Meta's Massive, Underappreciated Bet on AI | Sharp Tech with Ben Thompson

0:00 But like that's on a the B on a $200 billion company that's incredible.

0:06 That's like their best growth in like five years.

0:11 Yeah.

0:10 So that more than anything I suspect undergurs why suddenly everyone's piling

0:16 back into Meta like they've and this is part of the meta pattern.

0:19 They continually surprise Wall Street and surprise investors

0:23 with their ability to grow from ever larger bases.

0:27 Every time this happens and they show these incredible growth numbers,

0:31 it's off of a base that is dramatically larger

0:34 than the last time there was a crisis, right?

0:36 And and so the uh I think it this

0:39 every 18 months people say, "Oh,

0:42 it turns out that's actually a pretty good business."

0:44 Yeah, it's more like three to four years.

0:46 But yes, it happened it happened with stories, it happened with reals,

0:49 and now it's happening with uh actually we can

0:51 just shove ads everywhere as [laughter] much as we want.

0:53 So um and there's nobody that can touch us.

0:56 our users aren't going anywhere.

0:58 So, it's all really really impressive.

0:59 What did you think of the results, though?

1:02 And the idea I mean,

1:03 I I was struck by a note in your daily update on Thursday that the almost

1:09 the entire free cash flow that Meta has is going to go to AI Capex in 2026.

1:16 Yeah.

1:16 My takeaway is it's actually underappreciated the extent to which

1:21 Mark Zuckerberg is burning the boats, as it were.

1:25 Mhm.

1:26 Like I I think I don't know that people

1:28 have fully accepted or internalized the implication of this spent.

1:34 How do they come up with 135 billion at the top end?

1:37 I actually think it's pretty simple.

1:39 135 billion is how much free cash flow they're projected to have this year.

1:42 And so we're talking about putting basically

1:44 every single dollar of cash they have towards

1:47 buying data centers and chips and all the sorts of things that go into that.

1:52 And to put that in context,

1:54 a few years ago they were spending I don't have the number in front of me,

1:57 but I would say 12-$15 billion sort of in capex.

2:00 The increase from like three years ago to this year's projected number is

2:05 basically more than they've lost in reality labs over the last 13 years,

2:09 this supposed albatross where they're spending all this money.

2:11 So like like the like reality labs which is like

2:14 this symbol of meta sort of excess and Mark Zuckerberg like

2:20 honor and fine if he keeps turning out results will

2:23 let us spend [laughter] money uh is a footnote compared to

2:28 the amount of money they are they are projecting to spend

2:32 and that is yeah what it what it is is and it was this earnings call actually I

2:39 think I might have literally laughed um I laugh

2:40 at very stupid things because I'm a business nerd.

2:43 But the very first analyst question is uh

2:47 can you give us some examples of how you

2:49 expect to get sort of a return on invested

2:51 capital and over the next three five years?

2:52 This was my favorite part of your daily

2:54 update just for the record and what did say he's like no.

2:59 Yeah, exactly.

3:00 [laughter]

3:01 It was like a paragraph long no from Mark Zuckerberg as far

3:07 after like it was very honest the part that I quoted.

3:09 There was another like few hundred words.

3:12 Oh, my voice is fading.

3:13 You can hear it.

3:13 [clears throat] I can't wait.

3:16 Over 90 minutes here

3:17 of him actually then trying to come up with examples of stuff they might do.

3:22 But it was appropriately should be appropriately taken with a grain of salt.

3:25 The grain of salt that Mark Zuckerberg is like that meme.

3:28 Remember the salt guy meme of the guy like just flinging salt?

3:32 That was Mark Zuckerberg at the beginning of this answering

3:37 details and ideas.

3:38 Everything I say should be taken with a grain of salt and let me throw the salt

3:41 on this on the steak for you because uh you're going to need a lot of it.

3:45 Well, we got a question from Adam that tracks with some of my instinctive

3:49 skepticism that I gave away when I was reading the Bloomberg article.

3:53 Um so Adam says, "Ben, in your update Thursday,

3:57 I would have liked some discussion of whether Zuckerberg ought to come to terms

4:00 with the idea of being just an app or a set of apps.

