI with his dataset.

But Meta found a unique workaround by launching its rival, Threads.

It’s too soon to know if Threads will be a success as a social media platform. One hundred million users is an impressive milestone. But Threads will have to keep these users engaged… and recruit even more if it’s to be successful.

But even if it is moderately successful as a platform, it should still be enough to help inform Meta’s LLMs.

That’s the real motivation here. Threads may or may not be a wildly profitable platform. But it hardly matters. It’s a data farm for Meta’s future AI ambitions, which will undoubtedly be a huge opportunity in the years ahead.

Regards,

Colin Tedards
Editor, The Bleeding Edge%