The AI boom is hitting real limits. In this episode, the Newcomer team breaks down why data centers are running out of power, what is really happening inside OpenAI after its latest shakeup, and why neocloud players like CoreWeave may be heading toward a financial crunch.
Eric Newcomer, Tom Dotan, and Maline Renberg explain the investor panic behind the scenes, the brutal GPU economics, and what these cracks mean for the future of AI infrastructure.
00:00:00
The team here at Newcomer digs into our favorite predictions
00:00:03
for 2026 parlor game for the tech industry.
00:00:06
We decided to look at some of the top predictions from the
00:00:09
information from investor Fred Wilson, and we've got some of
00:00:12
our own. Lots of speculation about
00:00:14
mergers, Elon companies going public and a shake up at open
00:00:18
AI. Stick around for our
00:00:20
conversation. I talk with Tom and Madeline
00:00:22
about what's going to happen this year in tech.
00:00:25
This is the Newcomer podcast. Welcome to the Newcomer podcast.
00:00:36
I'm here with Tom Doton and Madeleine Renvarger.
00:00:39
We are going to spend some time engaging with other people's
00:00:44
predictions. We've got a great list, our
00:00:46
favorite three from the information and our favorite 3
00:00:49
from investor Fred Wilson, epic investor in Airbnb and Coinbase.
00:00:55
And he always has some spicy, sometimes self interested
00:01:00
predictions about the future. Much more specific I think than
00:01:04
a lot of venture predictions out there.
00:01:06
So we're going to get into it. We're going to tick through
00:01:10
these outside predictions and then maybe Madeline, Tom and I
00:01:15
will offer some of our own. I was saying before this episode
00:01:18
what my, my daughter is now three months old.
00:01:21
I'm like, what, what do I know about the world outside of my
00:01:24
house to, to see into the future?
00:01:26
But maybe I've thought about it some and we'll, we'll see.
00:01:31
We'll see what spicy thing I I, I land on at the end of this.
00:01:34
But. Happy 2026.
00:01:36
American empire is back. Hard power has returned.
00:01:39
We are deposing leaders. This is this is the year of
00:01:42
action. Welcome to the American
00:01:44
Hemisphere. Over here we're recording live.
00:01:46
Yeah, what is text play in the Don Row doctrine?
00:01:49
Yeah, exactly. It's a it's a new era.
00:01:52
Anything could happen. Maybe just an overall wild, wild
00:01:57
year. All right, Madeline, you want to
00:01:59
hit us with our first prediction?
00:02:01
Sure. Let's start with the
00:02:03
informations list. Because they had, you know,
00:02:05
their whole editorial team put together a list of predictions
00:02:08
and we picked three that we feel strongly about debating.
00:02:13
Eric, when you were there, did you, were you around at the
00:02:15
information long enough to do a year end prediction thing?
00:02:17
Because it's kind of a, it's a big tradition there.
00:02:19
It's like big year end content cycle.
00:02:21
I think it might have started in the beginning.
00:02:23
I mean, I was there a year and a half from the very beginning.
00:02:26
Harder. I don't remember.
00:02:28
I did one my first year there and I predicted this is a real
00:02:31
2014 moment, but I predicted that Yahoo was going to buy CNN.
00:02:38
And you were never held accountable for getting that
00:02:40
wrong, you know? But not only was I not held
00:02:42
accountable, it was taken seriously by some news outlets
00:02:45
as like reported news, which we'll get to as a thing that
00:02:48
happened to information again. And like Business Insider, of
00:02:51
course, like aggregated it and they had to like scramble The
00:02:54
Jets. That's CNN and Yahoo.
00:02:55
Be like we are not buying CNN. You're not getting acquired by
00:02:58
Yahoo. So for whatever reason, year end
00:03:02
news cycle is slow. People take these things far
00:03:05
more seriously than they should. Well, this, yeah, they're,
00:03:09
they're things that obviously the reporter doesn't have the
00:03:12
sourcing to report as fact. That's why.
00:03:16
They're predictions. There's a science of predictions
00:03:18
of like. I couldn't confirm this, so let
00:03:20
me throw it out there and just. Throw out my hunch.
00:03:23
And then some of them are just sort of speculative wild out of
00:03:26
the blue, like, sure, why not? And they're all in the same
00:03:28
list, so you're always like, which is which?
00:03:31
But yeah, should we jump into the list?
00:03:32
Should we do? This.
00:03:33
Let's do it. Let's do it.
00:03:35
Our first prediction from the Information Open AI buys
00:03:40
Pinterest to bring in its shopping and AD businesses in
00:03:43
house. I appreciate the gumption.
00:03:45
That's a good prediction, right? Like you, you want it's two big
00:03:47
names, it's an acquisition. It's something you can be proved
00:03:50
right or wrong on. So like hats off to them for for
00:03:53
doing that. However, insane, just just
00:03:56
completely way off base. Nothing, nothing going there at
00:04:00
all that, you know, Pinterest is like a $18 billion company right
00:04:04
now. So that would be the largest
00:04:05
acquisition in their history. They're not even sure they want
00:04:08
to do ads at the moment because they're fighting on the LLM
00:04:10
side. It's a completely adjacent
00:04:12
industry. It horrible move for them to do.
00:04:16
I just complete complete nonsense again.
00:04:18
I. Appreciate the AI you're saying.
00:04:21
Yeah, yeah, yeah. In a world in.
00:04:23
Which Pinterest would love it? What they're Yeah, they're worth
00:04:25
18 billion, which, you know, isn't insubstantial, but they're
00:04:29
down 61% over the last five years.
00:04:36
Well, yeah, it'd be great for Pinterest.
00:04:37
Yeah, sort of a languishing company.
00:04:40
They had a window. Microsoft wanted to buy them.
00:04:42
I think PayPal wanted to buy them at a time.
00:04:45
They they, they, I don't remember what happened with
00:04:47
those deals. Like they are the perpetual, you
00:04:49
know, player on the market, it seems.