4:04 I'm skeptical of Meta's right to win in Gen AI even as a capability.

4:09 Isn't all the ad revenue just funding Zuck's ongoing

4:12 yearning to be more than a set of apps?

4:15 What do you think of that?

4:18 I do think it continues to be underrated and underappreciated

4:23 the extent to which Apple saved Meta from itself,

4:26 saved Mark Zuckerberg from himself.

4:28 People forget the Facebook platform.

4:30 like the idea that we're going to be this place

4:32 for other apps and we're going to have Facebook credits

4:34 and Farmville and all these sorts of things that was

4:37 sort of the stated goal in what the company was doing.

4:42 And what happened was the phone comes along.

4:44 They underinvest in it and they actually try

4:47 this HTML 5 sort of approach that would theoretically

4:50 give them more flexibility to do this sort

4:52 of stuff in the future and got killed for it.

4:55 And then in 2012 or 2013, they just restart everything, redo the app.

5:00 It's going to be a great app on the phone.

5:04 And as an app on the phone, you don't get to be a platform.

5:08 The phone is the platform.

5:09 You don't get to do a platform on top of a platform.

5:11 Not just because of Apple's policies, which are certainly part of it,

5:14 but also because you're dealing with a 3.5 in screen.

5:17 What are you going to actually do with it?

5:18 And the reality is is that forced Facebook into their advertising business,

5:24 which is incredible.

5:25 And I pointed this out way back in 2013 when I first started.

5:28 I think I wrote an article, mobile makes Facebook just an app.

5:31 Like basically state making this case

5:33 like Facebook's being saved from themselves.

5:36 And what was incredible even back then

5:39 and it's only being extent like this point is,

5:42 you know, is a good point because it's 100.

5:44 It's even more true today in 2026.

5:47 directly ties to these results.

5:48 I pointed out back in 2013 that what

5:52 is incredible from a business perspective about Facebook

5:56 is that they somehow have gotten permission

5:58 of users to continually show them full screen ads.

6:05 Yep.

6:04 Like you scroll the feed and for some fraction amount of time every single

6:08 pixel on this device in your hand on which you're focused is an ad.

6:11 Is an ad.

6:12 Yeah.

6:12 and and actually users preferred it and liked it and responded to it

6:16 dramatically better than they did like

6:18 the banner adine advertising on newspapers.

6:21 Yeah, totally.

6:23 Like the banner ads were actually this is a great example of how people

6:26 think about these things so wrong and what

6:29 actually happens can completely upset your assumptions.

6:31 People would think, oh,

6:32 the banner ads we then they can see

6:34 their content and they're not so bothered by like this.

6:37 It's like the negative view of advertising I think is is is persistent.

6:40 People are like ashamed of advertising.

6:42 It's like, well, we'll just put it on the side.

6:43 We're not going to bother you too much, but can you look over here?

6:46 Like, this is basically the newspaper industry in a nutshell.

6:48 And what Meta achieved, Facebook back then achieved with the feed

6:53 and Apple forcing them into the constraint

6:55 of a 3.5 in screen is they figured out how to actually not tiptoe around ads,

7:02 but shove them in your face to the extent

7:05 that people weren't bothered and they actually liked It's really true.

7:10 As a user, it is a preferable experience to see a full screen ad

7:15 and then go back to the full screen content as opposed to seeing all screen.

7:21 That's the thing.

7:21 It's all content.

7:22 And that's true.

7:23 Well, and the Instagram ads content as well.

7:26 This is what happened last quarter.

7:27 Last quarter basically was a revalidation of the point I made in 2013.

7:33 It it it not only are the ads

7:35 way better now because they're all video and they're

7:37 sort of immersive and you can click through

7:38 and do all these things you couldn't do back then,

7:41 but actually we have way more capacity to shove

7:44 this in people's faces than we thought we did.

7:48 We could put more ads taking over the full screen.

7:51 And by the way, now it's not like a scroll and you're going past it.

