Inside Meta's AI Strategy: Acquisitions and Innovations

0:00Meta is making some big moves. They are going full gangbusters for AI. They're making tons of changes, buying companies, building data centers, all kinds of stuff. We're going to get into all that today. But before we do, Jaeden, why don't you tell them about our school community? yeah so every week we record exclusive content and we post it on our school community that we don't post anywhere else we have over 300 members an amazing conversation on there and this week jamie and i did an entire video breaking down the latest updates in um youtube and meta's short form video algorithms a bunch of really interesting things i've seen that have helped someone i know on their first uh short form video that they've made on their channel in like a year, get over 200 ,000 views.

0:45So there's a whole bunch of really interesting things. If you do anything with video, make sure to go check that out over on the AI Hustle School community. It's$19 a month. And if you lock in that price, when we raise it in the future, it won't be raised on you. And we also have over 50 videos. We post them every single week, exclusive videos, breaking down things we don't share publicly, all the inside scoop on our personal businesses that we're growing and scaling with AI tools. So go check that out if that's interesting to you. All right, let's get into what Meta is doing here. The thing that I think is the most interesting, or I guess one of the things I think is the most exciting is that they have acquired Play AI.

1:23So this is a company, Play AI has been around for a little while. They just, you know, it's like 11 Labs or any of these other platforms that generate really realistic sounding voices. And they're used by a ton of huge companies, including, I think, like Ford and IBM and Walgreens and Amazon and Salesforce, like all the big companies you could think about. A lot of them are using Play AI. It's one that I've heard, you know, quoted a ton. It's actually not, ironically, the one that I personally have used the most. And I do like a lot of AI voice stuff. have not been like, it hasn't been my number one, but they have apparently acquired it.

2:01And what's interesting is it feels to me like they probably want the tech because it integrates with a lot of the stuff they're doing, but it feels to me like they really met like Zuckerberg just wants to acquire any AI company with good talent. He wants to grab as many people as he can to build his super intelligence team. And he is spending hundreds of millions of dollars. He's giving a hundred million dollar contracts to some of these AI researchers. Now these are multi-year contracts, but still, I mean, come on, like, let's say it's five years, a hundred million dollars for a five-year contract.

2:29Like that's an absolutely insane. So he's doing that. And he's also acquiring a ton of companies. And I think just straight up just acquires the company, takes the team, maybe uses the tech, maybe doesn't use the tech. It just seems so crazy to me, but I think he, he realizes that there's a serious talent war. And if he doesn't have the best talent, meta is falling behind in AI. So, so crazy to me. Yeah. And I think this, I think this play AI one is a really interesting acquisition. I think it's actually really needed because, you know, you can see, uh, meta and their use of AI is, is related around media primarily, or at least they want it to be.

3:06And I think one thing is that's lacking in some of these video creation models we've seen is the actual audio. So take a Google VO three, for example, has amazing, amazing visual generations. Um, but when you, and it can add audio to it, But the audio just is what almost at this point is what the only thing you can tell that's fake, you know, so an audio really does make a huge difference in production quality of video. So I could see this pairing really well with, you know, a really solid video generation model. And and so I think they're, like you said, kind of just acquiring what they can and building going to be building some really cool stuff with some of these members.

3:48Yeah. So I think this is super, super smart. They've also recently acquired Scale AI. They got their CEO over who's now leading their team. So they're making a lot of these acquisitions. I think it's going to be interesting. Character AI, Meta AI, their wearables, a lot of their audio stuff. It's going to be interesting what they're able to do there. The other thing that's crazy that Facebook is doing right now, Zuckerberg just announced that Meta is going to build a five gigawatt AI data center. This is absolutely massive. And it's hard to like really understand maybe. And if you're not, I don't know, into data centers, you may not know what five gigawatts is.

4:29Apparently the footprint for this data center, Zuckerberg said, is going to be about big enough to cover all of Manhattan. Like that's how big this data center is. So it is going to be insane, like astronomically massive. What's interesting is they have to get like the electric, like the electrical figured out because it uses so much electricity. There's so many different elements that they're, they have to figure out with this, but essentially Zuckerberg is not pulling is he's pulling all the punches to try to compete with a couple different companies. So he's got like the talent side that we see they're working on.

5:05He's got the data center side. They got to do the AI. What's interesting with the data centers is Facebook was sort of falling behind on AI, it felt like for a little bit there. And they've now started to really try to compete because a lot of their competitors, like for example, Google, right? Google's creating some really impressive top-notch AI models. Well, how is Google able to make Gemini 2.5 Pro so good when it came out? It's because Google has Google Cloud. So they have so much compute in their, you know, literally an entire platform for it. They can tap into that when they need to scale up a lot of what they're doing to train a model.

5:42Meta has some data centers, but nothing to the scale that Google did, who has a whole cloud platform. And so you either have the situation that Anthropic found itself in, where Anthropic had to make a deal with Amazon and AWS to train their model. Or you got to be like XAI, who went and built Colossus, which is their own data center. And I think they built it in Memphis, Tennessee with a hundred thousand GPUs. And they put the whole thing up and just like grabbed a whole bunch of diesel generators to fuel this thing. And it was like an old, you know, abandoned, just like, like warehouse, basically that just threw a whole bunch of GPUs in there and a whole bunch of generators.

6:19And it's like, you know, the Frankenstein, but they cracked it together and that's how they trained Grok for so fast. So meta is trying to, I think, find a balance between the two. I don't think they want to go diesel generators, but like you can see the benefit that Grok was able to do that so quickly. So Facebook isn't quite as scrappy, I would say, but they're still going to build scale. So it's going to be absolutely massive and it's going to be a huge investment. So that's really, really crazy. Yeah, some of the concerns with this are water and energy consumption. This is a really interesting fact related to data centers, but they think...

6:56So in 2022, data centers used up approximately two and a half percent of our energy. So that's how much of our percentage of our energy data centers used. They think data centers could account for 20 % of America's energy consumption by 2030. I really do not doubt it. I think it's, yeah, we obviously need so much more energy. It's going to be a huge problem unless we address it. One interesting thing I will say with like in the case of Grok, there wasn't enough energy on the grid. So they're just like or they didn't have the permits to connect it or something. So they're just like, fine, we'll just get diesel generators and fuel it ourselves.

7:34In that case, like they just created their own energy. Right. But obviously, I don't think that's a sustainable solution forever is millions of diesel generators fueling these things. So I think we're going to have to come up with other like we're going to have to scale the energy production. You know, when they say like it could account for 20 % of America's energy consumption, you'd hope that we have grown our energy consumption, like our energy output a lot to be able to keep up with that. So it should be really interesting. I know we're looking at nuclear, coal, geothermal, natural gas, a lot of other things.

8:10We had the U.S. Secretary of Energy, Chris Wright, was commenting and saying that those are like things that they know they need to scale up for AI. So it's going to be interesting to see how that all actually plays out in the end. For sure. Well, if you enjoyed this episode, be sure to give us a review wherever you're listening. I really appreciate those and they help our podcast out a lot. And also, again, if you're interested in learning to grow your business or make money on the side using AI, go check out AI Hustle, our school community. Thanks for listening and we'll see you next time.