00:04:51
But for open AI to do it is just like a mashing of a, you know,
00:04:56
company that seems to be on the market and a famous company in
00:05:00
the news. But logically, Nope.
00:05:04
I I do a lot of shopping in open AI.
00:05:08
Really. Have you found it to be really
00:05:10
effective? I haven't tried it as much yet.
00:05:12
I, I feel like it's more window shopping, you know, I don't know
00:05:15
that I'm pulling the trigger on a lot of stuff.
00:05:17
I was trying to buy my sister a watch for Christmas and it was,
00:05:21
it was OK. All right.
00:05:23
You guys don't use it for shopping at all?
00:05:26
I haven't really I've used it a little bit in some products
00:05:30
review, you know, kind of aggregating there, but not
00:05:35
really like really in depth for shopping.
00:05:38
I should I should try it out? I should check it?
00:05:39
Out I mean I I don't think it's good enough.
00:05:41
You know it's it's it's the same it's like compelling.
00:05:44
You're like, I want a leather band watch that's silver in this
00:05:48
price point. Give me some options and it'll
00:05:50
give you some, but then you know it's not rigorous enough to
00:05:54
stick to the the guidelines. My thing I'm trying to do with
00:05:57
it is get it for travel planning.
00:05:59
Like I always want to see if it can pull some unlisted link for
00:06:02
some deal at a hotel or a flight or something.
00:06:04
So I plug it in specific dates for travel or destinations with
00:06:09
cheap flights at this time. And I'm waiting for the day
00:06:11
where it actually gets me like a $300 international flight, but
00:06:15
not yet. Maybe they're catching on on how
00:06:17
to hack around that, but some people are have been able to do
00:06:19
that. So that's my shopping use, I
00:06:21
guess is for traveling. I, I tried a deep research once
00:06:24
when I wanted to buy a new bag when I turned 40, I decided I
00:06:27
shouldn't be wearing a backpack anymore.
00:06:29
And so I had it run down like, you know, nice leather satchel
00:06:34
bags and, and it came with a bunch of stuff and I researched
00:06:37
it and then I was Googling some of these things and there was a
00:06:39
sponsored Google ad at the top and I was like, oh, that looks
00:06:42
cool. That's way better.
00:06:43
And I ended up buying that bag so.
00:06:45
Google strikes. Again, ad dominance.
00:06:47
Yeah, they're destroying of search with sponsored content
00:06:50
actually worked on me and and I'm enjoying my satchel and page
00:06:54
but no, not for me. All right, we're not buying the
00:06:58
open AI, buying Pinterest. Yeah, next prediction, Madeline.
00:07:01
But hats off good prediction. I I, I appreciate it.
00:07:03
It's. Just but great prediction in the
00:07:05
spirit. Of prediction that's.
00:07:07
True predictions. Yes.
00:07:08
OK, moving on prediction #2A. Speaking of predictions, a Stock
00:07:13
Exchange buys a prediction market.
00:07:16
Meta. Very meta.
00:07:18
I think that this, I'm skeptical, but you know, we've
00:07:22
been really seeing the rise of prediction markets in the last
00:07:26
few months and I could see it as a pathway for legitimization if
00:07:32
they wanted to take it. But I don't know.
00:07:34
What do you all think? I think this is going to be the
00:07:36
year of prediction markets. I just, there's so much momentum
00:07:39
and investor interest in these things.
00:07:41
I from a personal and moral standpoint, I'm, I'm pretty
00:07:44
horrified by it. But there's no question that
00:07:47
like, I think this will come away.
00:07:49
My prediction is that predictions matter in 26 and the
00:07:51
prediction markets. So I like the idea of it.
00:07:54
I don't know for stock exchanges how many acquisitions they make.
00:07:58
Like how often do you see? The real question is, does a
00:08:03
Stock Exchange want to get in the gambling business?
00:08:07
Because I think the reality of these prediction markets is the
00:08:12
reality of this these prediction markets is that they're, they're
00:08:16
gambling sites more and more that that's the real business
00:08:19
opportunity and their brands are eventually, you know.
00:08:24
Well, they're they're gambling breeze with the occasional
00:08:27
insider trading, Eric, they have occasional insider.
00:08:30
Trading as well, but. Yeah.
00:08:32
It's to me, it's, it's the intellectual argument behind
00:08:36
NFTS and you know, that side of tokenization put into, you know,
00:08:41
a, a trading market that's not even like that big of an
00:08:43
insight. That's literally what it is.
00:08:45
And you're kind of seeing that group we funnel around
00:08:48
prediction markets as they moved off of kind of NFTS, so.
00:08:52
I think it's more likely a casino buys a prediction market
00:08:56
than a Stock Exchange because I feel like a Stock Exchange
00:09:00
doesn't want to get in the gambling business.
00:09:02
But if the gambling companies decide, oh man, this regulatory
00:09:08
arbitrage is going to persist, like we can't get the government
00:09:12
to stop them, then then they might say we need to get in this
00:09:16
business right now. Well, that's definitely like, if
00:09:18
you can't beat them, join them because so many of these
00:09:20
gambling, you know, casinos and, and and sports gambling sites
00:09:25
are pissed off because they fought so hard to get regulatory
00:09:28
approval in different states. And then prediction markets just
00:09:31
barrel through to, you know, find a loophole to get them
00:09:34
done. So right.
00:09:35
If they just if they can't get them banned, you might as well
00:09:37
just buy it. I I agree.
00:09:38
There's definitely gonna be juice around prediction markets.
00:09:40
Yeah, I. Love it, no question.
00:09:43
Moving on to prediction reaction #3 Ilya Sutzkever sells safe
00:09:48
super intelligence. Smart.
00:09:50
I mean, sure, yeah. They, I don't know what
00:09:53
investors are expecting at this point from them.
00:09:56
Like their their goal is like basically like we need to
00:09:59
terraform Mars well. They also said that they were
00:10:02
not going to release a product until they reached super
00:10:06
intelligence, whatever standard that's measured by.