7:54 It's like a whole video that's locked like you're when you're in a reel

7:58 and you flick and it's going through a video and you're like,

8:00 "Oh, I just spent 20 seconds watching an ad when when actually

8:03 it was totally in my control to skip it the whole time,

8:05 but the ad was compelling and interesting and actually I kind of want to buy

8:08 this thing." Like it's it's this remains underrated

8:12 and I think it remains underrated by Facebook itself.

8:14 They like they are insufficiently grateful to Apple for forcing

8:18 them to do what they didn't want to do.

8:19 Now, I think they have much more valid points later on like

8:23 we've talked about at I think what Apple does in terms of APIs

8:26 for devices where they favor their own stuff is is clearly problematic

8:31 and for particular for a company trying to build a device business.

8:34 But overall in the grand arc of Facebook,

8:37 if I'm writing the Facebook without Apple function

8:41 is arguably the most important company to their success,

8:45 which they will never ever want to admit.

8:47 Well, and look, in terms of what Facebook/Meta is now trying to do in AI,

8:53 the reason I'm skeptical and the reason I identify

8:57 with some of the questions that Adam is asking,

9:00 like I'm not a Meta investor and so far be it

9:03 for me to tell Mark Zuckerberg how to spend $135 billion.

9:08 But on its face, Meta's AI strategy,

9:11 it looks like a business that's been sketched on the back of a napkin where

9:15 No, it doesn't look like a business at all.

9:16 I think you're actually giving them too much credit.

9:18 But here's here's sort of the defense of it.

9:23 Okay.

9:22 If you truly believe that AI is a total paradigm shifter

9:28 and that everything that yes,

9:30 we'll look back on the smartphone and the smartphone

9:32 will still be around like the PC is still around,

9:34 not actually pertinent to the way people interact in the future.

9:38 And you know, this idea of just in time AI,

9:41 whether it be Blurvia glasses or or little handheld

9:45 devices or earpieces or or whatever it might be, you'll need the infrastructure.

9:51 You'll need the muscle.

9:52 You'll need to matter.

9:54 That's right.

9:55 And it's actually interesting.

9:57 The contrast between Meta and Apple right now is extremely striking.

10:03 No kidding.

10:04 Apple has basically abandoned this.

10:08 They're they're outsourcing to to Google like uh John Gruber and Ding

10:14 just pointed out like you go back to the the Tim Cook doctrine.

10:17 One of the core things is we have to own

10:19 and control the technologies that are like critical to our business.

10:23 Apple is basically either giving up

10:26 on that doctrine or stating that AI isn't actually

10:30 going to be that big of a deal in terms of disrupting their core business,

10:34 which by the way all might be true, right?

10:36 Like like maybe all along totally.

10:40 I mean, and it's fascinating to me watching Meta because we've kind of rad

10:45 over the coals for its complacency over the last two and a half years.

10:49 Even as every conversation does include the caveat,

10:53 this might be smart, but it's less interesting from a technology standpoint.

10:57 Right.

10:57 So, so that's why I'm going to push back on you laughing at this.

11:00 Like if you actually if you want if you're going to rake up over the coals,

11:04 shouldn't you be lauding Mark Zuckerberg for basically saying we

11:10 we are going to

11:12 push and squeeze our core business and we're not going to dillydally around.

11:17 We're actually going to take every single dollar from our existing business

11:21 and put it into being competitive and controlling our own destiny in the future.

11:28 Totally.

11:28 You interrupted me before I could say actually

11:30 I've come to respect Apple stance more when

11:33 I look at what Meta is doing because I look at the napkin and it says

11:37 capex plus super intelligence equals profit is basically

11:42 what I'm taking away from a lot of what

11:43 Mark Zuckerberg has had to say over the last year or so and I find myself

11:48 sitting here being like all right so what's

11:50 the product and why is that product going

11:53 to be differentiated from any of the companies

11:55 that are already dominating in this space And I

11:58 think one argument is that performance will be

12:01 correlated to compute and the infrastructure that you

12:04 have and so they'll eventually take the lead

12:06 that way or get near the top that way.

