Deep Dive: Meta Invests $14.3 Billion in Scale AI

Meta Platforms Inc. has finalized a multibillion-dollar investment in Scale AI and recruited the startup’s chief executive officer to join its artificial intelligence efforts — an unusual deal that signals a heightened push by the social media giant to catch up on AI development.   

Meta said Thursday that it has backed Scale, without including details. The size of the investment was $14.3 billion, according to a person familiar with the matter. The deal values the startup at more than $29 billion, including the money raised, Scale said in a blog post Thursday.   

As part of the investment, Scale CEO and co-founder Alexandr Wang is set to take on a new role at Meta on its AI team. Wang will join the company’s “superintelligence” unit, focused on building AI that performs as well as humans, a hypothetical advance often referred to as artificial general intelligence. Wang will stay on at Scale as a board member.   


Carol Massar and Tim Stenovec discuss the move with Bloomberg Opinion Columnist Dave Lee, who wrote the following column on the matter:   

(Bloomberg Opinion) -- It sounded like something that should have come from the sports desk — a $14.3 billion transfer fee for a young up-and-coming prospect as Meta Platforms Inc. looks to rebuild its team for the tough season ahead. The head coach is an under-pressure Mark Zuckerberg, and the hot talent is Alexandr Wang, 28. His company is Scale AI, and Meta is taking a 49% stake, it was confirmed last week. 

Were this an acquisition, it would be the second largest in Meta’s history after its $19 billion purchase of WhatsApp in 2014. But it’s not an acquisition, so don’t call it that, even though it bears many of the hallmarks of one. 

Wang is going to join Meta as a top executive tasked with running a crack team to build an AI superintelligence, sitting next to Zuckerberg at Meta’s headquarters. Other Scale AI employees will join, too, according to multiple reports. So — definitely not an acquisition, just an investment that also includes putting the the company’s top talent on Meta’s payroll. Meanwhile, Meta has been trying to poach AI talent from Google and OpenAI with the promise of “seven- to nine-figure” salaries, the New York Times reported. 

In its defense, Meta is hardly a pioneer here. As Bloomberg Tech’s Jackie Davalos mentioned in her analysis, this kind of squad building is becoming a regular occurrence. Microsoft Corp. signed Inflection AI’s co-founders; Alphabet Inc. hired Character.AI’s founders; Amazon.com Inc. took on Adept AI’s chief executive officer. On none of these occasions did they acquire the actual companies.

Two forces are driving this approach. The first, glaringly, is that the big companies are particularly keen to avoid being seen to be making acquisitions right now when judges are deep in consideration over whether earlier actions, such as Meta’s purchases of Instagram and WhatsApp, should be deemed illegal. For Meta, structuring the Scale AI deal as an investment means avoiding a long, turbulent timeline that would come with a buyout effort.

But the second factor is what I find more interesting. Ever since the launch of ChatGPT, there’s been no shortage of soul-searching among big tech firms as to why they didn’t get there first. How could it be that the pioneering work was done outside of their campuses by individuals and companies with relative pennies compared with their R&D budgets? The reason, as evidenced by these hirings and investments, is the very nature of bigness. Now that the big tech companies are mature businesses, never has thei

Episode Transcript

Speaker 1 (00:02):
Bloomberg Audio Studios, podcasts, radio news, The.

Speaker 2 (00:08):
Stock Movers podcast, your roundup of companies making moves in
the stock market harnessing the power of Bloomberg Data.

Speaker 3 (00:17):
I'm Tim Stenebeck along with Carol Master. Let's take a
look at some stocks. Meta Platforms taking a forty nine
percent steak in scale AI with Alexander Wang, who's twenty
eight years old, joining Meta as a top executive to
build AI super intelligent.

Speaker 2 (00:33):
Mistake, Right, it's a steak's g ak.

Speaker 3 (00:36):
It's not an acquisition.

Speaker 1 (00:38):
Yet, Okay, If it walks like an acquisition.