00:10:09
But then they have backtracked that and I think in the last
00:10:11
couple months have said that they would explore releasing
00:10:14
some products before they reach that goal.
00:10:16
So clearly there's some need to make money of some kind.
00:10:20
It's not just, you know, hanging out in that ecosystem, but
00:10:22
whether or not that works in the foundation model race, I'm not
00:10:25
sure. This would certainly be the most
00:10:28
extreme license and poach, right?
00:10:31
It's like obviously this would be done in the not an
00:10:34
acquisition acquisition when you know, half of the value is
00:10:38
probably just hiring Ilya and the other half is getting the
00:10:41
team. Yeah, very plausible.
00:10:47
We just don't have very much visibility into what's going on
00:10:49
over an SSI. Yeah, it's almost like the most
00:10:53
cynical argument for a licensing Aqua hire because I don't even
00:10:56
know what there would be to license at this point.
00:10:59
We don't even know what IP they have.
00:11:00
Just Ilya is Ilya is LeBron. Well, he's older, but you know
00:11:05
he's who's. I don't know, I'm not watching
00:11:06
enough NBA these days, but he's, you know, he, he, he's that big.
00:11:11
So whoever gets or. Is he like, he's like Kevin
00:11:13
Durant or like I don't ever. Again, that's old like I don't.
00:11:17
I don't know exactly. All our narratives are from five
00:11:20
years old. We don't we.
00:11:21
Don't have. We're on the combo.
00:11:22
Is he Luca? Anyway.
00:11:26
Yeah. But Luca's dominant.
00:11:28
I I'm sort of saying, I think he's.
00:11:30
Oh, he's post. Or he's.
00:11:32
Post dominant. I don't I.
00:11:35
Don't know, I don't know. You see like Noam Shazir going
00:11:37
into Google and getting a lot of credit for the Gemini turn
00:11:40
around. Let's put it like this.
00:11:43
There's no question that there is a multi billion dollar deal
00:11:45
just to get Ilya at any point in time.
00:11:49
If he decides one day or his investors push him one day to do
00:11:52
this like it's going to be a massive outcome for him.
00:11:55
So it's really it's really just up to him and his investors and
00:11:59
but it's my smart pick. I could see it happening.
00:12:01
All right, those are three of The Information's predictions.
00:12:03
We'll link to them maybe in the episode notes.
00:12:07
Here we got Fred Wilson, epic famous venture capitalist.
00:12:12
Moving on. Spicy predictions from Fred.
00:12:15
You like, you like Fred's predictions.
00:12:16
He does them every year. And he's he seems like he's much
00:12:18
more interested in being a personality than an investor
00:12:21
these days, right? Or where do you, where do you
00:12:22
situate Fred? And he's always used his writing
00:12:27
to, to stay out there. I think he's he's a good writer.
00:12:30
He's succinct. I think, you know, he knows to
00:12:35
make the predictions spicy. And sometimes, you know, like
00:12:39
many predictions, they're guilty of wish casting a little bit,
00:12:42
you know, forecasting what you want to have happen.
00:12:45
But I think he's a smart guy. I think they're interesting.
00:12:47
We can. We can argue about them.
00:12:49
Yeah, he has a great history and he is known, I would say as a VC
00:12:54
influencer of sorts with his writing.
00:12:56
I think it's pretty influential. So moving on to the first Fred
00:12:59
Wilson prediction, we will debate smart glasses finally
00:13:02
reach product market fit in 2026, but it won't be Meta that
00:13:06
delivers the winning approach. I don't think this is true.
00:13:09
I think maybe I'm just very bearish on the sort of like VRAR
00:13:14
working anytime soon and perhaps the glasses are the closest
00:13:18
we're going to get to get it to work.
00:13:19
But I just don't think people like stuff on their face.
00:13:22
I think Airpods work because they're in ear.
00:13:25
I think other devices. Wearing glasses.
00:13:27
I but smart glasses and things floating around my cycle of
00:13:31
vision. My glasses will look dirty.
00:13:33
I just don't see the appeal of that.
00:13:35
As the lifelong glasses wearer, what do y'all think?
00:13:38
I don't see it. Well, I guess the the problem I
00:13:41
have is like if there is a product market fit, it almost
00:13:44
has to come from Meta because they're the only one with any
00:13:47
sort of scale. Like who's going to come out of
00:13:48
the blue? Well, product market fit is the
00:13:50
biggest wiggle wiggle room claim because it's like, you know, not
00:13:58
not that it's market penetration or anything.
00:14:01
It's like, oh, it people start to see the value.
00:14:04
Yeah. I don't know what company he's
00:14:06
even envisioning here. I'm sure he's got one in mind.
00:14:09
Yeah. I mean, it's a really like from
00:14:11
a start up perspective, there aren't that many that I think
00:14:13
have the cash sort of compete at the hardware scale.
00:14:17
I mean, Snapchat is like snap is like kind of still plugging away
00:14:21
at this thing. Really.
00:14:22
It's sort of quixotically, you know, like no one really can
00:14:25
control Evan as he continues to invest money in this thing that
00:14:29
he's done forever. And then I don't know who else
00:14:32
there is. Like, like I, I will admit I've
00:14:33
seen more meta, you know what, what are they called the the
00:14:37
smart glasses they have. Well, there there's the Ray Ban
00:14:40
glasses, but I. Don't know.
00:14:41
Yeah, the Ray Ban glasses, Yeah, yeah, whatever.
00:14:43
Like I've seen more of them around than I would expect.
00:14:47
All right, I I've got a prediction, Maybe I'll come up
00:14:49
with another one in my official slot, but this is very relevant
00:14:53
to this. I think the open AI device with
00:14:57
Johnny I've will be a pen. This has been my friend of the
00:15:02
podcast Max child has been hammering that it's going to be
00:15:06
a pen. I think Bloomberg literally did
00:15:08
a column based on Max's conviction.
00:15:11
And then there was some sketchy report that that seemed to
00:15:14
substantiate it. Max isn't taking a victory lap
00:15:18
yet, but I I'm listening to them.
