Projects
![]()
No projects found
Investors
New
DAO Treasuries
DATs
Lend / Borrow APRs
Chains
Categories
Spaces
The year ahead 2026
Recorded: Dec. 31, 2025 Duration: 1:33:40Space Recording
Short Summary
In a dynamic discussion, participants explored significant trends and growth opportunities in the crypto and tech sectors, including the impact of AI on software development, the implications of geopolitical tensions, and the evolving landscape of project launches and market fluctuations.
Full Transcription
Thank you. Thank you. Man, how times have changed, I will say this much.
I remember last year, I organized about a week in advance, a whole bunch of spaces,
so that people would kind of like be together, and we'd have a bit of a crossover event
to try and boost listener numbers as much as like possible. I
think it was no sorry it was a year before last year. Last year was an
interesting kind of introduction into the world of narrative engineering when
you're at the forefront of it right and so this was kind of like what my year
was about so far. I'm gonna deal with with, I'm going to like start with a little bit of introspection first and foremost, right?
Because this year for me has largely been one of finance, I would say, right?
It was one where I had to deal with the reality of the situation,
that time was money and that I was effectively spending a lot of my time in a manner that wasn't productive, that I was effectively wasting it.
One of the solutions to this, one of the best solutions to this, was to say, well, there's
lots of people who actually care and would like to contribute to kind of like, you know,
the general mission that I have and be part of it in some ways as well.
And so I started, of course, you know, the Noetic Order which I did I think in the later half of the
year of this year. And it's been growing, it's very successful, it's very nice.
My most dedicated following is in there which is very nice. I see some of
the people here even as a matter of fact in this space as well so that's pretty
cool. The advanced members subscription has enabled me to basically become independent of X in terms of revenue.
Because at the moment, X's revenue is exceedingly subpar due to one very particular reason.
I refuse to post politics.
I refuse.
And I've done so since February. I went to New Zealand in February,
right? And since then, people could kind of point to like a moment where things shifted,
where I decided I'm no longer doing anything even closely related to politics at all.
Because I realized that it was cringe and that it didn't matter. And no matter what you did,
no matter what you would think of or speak about, if it has anything to do with politics,
it's almost like sports. It's irrelevant. It doesn't mean anything. It's all just about
narrative engineering. Because at the end of the day, the only reason why you'd ever have a
political affiliation of any kind is because you're getting something out of it. And most of
the time, what you're getting out of it is maybe some sort of leniency, right? Let's say, for instance, I have to kind
of kiss ass of a certain administration so that they give me some sort of benefits.
These benefits can be anything. It doesn't really matter, you know, if they're worth it to you,
they're worth it to you. And you're going to do what it takes to get what it is that you need.
And so this is kind of like what I see the current administration is that that's that's the
that's the current meme you saw even companies ask us very differently uh you know about two years
ago and now they do so today in a manner where it's like wow this is very uncharacteristic or
two years ago isn't it right so all of these organizations and all of these entities,
they all serve their own interests as they actually should.
This is the thing that people often don't understand,
is that each and every single one of the entities are entities of their own
because they must be, because they have to exist somehow.
So of course their own interests must be served,
and they will serve their own interests regardless of what happens,
because they have to. So it's even understandable in some ways it's the good
pr move right um can't really shade them as much for that it just depends on how far off the deep
end they go with exactly this and i and ideally you would like to create something that's
independent of these uh types of you know like influences But the bigger you get, the harder that becomes. And the more you
become dependent on these kinds of things, especially if you're becoming very integrated
into a nation, let's say, for instance, that could be as any form of company or like some sort of
governance structure or whatever it be, right? So the choices are endless for you, all of these
paths. Either way, I decided to stay away from that.
It was an interesting time because prior to even January of this year, there was an offer that was
given by someone, and I can't name whom or what the nature of it was, but I rejected it, and that
was part of the shift away from politics that I already had because I just didn't like it. And I wanted to focus on
something that's much more constructive, like technology, manufacturing, perhaps even matters
of the soul and the spirit and God in some ways, right? A lot of people have deviated away from
that. They just look at the thing and they say, hey, I look at the iconography and therefore I
worship. And it's like, no, it is a lot more deeper than that. And you have to understand this, right? Even if you were to look at the iconography, you only ever see it's like no is is a lot more deeper than that and you have
to understand this right even if you were to look at the iconography you only ever see the very
surface of it you don't even see the rest of it you don't see the depth of it it's like an iceberg
man there's so much depth to it there are mathematics and and it gets lost on the masses
that think that they're worshiping right and so it's things like that. And this has largely contributed to a more sane approach to what I've been doing for about the last two to three years
up until now. And one of the sane things was, how quantitative can you be with the things
that you're doing? And this will be very important for next year, or like the coming year,
which is going to cross over for some of us fairly soon.
UTC 14, it has already crossed over.
That's why on the Timeless Martian, I have posted that the 2025 is completed, right?
So I'm going to be posting on all the main accounts and wherever else
as soon as like various time zones hit, including mine here.
So it's like, for those,
it's already gone.
And for me, the new year is coming
in like 12 hours, basically.