Speaker 3 (00:40):
Dave Lie says, do not call it an acquisition for
a host of reasons. Okay, yeah. Davely is US technology
columnists for Bloomberg Opinion. He spent about fifteen years reporting
on tech at the FT and BBC before joining Bloomberg.
He joins us, here, do we just not call this
an acquisition? Like? Why why are we not acquired? Why
aren't companies not acquiring things right now?

Speaker 4 (00:56):
Because they're afraid. They're afraid that the courts are going
to undo their acquisition. So at this time they're thinking
probably too risky to complicated. If the stake was fifty
percent as opposed the forty nine percent. That would of
course be a takeover rather than an investment, and immediately
that would trigger all kinds of reviews. They'd have to wait,
they'd have to get clearances, not only in the US

(01:17):
but also around the world different regulators.

Speaker 5 (01:20):
There's no time for that in the AI era.

Speaker 4 (01:22):
They've got they've got to get to work straight away.
And so we're seeing this sort of new approach which
I sort of likened in a column I wrote to
sort sort of when a football team buys a great
young prospect, you know, they pay a transfer fee to
compensate wherever he's leaving. In this case, for Wang, it's about,
you know, making investors whole at the scale AI the
company he founded, and they bring on this young talent

(01:43):
with the view that he's going to be, you know,
the what.

Speaker 5 (01:45):
He's going to run a super.

Speaker 4 (01:47):
Intelligence team at Meta and this is obviously what Mark
Zuckerbo thinks. He's needed to bring his team up to
up to speed to compete against the other big teams.

Speaker 3 (01:57):
He's going to be sitting by. Mark Zuckerberg reportedly what
is superintelligence?

Speaker 2 (02:01):
What is?

Speaker 3 (02:02):
And also scally Eyes not building its own superintelligence either of.

Speaker 5 (02:07):
That's what's danger. That's what's confusing to me. So Scale
Eye is an interesting company.

Speaker 4 (02:12):
It's kind of right at the very early stages of
AI development. It's about hiring humans to do data labeling
and say that's a tree, that's a bus, and so forth.

Speaker 3 (02:20):
It doesn't sound like super intelligence to me.

Speaker 4 (02:23):
The crucial thing and a colleague, one of my colleagues,
Ellen who Hewitt, wrote this fantastic piece about Alexander Wang
today sort of talking about how this guy knows everybody
in AI and he knows everything that's going on. And
one of the things about SCALEI that makes it particularly
useful for Meta is that it's a company that other
companies would go to when they needed data. So they

(02:43):
were working with open Ai, they were working with Google,
They're working with several other companies, and so they have
a bit of an insight into what those companies are
working on and what sort of data they were needing
to have. So now Mark Zuckerberg gets a bit of
that intelligence as we want what you.

Speaker 1 (02:59):
Have, We want to we want what you know about
what kind of the whole landscape is doing when it
comes to AI.

Speaker 5 (03:05):
What have these companies come to you for yeah, exactly.

Speaker 4 (03:08):
And also Alexander Wung is just known as a great networker, right,
and so when so much of this AI race is
about talent and bringing people on, we're hearing Zuckerberg is
emailing people personally and saying, come and join us. I'll
give you ten million bucks if you do it. This
is this is just a signal of how fierce the

(03:28):
AI rate.

Speaker 1 (03:28):
Different type of signing bonus.

Speaker 4 (03:30):
Right, like a smart star, but considerably bigger in terms
of the numbers.

Speaker 3 (03:37):
What's the promise, though, I mean, what is this super
intelligence that they're working toward. What is the world Dave
that Meta Platforms wants to build and be ready for.

Speaker 5 (03:48):
I'm not sure I want to imagine it, frankly, but.

Speaker 3 (03:50):
It's you and me both.

Speaker 5 (03:51):
I mean, there's various names for it.

Speaker 2 (03:54):
Now.

Speaker 5 (03:54):
Superintelligence or artificial general intelligence is.

Speaker 3 (03:57):
Very a g I and super intelligence are the same thing.

Speaker 5 (04:00):
It's all basically the same goal.

Speaker 4 (04:01):
This idea that you can create this essentially one machine
or you know, one technology that people can use for
whatever sort of varied us as they want an AI
to do.