00:15:20
And so he's bullish. He's bullish on a pen, which,
00:15:25
like you're saying, is a sort of a different device.
00:15:27
And the idea is, you know, you keep the pen on you and it
00:15:29
records and maybe when your hand it takes in all your hand.
00:15:33
I was gonna say, would it analyze your handwriting as
00:15:35
well? Yeah, I think that's part of the
00:15:36
goal, yeah. Lock it in.
00:15:38
I mean that would that. Was the mistake that Humane made
00:15:41
They did a pin but not a pen, right?
00:15:43
Right. Like they were off by a letter.
00:15:45
Switch the vowel. And then, yeah.
00:15:47
Multi. Billion dollars?
00:15:48
Oh no, we met. We were doing a pen, not a pin.
00:15:52
I like bringing up the Johnny. I will get to open AI in a bit,
00:15:55
but but I, you know, something needs to come out of that whole
00:15:58
operation this year. So might as well be something.
00:16:02
But with the pen, like does it? Does it like you highlight text
00:16:06
or something like on paper and it would?
00:16:09
Well, I just think you're, you're like a doctor.
00:16:10
You're like walking around, you have your ambient thoughts, you
00:16:13
know, it's, it's all feeding into your model.
00:16:15
And then when you're ready to sit in front of the computer,
00:16:18
you know, you output it and you have all these conversations.
00:16:21
It's in some ways the dream of AI or you know, it's like we all
00:16:25
want to get away from screens, right?
00:16:27
So if you're building a device like that allows you to escape a
00:16:30
screen so that then when you know AI can digest it.
00:16:35
I don't know. Yeah.
00:16:36
Well, that's interesting. Having a screen free AI note
00:16:38
ticker but you would still be taking the notes.
00:16:40
I don't know. Interesting could.
00:16:41
It be the pen from Goldeneye and it's like a very tense situation
00:16:45
where every time they double click it, it arms the bomb and
00:16:47
double click it again, it unarms it and you have to keep track of
00:16:50
it until finally Boris explodes. That's the safe super
00:16:54
intelligence pen that they're working on.
00:16:56
Yeah. And moving on, 2nd Fred Wilson
00:17:00
prediction. We're going to debate the SpaceX
00:17:02
IPO, The largest ever marks the market top and we are in a bear
00:17:07
market by year's end afterwards. Solid, solid prediction.
00:17:11
I'm I'm not going to push too hard against that.
00:17:13
That is very reasonable. You guys are starting to look
00:17:17
into this more than me. OK.
00:17:18
The SpaceX IPO is apparently going to hinge on space
00:17:24
manufacturing, right? So the SpaceX story is obviously
00:17:27
we'll build rocket ships, the reusable rocket ships.
00:17:29
Then it's like, OK, we'll build satellites.
00:17:32
And that that is, you know, been a great success.
00:17:35
Starlink is generating a lot of money.
00:17:37
That's the real business, seemingly.
00:17:39
But it's an Elon company. He wants the valuation to be
00:17:43
what, like $1.5 trillion, like the largest IPO ever.
00:17:48
So you need, as Elon well understands, you always need
00:17:50
another story. And so it's going to be space
00:17:53
manufacturing. So are people going to buy like,
00:17:57
is a space manufacturing thing something that's coherent?
00:18:00
Do people buy it? There's a lot of hype right now
00:18:03
about it. I mean, the number of investors
00:18:05
who've been talking to me about space data centers is something
00:18:09
worth looking into. Sorry, is it space data centers
00:18:11
instead of manufacturing? Well, it's a lot of.
00:18:14
Things I mean I well also. It is space manufacturing.
00:18:16
As well. That's that's what we.
00:18:17
Do yeah, no. It's it's just space.
00:18:20
You know, it's all going to happen in space.
00:18:21
Everything's, everything's. We're doing shit in.
00:18:22
Space. Now we're done here.
00:18:25
We've moved outside the orbit or we're in the orbit, but we're
00:18:27
off. We're off Terra, Terra Firma.
00:18:29
The things that you were doing down here, we can now do it up
00:18:31
there. Otherwise it's it's sort of the
00:18:33
same. We're no longer in the arena,
00:18:35
but we are in orbit. Yeah.
00:18:37
Do you think, do you think Mars is gonna be in the prospectus or
00:18:41
some of the Elon fans are a little disillusioned by
00:18:44
everything? Good prediction.
00:18:45
I would say 98% likely Mars has at least one if you control F on
00:18:50
the S1. I hope so, right?
00:18:53
Like, I hope it's a risk factor. Risk factor.
00:18:55
Elon's real, real reason for building this company is to get
00:18:59
us to Mars. And we might burn endless sums
00:19:01
of money, like trying to get there.
00:19:03
I don't know, Elon fanboys. I mean, it depends which kind of
00:19:06
whether you're trying to make money on his companies or not.
00:19:08
But like, some people worry that he's lost sight of getting to
00:19:11
Mars, right? That that's what he's supposed
00:19:13
to be doing and that he's not like, committed to the cause.
00:19:18
Being a public company would definitely hurt that.
00:19:21
Like you're able to get by a lot more with private investors.
00:19:26
You know, having some sort of Mars like plan where as I think
00:19:30
having analysts asked you on 1/4 by quarter basis why there's
00:19:34
this much investment in the Mars program is is probably
00:19:37
frustrating to him. I mean Elon's been known to tell
00:19:39
investors on earnings calls to go fuck yourself, but I I would
00:19:45
be shocked if there is not at least some mention of Mars with
00:19:50
within the the S1. But for the time being it's just
00:19:53
manufacturing in space. Pills in space?
00:19:56
We we agree largest IPO ever. Hopefully in 2026 we're going to
00:20:00
waste a lot of energy talking about it if it's not in 2026.