12:09 But they seem to be operating from the assumption that none

12:12 of their present- day advantages will matter in an AI dominated future,

12:17 which itself is a pretty speculative premise and it's why

12:20 a lot of this is such a fascinating bet from Zuckerberg.

12:24 Um maybe yeah, I push back on that.

12:27 I I I actually think it is a defense of meta spending

12:32 that AI even as it is today has tremendous upsides for their business.

12:38 I wrote this uh two years ago that like

12:41 meta should be the biggest beneficiary of this.

12:43 We talked about like ad generating ads just in time or the better

12:47 targeting or in the long run every

12:49 single pixel on Instagram should be monetizable.

12:53 Like it shouldn't be content versus then you

12:55 go to an ad that no everything should be an ad like like basically that is

12:59 viable like with the technology as exists today.

13:01 You can identify everything on screen.

13:03 You could link it to something else.

13:04 You could have a buy like when I right now I take a screenshot of our podcast.

13:09 I'm looking at you.

13:09 I'm like what kind of mic does Andrew have?

13:11 I should be able to click on that.

13:13 Buy that.

13:13 I should be able to buy that.

13:14 Like and that should be monetizable.

13:16 Like like there's no reason given technology

13:19 as exists right now that that couldn't be done.

13:21 So meta more than any other company we talk about the product

13:24 overhang of the model capabilities and all the things that could

13:26 be built just given what exists today never mind what might

13:29 exist in the future I actually buy the case that they

13:34 I do too I just don't know you I don't

13:36 know that you need to spend $130 billion to get there

13:39 why not like so we're talking about AI exists today what is

13:42 AI going to look like in two to three to four years

13:46 and if you're not investing today then all you're doing is making

13:50 it that much further and fall and falling behind like like what

13:56 I guess my defense of this is what do you want them to save the money for?

14:00 Uh that's a good question too.

14:02 Again, I'm not I'm not a Meta

14:04 shareholder and this makes for more interesting podcasting,

14:07 but I'm just not sure where the money leads.

14:09 I think is a fair question to ask at this point

14:12 given the lack of clarity from Meta and Zuckerberg.

14:15 Yeah, but if if it was super clear, again,

14:18 I do think there's real things in terms of their current business.

14:20 So I I I I think you're actually Yes, we laughed at his sort of I don't you know

14:27 I'm not going to give you a clear answer here,

14:29 but he did say in the answer it'll benefit our current business.

14:32 And I do think that's super clear.

14:33 Actually, if anything, I think he understates this.

14:36 And I would argue doesn't fully understand this in part because one

14:42 of my critiques of Zuckerberg is he doesn't really like ads either.

14:47 Yeah.

14:46 And uh doesn't fully understand or think about the potential there.

14:49 Like I I feel I am much more enthusiastic about meta ads than

14:53 Mark Zuckerberg is which is kind of a problem advocate [laughter] for them.

14:57 You're enthusiastic about all the AI

14:59 opportunities back going back to the interview

15:01 with Zuckerberg about a year ago year and a half ago right?

15:04 Well I just this ad angle in particular I

15:07 think he could actually sell this much better if

15:09 he was deeper into and cared more about ads

15:11 because I just think the opportunity here is astronomical.

15:14 Like just again with LLMs as they are and Facebook

15:17 inventory as it is like I just think like you look

15:21 now look yes there is an aspect of their growth

15:26 that is from shoving more ads in front of people but number

15:30 one you get away with that because your recommendation gets so

15:34 good that people don't mind and also just they talk about

15:39 like 1% increase 5% increase that level of increase in terms

15:44 of effectiveness and targeting and clickthroughs on a $200 billion business

15:49 is astronomical like like so like these these are very

15:53 real gains that they are achieving today and I think it's

15:56 completely legitimate when they talk about building new targeting built around

16:01 LMS which is basically what they're saying is we can understand

16:04 the content and they don't oh and we're going to build

16:07 new ones where you can go back further in people's history

16:10 to understand what what is interesting to them that's about increasing

16:12 context text windows and like understanding all the stuff that's going on.

16:15 All this is totally legitimate.

16:18 These are legitimate arguments.

