The bull case for Meta
Meta is doing a lot more right than they are wrong. While they were hammered in the market in 2021 and 2022, I think their future, especially in an AI-future, is far stronger than their past. Like in the Bull Case for Apple, I think there are 3 drivers for Meta’s future.
Obviously, a TikTok ban would benefit Meta. It would increase time on app increasing the ad placement inventory, especially in Reels, increasing lock-in. That said, there will also be new entrants to the short-form content space, and I think lock in is much more difficult when operating on an interest-graph (you see more content that you like) than a network-graph (you see content from your network). Other companies can build interest graphs too, get creators on the platform and then spin up ad networks. This business strategy is obviously oversimplified, but it’s much more difficult for you to convince all your friends and family to switch from the Facebook family of apps to a completely new family of apps.
That said, a ban is not really in Meta’s control, and so I don’t think it’s worth spending much time on.
The B2C Bull Case – AI drives higher engagement on more modalities
This is a two-stakeholder case. Stakeholder 1 is for Meta and Stakeholder 2 is for creators.
For Meta, better AI drives deeper engagement on things you care about. A better Reels algorithm keeps you on Reels longer. A better Instagram algorithm keeps you on Instagram longer. Also, a better Meta AI assistant that leverages more and more of your data is more useful to you. A WhatsApp bot that you use as a conversation partner, a thinking partner, or travel agent, or a smart assistant like Alexa is a very useful tool. “Deep engagement” doesn’t necessarily mean increasing the total-time-spent in the family of apps. It also means the quality of that time, or, better yet, time-not-regretted-using-Meta’s-services. AI like LLaMa driving higher-value engagement in the apps leads to your turning to Meta more and more and more. It’s similar to Prime driving higher-value delivery predictability drove customers to turn to Amazon more and more and more.
For creators, AI tools to create content faster and cheaper will transform Meta’s recommendation algorithms. Right now, after about an hour on Reels, the content quality falls off a cliff. The limiting factor for Meta’s recommendation algorithms isn’t quantity content, it’s quality content. These AI tools, especially in multi-modality like LLaMa 3, will drive way higher quality content and, more importantly, raise the benchmark for “average.”
Especially in short-form content, B-roll is valuable. More than half the time, a Reel is some audio with some text placed on top of B-roll, and that’s okay for many circumstances. Grade-A storytelling also deserves a place on the Meta platform, as they previously attempted with IGTV (which they then rolled into Instagram), but that’s fundamentally not Reels. Reels is fundamentally a quick hit. It’s not storytelling; it’s more like improv.
The other element for Reels creators where generative AI will crush it is in editing. AI that can intelligently edit the Reel will drastically increase the number of Reels posted to the platform because the barriers will entry will be reduced to near-zero.
With each improvement, Meta will keep users on the platforms longer, which will increase ad inventory. Which brings me to the next point: filling that inventory with ads.
Case 2: Incremental Increases Drive New Ad Spend
Imagine you, overnight, doubled your operational capacity and it was completely free. What’s the next thing you’d do?
You’d spend more on marketing and sales to drive new customers.
Enter: open-source LLaMa.
Meta has every incentive in the world to open source their large language model, just as they did when they open sourced React.
React is a development framework that allows teams to build desktop and mobile web applications as well as cross-platform native mobile applications. Essentially, one code base can power desktop web, desktop mobile, native iOS, and native Android.
Naturally, the Facebook apps are built on React, but, now, much of the internet is as well. We build most of our applications on React. One Click RFP is built with React. It’s the fastest, easiest, and most scalable way to build an application.
What’s Meta’s incentive to open-source React?
The easier and cheaper it is to spin up a new software company, then, naturally, the more money will be spent on marketing that company. Imagine I raise $1,000,000 to build a tech company. If I spend $500,000 to build an app, I have a way smaller marketing budget than if I spend $100,000 to build an app.
Making LLaMa open-source is the exact same thing, except with AI agents.
All indications are pointing to AI agents, powered by large language models like LLaMa, automating or augmenting many tasks for companies in the not-so-distant future. If there are no good open-source LLMs to build on top of, then companies must use closed-source LLMs, which will be 10X – 100X more expensive over time. Instead, if Meta can democratize the LLM technology, then they make the tech table-stakes in your business.
AI agents will probably double many company’s operational capacities, and it’ll have the biggest impact on small businesses which are resource-constrained. Making the technology free-to-use and free-to-modify is a rising tide, except that, crucially, Meta will see a large percentage of the gains in ads.
Case 3: Open Source LLaMa drives free R&D
By making the best models open-source, the developer community, for free, will tinker with it and will publish their findings. That’s how open source works.
This is the crucial feedback loop for Meta. They do the heavy lifting. They buy the GPUs. They take the technology from 0 to 1, but they are allowing the developer community to take it from 1 to infinity, but, since it’s theirs, they can use all those gains.
In short, cases 1 and 2 continue to grow thanks to case 3, and, with their scale, their ROI will be measured in tens or hundreds of billions.
Feature image credit: ChatGPT, with the following prompt: “An image representing Meta, the company. The company is on an upward trajectory. The style is friendly and approachable and in one that wins hearts and minds.“