00:20:04
And then I think the thing we don't know is, is it the top
00:20:08
tick? I I definitely think there's, I
00:20:11
don't know, it's not we work, there's a real Starlink
00:20:13
business, but. There's definitely a risk that
00:20:16
just like it's a similar founder situation where it's like shout
00:20:20
the biggest number you can and hope everybody continues to
00:20:23
believe you. And you know, some of the air
00:20:26
could come out of that ahead of the IPO.
00:20:30
There's already so much speculation in Tesla as to what
00:20:34
the future of that company is really about.
00:20:36
I mean, a huge percentage of its value comes from robots right
00:20:40
now. And like the Optimist program,
00:20:42
which that hasn't made a cent and isn't at even close to real
00:20:46
world level production. So yeah, I mean, people just
00:20:50
that's part of the Elon gambit and it's worked for him as a
00:20:53
public company is to like have these unrealistic sounding
00:20:56
programs that people continue to believe are highly valued.
00:21:01
So yeah, Mars, why not? For our final Fred Wilson
00:21:05
prediction to debate today, an AI generated song is nominated
00:21:10
for Song of the year in the 2027 Grammy nominations.
00:21:15
So I guess we're pushing a little ahead there, but I guess
00:21:17
the nominations come out the year before.
00:21:19
I don't think this is going to happen.
00:21:21
I think there is no way. I can't imagine the Recording
00:21:24
Academy getting behind AI in this way.
00:21:27
Tom, you're the Hollywood Whisperer, which is music
00:21:29
adjacent. What do you think?
00:21:31
I've never really reported the music industry.
00:21:33
It's so dirty. He he has the right idea, wrong
00:21:36
prediction. Like awards are the most
00:21:40
backward looking conservative groups out there.
00:21:43
They're the last people that would want to willingly accept
00:21:46
an anti artist, anti creativity technology and give it an award.
00:21:50
The only possibility I could see is like it gets revealed that a
00:21:54
song that everyone enjoyed, you know, had more AI composition
00:21:58
aspects of it. The thing that he should have
00:22:00
done, which I think is actually a decent prediction, is that an
00:22:03
AI generated song goes number one or like charts really highly
00:22:07
on the Spotify or Billboard charts that that I could totally
00:22:10
see happening. Well, songs have already
00:22:13
charted, just not #1 so. That's very.
00:22:16
Much in the role of possibility. Yeah, yeah.
00:22:18
And I already, you know, I love some of the AI songs out there.
00:22:21
Most of my music is AI generated these days.
00:22:24
I think it's it's, you know that that we are Charlie Kirk's.
00:22:27
Have you guys heard that one? Of course I've heard that one.
00:22:30
Yeah, yeah, that's a banger. And I think more to come.
00:22:34
It's taken over the nation, really.
00:22:37
Yeah, yeah, it's going to be sung at the next RNC.
00:22:40
The well that that is a big vacuum for AI art.
00:22:44
It's like Republicans who want to be pandered to but don't have
00:22:48
any artists in their camp desperate for somebody to make
00:22:53
the their their propaganda and I guess.
00:22:55
Joe Rogan is is is not campaigning.
00:22:58
He's, he likes it. He loves AI music.
00:23:00
He talks about right, enjoying like AI.
00:23:04
What is it like AI rap? Like sort of a a melding of rap
00:23:08
in Motown or something? Wow, a lot of the charting AI
00:23:12
music has been in is country western music or country pop, so
00:23:17
do with that what you will. Right.
00:23:20
I think it's very clear what's going on there.
00:23:24
Yeah. What's funny is that music
00:23:26
probably is not main. It's obviously mainstream, but
00:23:29
it's charting wise. I think it's been a while since
00:23:32
like a country song went number one.
00:23:34
Or usually it'll do it if it has like some hip hop.
00:23:37
Country songs go #1 all the time, but they're certainly
00:23:40
country pop songs. It's not, you know what we if
00:23:42
you're being a genre purist. No, it's basically pop with a
00:23:45
steel guitar and the right drum beat in the back.
00:23:49
Yeah, back to the prediction, I think much more likely that
00:23:52
there's an AI generated hit then it gets a Grammy nomination.
00:23:57
Like I I'm not convinced necessarily they'll be, you
00:24:00
know, a top ten hit this year, but I'm surprised he's going
00:24:05
with this thing that's gate kept by humans.
00:24:09
Yeah, yeah, agree, agree. And by the way, if you want to
00:24:13
lean into the dystopia of that an AI generated number one hit
00:24:16
that you know, that offends a lot of people.
00:24:19
That is kind of a concerning concept that we are being
00:24:21
entertained by our algorithms in that way.
00:24:24
But yeah, I agree he's slightly off the mark in terms of the the
00:24:28
body to to hang it on. Hey folks, I'm Omar, product and
00:24:33
design lead at Google DeepMind. Have you ever wanted to build an
00:24:36
app for yourself, your friends, or finally launched that side
00:24:39
project you've been dreaming about?
00:24:41
Now you can bring any idea to life, no coding background
00:24:43
required, with Gemini 3 in Google AI Studio.
00:24:47
It's called Vibe Coding and we're making it dead simple.
00:24:50
Just describe your app and Gemini will wire up the right
00:24:52
models for you so you can focus on your creative vision.
00:24:55
Head to AI dot studio slash build to create your first app.
00:24:59
Now it's time for the newcomer predictions.
00:25:01
Our house predictions now is the well thought out.
00:25:04
Accurate. Ones not just shitting on other
00:25:06
people's predictions, but yeah. Take out your pens now, right,
00:25:10
your AI pens you're. Saying, all right, Tom, what's
00:25:12
your prediction? I made a couple, but the ones
00:25:17
that we should probably dig into here is it's going to be a tough
00:25:20
year for open. AII think they are facing a ton
00:25:23
more competition. They've raised so much money in
00:25:25
the last year. And the thing that I keep
00:25:28
wondering about is like executive stability and they
00:25:31
have in the last 18 months brought on new CFO, new head of
00:25:35
products, Johnny, I've whatever you want to call that.
00:25:39
And some of these people are from the Sam Altman orbit, some
00:25:43
of them are just general Silicon Valley talent.