Speaker 5 (04:11):
And there is this feeling that somebody will get there first.

Speaker 4 (04:14):
Right now, I'm of the view personally that there'll be
several players doing this and that's how it's going to
the land's going to turn out. However, I think there
is a sort of panic to make sure that we've
got to be you know, first, or very very close.

Speaker 5 (04:27):
Behind, to be fair.

Speaker 1 (04:29):
I mean Microsoft has done some similar moves right to
try and position themselves by acquiring you know, either different
people that seem to have certainly some AI insight or
know how. I think Alphabet has done the same thing right,
So they're all kind of jocking for position.

Speaker 4 (04:43):
They're all truking a position, and they're also taking the
same approach of we can't be seen to be acquiring
companies so we can try and acquire people.

Speaker 1 (04:51):
Now.

Speaker 4 (04:51):
I think they're not as clever as they think in
this regard. I think the Federal Trade Commission, the Department
of Justice, they know what's happening here. This is competition,
just in a different form. Because if you've got companies
that are spending crazy money on wages, does that mean
a smaller company just can't compete for talent? I guess not, right,
So there's going to be and we already know the

(05:13):
FEC is looking into some of these deals.

Speaker 5 (05:16):
I'm looking at what they're going to do.

Speaker 3 (05:17):
Yeah, what's going on when it comes to innovation at
these huge companies. I mean, I read your calm and
I think to myself, this is turning into like a
PEPSI buys Poppy for two billion dollars type of thing,
because we want to let the startups duke it out
for shelf space. When it comes to probiotic soda, we're
not going to develop it ourselves. Is that happening? A

(05:37):
decade ago, I thought Facebook killed the startup ecosystem.

Speaker 4 (05:41):
Well, I mean, look, I guess it's a question of
the shifting purpose of the ecosystem to those big companies, right.
I mean it used to be if you started a startup,
you had several exit opportunities U IPO, you get acquired.
You know, both of those channels have almost been closed off.
We're not seeing so many tech companies IPO anymore. That's
really dried up, and it's becoming much harder for tech

(06:02):
companies to acquire companies at all. But at the same time,
the startup ecosystem still has this fantastic ability to work
quickly and bring out new ideas. And I wrote in
that Columny reference that, you know, I'm not sure if
Alexander Wang joins Meta as like a junior engineer, I'm
not sure he writeses through the ranks.

Speaker 5 (06:22):
That becomes this guy.

Speaker 4 (06:22):
I think that's only something you could do outside of
the company, and that's just the nature of startups. And
one of the sort of moments of real reflection in
Silica Value was when open ai came along and launched
chat GBT and everyone's going, well, hold on a second,
why Google's been working on this for years? You know,
Microsoft's been doing it. Apple, you know, you'd think would
have been doing it, and it just wasn't at all really,

(06:44):
So there is this sort of understanding that the way
to incubate a lot of this talent, particularly of AI,
which is so talent and engineering driven, it's easier to
find that in the ecosystem rather than spend all your
own R and D for you know, ideas that might
come toughing.

Speaker 1 (06:58):
It's funny, I was thinking about Amazon when they used
to just buy all these companies. I feel like kind
of almost a catch and kill kind of thing, right
to make sure that they didn't kind of lose out
on a marketplace. This sounds like it's kind of catch
and not miss out as well when it comes to
AI a little bit right, they want to make sure
that rather, you know, rather than take something out so
that it's not competitive. It's to like, wait, what am

(07:19):
I missing with people who really might have some insight
into this?

Speaker 4 (07:23):
Just quickly, Yeah, I think it's it's they're not worried
that these smaller companies are going to steal their lunch,
worried that these smaller companies might go and help.

Speaker 5 (07:31):
Somebody else they steal that lunch.

Speaker 2 (07:32):
Check the Stock Movers podcast from Bloomberg Radio. Check back
with us throughout the day for the latest roundup of
companies making news on Wall Street and for the latest
market moving headlines. Listen to Bloomberg Radio Live. Catch us
on YouTube, Bloomberg dot com, and on Applecarplay and Android
Auto with the Bloomberg Business app.