16:19 And I actually think my critique of Zuckerberg is he

16:24 doesn't make these arguments well enough because he doesn't care.

16:26 I think Mark Zuckerberg just wants to control

16:28 the future technology which he believes is AI.

16:30 And it's working out because that is going he has smart

16:34 people who do care about ads and and build them out.

16:38 Uh but you know at least he's better than like you know open AI in that he

16:43 accepts the reality of ads and lets those people do what they need to do

16:47 right we do need money in order to sell any of this that makes sense.

16:51 Uh so yeah so I so I do think the future opportunity is

16:53 real but why do you need and this is I guess the critical thing.

17:00 Mhm.

16:59 You want let's take consider regulation.

17:05 Okay.

17:04 You love regulation.

17:05 I don't.

17:06 Uh let's depends on the regulation for the record.

17:09 A lot of bad EU regulation out there but sure.

17:12 What is my core concern with regulation?

17:16 Um, the core concern drawing on about 50

17:20 different Sharptech episodes over the last couple of years.

17:23 The core concern with regulation is that it

17:26 stifles innovation and we don't we fail to account

17:31 for the products we don't see and the businesses

17:34 we don't see in an overregulated environment.

17:37 Is that right?

17:38 Excellent.

17:39 Great job.

17:39 [laughter] A+ for you.

17:41 Uh, you know, it's always tricky at the end of the semester.

17:44 You're tired, worn down, and you have your final and you're a little sick.

17:48 Can you actually Starbucks through in the clutch?

17:50 Can [laughter] you come through the clutch Starbucks and give the right answer?

17:53 And you did it.

17:54 I'm extremely proud of you.

17:58 Okay, so the point is it's a hard argument to be

18:00 on the anti-regulatory side because on the pro-regulatory side,

18:03 you get to point to actual harms that happen.

18:06 Mhm.

18:06 And on the anti-weather side,

18:07 you have to point to things that didn't happen that were foreclosed,

18:12 which by definition, you don't actually know what they are.

18:16 And that's what is so frustrating and also concerning.

18:21 You don't actually we don't actually have

18:23 a measure of everything that didn't happen.

18:26 And and it could be actually way larger than we think it might have been.

18:32 And but but then but but then people like you who want more laws,

18:35 they're like, "Well, give me an example.

18:37 Give me what is it?" Huh?

18:39 Yeah.

18:39 But that's what you're doing to Facebook and AI.

18:41 That's what you're doing to Facebook and AI.

18:42 Right.

18:43 So there there is an aspect of the more transformative

18:47 that you think AI is almost by definition the less

18:52 you can say what the transformative aspects are going to be

18:56 simply because what is possible in the future wasn't even

19:01 imaginable previously and so if you're I I have sympathy for Mark Zuckerberg

19:07 in that he clearly believes this is that level of transformative he believes

19:12 to the extent they're going to spend every

19:14 single dollar of free cash flow they have.

19:17 But implied in that level of belief is that we're pursuing things that are

19:25 definitionally impossible to say what they are

19:27 because that's how transformed this technology is.

19:32 Yeah.

19:32 Now maybe an insane perspective for a public company CEO,

19:36 but that's why that's the whole point of having founder control.

19:39 Like so totally I mean I I respect where he's coming from.

19:42 I do think Adam nailed it that Zuck at his core

19:45 is a technologist and a dreamer and he feels like this is

19:49 the next big thing and he wants Meta to be at the center

19:52 of it and it's his company so it's his prerogative to spend.

19:57 I just am not sure it's an existential concern

20:01 to the extent you need to spend every last I know but but the point where you

20:05 validate whether it's existential or not is too late.

20:08 M that's like that's what Apple's that that like we're running an experiment.

20:12 Let's let's let's schedule our podcast for 2036

20:17 and see who was right, Apple or Meta.

20:20 And I think ultimately my uh skepticism here is a vote

20:23 of confidence in the moat that Meta has with Instagram

20:27 and Facebook and how durable those businesses are going

20:30 to be even in a paradigmshifted environment 10 years from now.

20:35 But maybe I haven't read enough Daario essays to fully

20:39 internalize the revolution that's coming for all of us.