00:25:45
But that's a very chaotic environment to work at.
00:25:48
It's just historically, it's just built on, you know, like AI
00:25:52
don't know, like a Native American tribal ground or
00:25:54
something. There's just always a sense of
00:25:57
unrest and, and I don't know, darkness pervading Open AI Think
00:26:03
there's going to be a big shake up at some point there as they
00:26:07
start slowing down in terms of their growth as raising money
00:26:10
becomes more problematic. And you know, I don't want to
00:26:13
name names too much 'cause it's a maybe a little unfair to say
00:26:16
so and so is going to get fired this year.
00:26:18
But like if Open AI does struggle to raise more money and
00:26:23
they will need to raise money this year, there's no question
00:26:25
the finance team will be an interesting thing to.
00:26:27
Learn who's to blame. Who's the person who made a big
00:26:32
public speaking blunder? Yeah.
00:26:36
In 2025, who's the person who was the CEO of a company, but
00:26:40
then Open AI decided to bring in another former CEO.
00:26:46
Yeah, we're talking about we're not going to be too cute.
00:26:49
This is Sarah Fryer, CFO of Open AI, former Next Door CEO.
00:26:54
Tom, are you, are you predicting that?
00:26:56
Yeah, and and also not the CEO of Square, I just think she'll
00:26:59
be person to watch this year. I think if things are going
00:27:03
great for open AI and they can raise more rounds and their data
00:27:06
center deals keep going off without a hitch, then she's
00:27:09
probably solid. But I think if things do slow
00:27:13
down this year, that's just an area of extreme tension.
00:27:16
Though if Open AI can't fundraise, there's one person
00:27:19
responsible. That is Sam Altman.
00:27:21
He's sold the world on on that company.
00:27:24
You know he is their fundraising machine, but you know.
00:27:29
Yeah, it's hard to and, and look, Sarah Fryer is a very well
00:27:31
respected executive in Silicon Valley.
00:27:34
I mean, she really wants a ton of credibility to square when
00:27:37
she was there helping to bring that company public and, you
00:27:40
know, next door was problematic, but that company was
00:27:43
problematic. Like they were struggling in
00:27:45
general. So I don't want to say she's
00:27:47
like anything, but, you know, an example of the high quality of
00:27:52
talent they have at the executive level.
00:27:54
That said, you know, there's a lot of pressure on her and
00:27:57
there's often scapegoats when companies struggle.
00:28:01
And it just wouldn't surprise me if that is the area of focus.
00:28:04
Do you think Anthropic passes Opening eye in valuation in
00:28:08
2026? Well, I think that depends on
00:28:11
what this supposed round that was coming together at the end
00:28:14
of last year, the one that apparently it's at the $800
00:28:17
billion plus level. Like do I think in which
00:28:20
Anthropic I think is at like 3, like late 3 billions.
00:28:23
Like are they going to more than double their valuation this
00:28:26
year? I don't think so.
00:28:29
But I think Anthropic's credibility as a company, which
00:28:34
is like a non, you know, it's not like an objective metric,
00:28:37
but I think their credibility as a company could surpass open AI.
00:28:40
Yeah, I think people might not equity, that's not.
00:28:44
Yeah. Reputation.
00:28:45
OK. Let's put it like this.
00:28:47
That's a cool. Factor.
00:28:48
Yeah, which? Which way?
00:28:49
Incredibility. Well, they're definitely not
00:28:50
cool. They're the Dario company.
00:28:53
The guy was like stuffed animals everywhere.
00:28:55
I think that well, I don't know what's like the objective
00:28:58
metric. I could I could base this on if
00:29:00
his evaluation. No.
00:29:01
Is it like quality of model? They've already surpassed them,
00:29:04
right? I think.
00:29:07
You answered right, you're not predicting Anthropic to pass
00:29:09
open AI evaluation in 20. 20 No, but I do think the general.
00:29:13
Open AI has a huge consumer business which is can't take it
00:29:17
away. Right, right.
00:29:19
All right, Madeline prediction. All right.
00:29:21
My prediction, I think Defense tech which was really hot all of
00:29:26
this last year is going to stay hot this year.
00:29:29
But the first big wave of M&A of some of these companies will
00:29:33
start and I think the buyers will be more other start-ups
00:29:39
rather than the primes. I think there will be some
00:29:42
consolidation on the private side rather than like Lockheed
00:29:45
going in and buying some of these companies.
00:29:48
I do think that. You make boats, I make drones.
00:29:51
We both need a huge lobbying operation to win government
00:29:54
contracts. Let's.
00:29:56
Let's team up and then we can do the boat drone company together.
00:30:00
I think that is no question going to happen.
00:30:02
The piece that I'm skeptical on is if anderill is really going
00:30:07
to buy one of the other big players in the space because
00:30:10
they don't need to. So will they buy like one of
00:30:13
these one off you know, satellite or feature companies
00:30:16
to do one thing? Possibly, yes, I could see that
00:30:18
being where they become the acquirer, but I do.
00:30:20
Not you're betting. By I'm betting against them
00:30:23
buying like a saronic or like a. Shield or something they simply.
00:30:26
Don't need. To so instead they're going to
00:30:28
do it themselves and say, why are you being annoying like
00:30:30
we're we're the leader here. I do think also that if there is
00:30:35
a ceasefire in Ukraine, that could be a big slow down for the
00:30:40
market because that has been, you know, the main place where a
00:30:43
lot of this tech is being tested and where people are getting
00:30:46
contracts to do experimental, you know, work there.
00:30:49
However, if we go to war with Venezuela this year, then, you
00:30:53
know, new markets door wide open.
00:30:56
So I don't think that necessarily just Ukraine winding
00:31:00
down would lead to to a slow down in the space anymore, I
00:31:04
thought. You were going to say.
00:31:06
Taiwan. I thought you were going.
00:31:07
To say with China or Greenland, yeah.
00:31:09
Well, I was going to say there's so much more, you know, as as we
00:31:12
are the American hemisphere and launching our, you know,
00:31:15
missions, the the defense tech market only grows so.
00:31:19
Yeah, the drum beat for war escalates in, I don't know,
00:31:24
Northern Europe and also Central America and also northern South
00:31:27
America. Is is Andrew going to go public,
00:31:29
you think? That I think they could.
00:31:32
I think it's possible. I'm not sure if it's going in
00:31:33
2026, it might be early 27. It depends on the market.
00:31:37
They've you know, done some looking into that.
00:31:40
But I don't think I'm skeptical that it actually happens.
00:31:43
Feels like we'd hear the buzz, you know?
00:31:45
Yeah, I think, you know, there's always, you know, the data
00:31:48
bricks going public conversation.
00:31:50
If we're talking about equally huge DECA corns and data bricks,
00:31:54
seems to me like one that would be a 27 at least and not 26.
00:31:58
So I'm going to stick with that for Anderel as well.
00:32:01
All right. I'll end with my prediction.
00:32:06
I have a couple now that I'm tying up the loose ends of
00:32:09
everything that people didn't see coming in 2026.
00:32:13
I mean I I think macro bullish on AIAI going strong still in
00:32:23
2026. I think 2 specific bad things
00:32:27
happen because AI is going strong.
00:32:30
One, I think we're going to get like a huge international hoax
00:32:35
enabled by AII. Don't know if you saw the piece,
00:32:39
Casey Newton in platformer. There was this viral screenshot
00:32:43
that seemed to be about DoorDash or Uber Eats from a worker, sort
00:32:47
of a white collar worker saying, you know, oh, we're doing all
00:32:51
these terrible things. We're giving workers a
00:32:54
desperation score. And then Casey Newton reached
00:32:57
out to the person who posted that to try and get
00:32:59
substantiating documents. And the person apparently sent
00:33:02
him like a whole report. And then it turned out the whole
00:33:06
thing was fake and it was basically an attempt to
00:33:09
substantiate what was, you know, basically an AI generated thing.
00:33:13
And now you can create an AI generated report to try and
00:33:17
substantiate it in in the business.
00:33:20
Reporters would call this like rat fucking when people go to a
00:33:22
sort of elaborate lengths to try and con reporters and whether
00:33:26
whether it's a reporter or not. I mean, I think we're going to
00:33:29
see a huge international hoax. I mean, I think maybe I'll go so
00:33:34
far as to say major media outlets sends a push
00:33:37
notification about something that's not true.
00:33:42
I don't know. What do you guys think about
00:33:43
that? One, I buy that if people fell
00:33:45
for the, you know, Balenciaga Pope, which frankly I did, then
00:33:49
I think we're going to see, you know, something much more
00:33:51
nefarious than that if that was already, you know, like a year
00:33:54
ago. Yeah, I wonder how long that,
00:33:56
like the, the DoorDash thing was interesting because it went on
00:34:00
for a decently long amount of time.
00:34:02
Like seeing AI generated or misleading posts on social media
00:34:06
is like just the fact of life right now.
00:34:10
It's very difficult to spend any time, certainly on Twitter or X
00:34:13
without seeing bullshit on there.
00:34:15
So I'm, I'm, I'm wondering what, Yeah, like an article will be
00:34:18
written at a major news outlet. You think that will be based on
00:34:22
an AI generated hoax? That's like where you're.
00:34:24
Going, I'm saying more of AI said a push notification.
00:34:27
Because I think it's more plausible that it's like
00:34:29
somebody. Becomes the.
00:34:30
Aggravator that it's like we got we got to move on or you know,
00:34:33
it's like, yeah, Business Insider sends a it's on the
00:34:36
break in this thing just happened breaking, you know,
00:34:38
before before it really gets metabolized, you know.
00:34:42
Yeah, I think it's totally possible.
00:34:44
We're already there. The technology is there.
00:34:46
It's just about like some bad, you know, bad judgement on an
00:34:50
editor's part at a big news outlet to not do the work.
00:34:54
And they're overworked. They're trying to beat the news
00:34:56
cycle. Yeah.
00:34:59
Inevitable. Sure.
00:35:00
I buy it. What?
00:35:01
What industry will it be about? Will it be about, like,
00:35:04
politics? Will it be war?
00:35:06
Will it be about like someone will say well.
00:35:08
Somebody's dead, some video, some fake video.
00:35:11
You know, I don't. Yeah, there are lots of options.
00:35:14
The other one is, and this would be more sophisticated, but like
00:35:18
an AI hack, like a basically a sort of technique enabled by
00:35:23
powered by, you know, AI, Either that's like, you know, scaled
00:35:29
phishing or, you know, there are, there are lots of lots of
00:35:32
things, but a big cyber problem because of AI in 2026.
00:35:38
Yeah. I mean, you know, we got a bunch
00:35:40
of a lot of VC firms like to send us their partners
00:35:42
predictions at the end of the year.
00:35:44
And and that actually was one from Rama Sikar at Menlo
00:35:48
basically saying there was going to be a self propagating attack
00:35:51
where an AI agent would autonomously, autonomously
00:35:54
exploit vulnerabilities in other agents.
00:35:57
That's kind of a more agent specific one, but it fits into
00:36:02
what you're describing. And as agents become more
00:36:05
prevalent and more business processes depend on them and
00:36:10
there's a vulnerability there, would make sense that agents.
00:36:12
Agents are talking to agents. And how we're implementing AI is
00:36:19
creating its own cybersecurity vulnerabilities, like everybody
00:36:21
giving clawed terminal access to our computers and all these
00:36:25
agents, you know, you're giving pretty deep access to startups.
00:36:31
Well, everyone's giving ChatGPT and Claude their health
00:36:34
information constantly, so that's a whole other can of
00:36:37
worms. But lots of personal data is
00:36:40
going in. Right.
00:36:41
The the risk is just we're exposing all this very sensitive
00:36:44
data much more cavalierly because of AI.
00:36:48
I'd actually be interested. Have you ever tried to ask
00:36:51
ChatGPT? Like if you were to guess my
00:36:53
passwords based on my information.
00:36:55
That's what you said. Never asked that.
00:36:56
Yeah, I'm. Gonna try that.
00:36:58
Yeah, that's good. That's a good idea.
00:36:59
Or at least your security questions, you know.
00:37:02
My last prediction. I'm curious what you guys think.
00:37:04
This is unrelated to the AI ones.
00:37:06
I I think we see another big Elon world merger like right now
00:37:11
with Grok Elon. They're out there bragging about
00:37:15
how many users X has as if that sort of, you know, their their
00:37:19
valuation per user is so much better than Anthropic, which is
00:37:23
absolutely insert. But he's seen a lot of this like
00:37:27
synergy of hype by having X and, and Grok part of the same
00:37:31
company. He needs to figure out something
00:37:34
to save, you know, Tesla. I I think it's very believable
00:37:39
that, you know, X and Tesla merge in 2020.
00:37:44
Six depends I sure Elon mergers make a lot of sense.
00:37:49
People have expected Tesla to yeah, I I think that's like on
00:37:54
people's expectation list. My only issue with that is like
00:37:58
Tesla's increasingly A robotics game.
00:38:01
Not necessarily. But that's robotics is AI
00:38:03
powered, you know they're. I yeah, sure it is.
00:38:06
Yes. It's all it's, it fits into the
00:38:08
same narrative. I think Grok is an LLM versus,
00:38:11
you know, these are, you know, real world robotics.
00:38:13
But it's an easy sell to investors.
00:38:16
The only problem is that it's one thing to merge X and XAI,
00:38:21
which had a lot of overlapping investors, and it was a private
00:38:23
market. Like you've got to get
00:38:25
shareholders on board with a Tesla thing and they're already
00:38:28
making a fairly thin case for valuation.
00:38:32
So true. It's kind of like it depends on
00:38:35
Elon's like juice this year. I'm sure he wants it.
00:38:40
Makes sense? But he doesn't want all his
00:38:42
companies public either, so. Yeah, I like it.
00:38:45
An Elon merger. I like it.
00:38:46
What about other ones? What about like The Boring
00:38:48
Company or or? Right.
00:38:50
Well, yeah, there's so many. Neural link.
00:38:53
Right, yeah, you can get grok and neural link so you can chat
00:38:57
in. Inside your head.
00:38:58
Exactly. Exactly.
00:38:59
Or boring coming in neural link like how do you borrow into
00:39:01
people's? Head you bore into your head.
00:39:03
Yeah. Those drills, they get big and
00:39:06
small. Yeah, the small micro drills.
00:39:08
You were thinking too big, Elon. It's not about tunnels under
00:39:12
Vegas. This is about tunneling into
00:39:14
your brain. Midterm elections are this year.
00:39:17
Oh yes. All right, it's going to be a
00:39:19
crazy, I mean, yeah, yeah, that alone will make it a crazy year,
00:39:22
right? I mean, Trump's like, if I lose,
00:39:24
I get impeached. It's going to be, you know, I, I
00:39:28
think in the business world, the big question is just like, does
00:39:30
the AI juice keep pumping? I I say yes in 2026.
00:39:36
Do you guys, you think NVIDIA is up at the end of the year or
00:39:40
down? I think that's that's the
00:39:42
ultimate question. What do you think I say up?
00:39:44
I think NVIDIA up because they're still making sales.
00:39:47
There's still so many people buying their chips that they
00:39:49
kill quarter by quarter. They've never really had a bad
00:39:52
quarter. It's all still insane.
00:39:55
I think 26 might be the last year.
00:39:57
Yeah, I think, I think it's all up this year.
00:39:59
I don't. I think we'll see in 2027 if you
00:40:02
know, the rubber hits the road around.
00:40:05
Exits for some of these more private companies and if people
00:40:08
would actually get acquired, but I think for NVIDIA being up at
00:40:12
the end of 26, yes. Another roaring market, and then
00:40:15
when the poor Democrats maybe sweep into power for next year
00:40:19
and the market and everybody unravels.
00:40:23
Right. Things were fine when when when
00:40:24
Republicans were in control of Congress.
00:40:26
But now? And now everything's turned.
00:40:29
The only push back I'd have on the market being up for AI is
00:40:33
I'm not so sure about about data center right in the neo, the neo
00:40:37
clouds. Yeah, I think that's very
00:40:40
sketchy. I think the debt markets will be
00:40:42
the thing I'm actually paying some of the most attention to
00:40:45
this year because these companies, I mean, core weave is
00:40:48
already, you know, filing these very concerning sounding filings
00:40:52
about, you know, the their need to a consolidate debt and, and
00:40:55
the changing rates and things like that.
00:40:57
And we're seeing already a lot of the lenders jacking up rates
00:41:02
on neo clouds because of their uncertainty of, of the of the
00:41:06
market. So I.
00:41:07
Think it's still worth $38.8 billion, up 95%?
00:41:13
Yeah, it was. It's still a from its IPO.
00:41:16
It's been down a lot recently, but you know, you've got Core
00:41:19
Weave down, you've got Oracle down.
00:41:22
You know, there are some other interesting companies that we've
00:41:24
talked about in that world that would be really good indicators
00:41:29
and they'll be the first ones I think to start seeing problems.
00:41:32
All right, space to watch Neo clouds, red flags from Tom.
00:41:37
Cool. All right, well, that's our
00:41:38
show. Thanks so much.
00:41:40
We'll see you in 2027. Yeah, see.
00:41:42
You next year, so you. Next.
00:41:44
Come back next week, please. We'll be back, I promise.
00:41:46
Yeah. Bye, see.
00:41:49
Ya. Thank you for tuning in to this
00:41:50
week's episode of the podcast. If you're new here, please like
00:41:53
and subscribe. It really helps out the channel.
00:41:55
Listen in for new episodes every week, wherever you get your
00:41:57
podcasts.