In 12 hours, I would have already crossed over
into 2026, because I live in Oceania,
which I can say now, because earlier this year,
I got doxxed next level, right?
in some ways, that was also a very freeing event, like I
keep talking about this sometimes, as people thought that was a negative thing, and it's like,
no, that was actually one of the best things that had ever happened, because everything that led up
to it was, it just kind of fell into place. I don't know why, it just kind of did, and it was great,
you know? I happened to be in the right places and had
absolutely the dumbest of people make assumptions that were so obviously disproven and then they
just like kept rolling with these assumptions and then in one fell swoop by the very entities that
they kept supporting they were undone i was like that's fucking hilarious like these people would
be shilling for all of these like like, legacy media institutions, for instance,
who'd make up total bullshit about, like, Elon's companies, for example.
And they would just try to hold that over everybody's head
whilst you're just, like, sitting there trying to quantitatively undo their bullshit,
going like, hey, that's actually not how any of that works, here's exactly why.
And then it's just like, this is fun content for me
because I like doing analysis of things, and I'm really good at it.
And so I like taking those fools apart.
It's a lot of fun, right?
Which also, of course, left behind a bit of an image, as I'm sure many of you are aware, right?
Because it's like, how does this one guy know so much about this particular topic?
Well, you just fucking read, man.
Everything's public, dude.
You just fucking start.
Even before AI was always there, it was probably even easier to get that kind of knowledge before AI
because indexing was more like,
like it didn't have as much bloat,
you know?
So like,
then these same entities kind of like turn on them
and just destroy the whole narrative.
And they just sit there and just go,
well,
what the fuck do we do now?
The answer is you do nothing and you just wait out and see what happens next.
And there's really nothing that's going to be done.
Because the problem is back then when they made a political argument, for instance, which they're trying to do today,
they would try to appeal to some sort of base, some sort of higher party affiliate of some sort,
to kind of like, you know, have a strikedown occur.
That doesn't exist, because that's not
how any of that works, right? If your political party's in power, it will work. But if there is
one that's in power that's against you, it won't work, right? But that's also why 2026 is going to
be important, because make no mistake, and look, I've made a lot of unpopular takes in the beginning
of the year, so here's another unpopular take for you. Especially one that I said before about,
like, you know, platforming and such, such you know people got mega mad at me for
like saying certain things they're like oh yeah all these fucking people that don't know what
they're talking about what i like to call the legacy maga all that shit it was like a big thing
it's like oh my god how dare he say this i'm like what do you mean how dare it's fucking i could just
say shit bro what's wrong with you um but like the midterms are going to be a disaster for the Republicans, I'm pretty sure.
Because a lot of people are very dissatisfied with what's happening.
And there's lots of narrative engineering where people say,
no, that's not true, that's not everybody who thinks this way.
I will say, however, it's going to be quite close for them.
There are going to be different interpretations of history.
That will be interesting, that both sides will need to somehow reconcile with.
Everything is just going to be kind of chaotic.
Like even the big football games are coming up.
All of that shit is going to be really interesting to see because it's all going to be political.
So this coming year will be very challenging for someone like myself because I'm going to go largely unnoticed, I think, even.
Because that's kind of part of the...
Maybe I'll become noticed in that I'm the only abstraction that's left.
Because I'm not going to be part of the bullshit.
But everyone's going to play some sort of game.
And my game is just like, hey, I'm going to observe what all you guys are doing and analyze the shit out of it.
And just talk about it and have fun fun and see what other people have interest.
Say, do you have interest in my assessment of the situation, how I monitor it, and do you see that as valuable?
So that's kind of what we're doing at the moment.
That's kind of what we're doing with a noetic order as well.
And this is how I decide to beat the whole political problem because there's
going to be a ton of narrative engineering make no mistake there will also be um
so because one of the major problems right now at the moment is how trump is very transactional
and the way that that impacts the world is very volatile in nature.
And so this will create kind of like this technological, somewhat geopolitical cold war that just is kind of felt but barely perceived.
Like you'll perceive it with your wallet more than anything, because you'll just be wondering, why the fuck has this gone up?
Why the fuck is this no longer a thing? Or like, wait, why is this cheaper all the time? Something like that. You'll just go, what's going on here? You know, there'll be lots of fluctuations. There'll also be plays with the Fed as well, because, you know, Jerome Powell is going to step down. He's still going to be on the board, though. Right? He's not going to be chair. He's going to step down as Fed chair, obviously. Right? That's going to be out his term ends. The Fed itself
will try to secure itself as much as possible, so that even if Trump were to put a loyalist
in there, you can't fuck around with the interest rates so much, which is going to be one of
the major things that he'll try to do because of the fact that, well, you have to somehow
control what's happening on an almost optic level
like, more so optical level
so that when, say, the term
is over, you can kind of like
let everything bounce back and rebalance itself
which isn't exactly pretty
but that's what's happening. Nothing that anyone
can do about it
because the fuck you're gonna do actually
real talk is nothing that's gonna happen
so it's like, that's happening.
China itself is becoming exceedingly aggressive, which is annoying.
Because you had this initial assessment going into the year that they were going to continue with their asymmetric form of warfare or manufacturing.
Their computers are going to get better, by the way. Much better.
I think they already have have to a certain extent. They've reached RTX 4060 levels of advancement, I think. So they're about two and a half years
behind. But that's pretty good. That's pretty damn fucking good. Only two years behind? That's
pretty good. So they're going to give some serious competition. They're also going to try and innovate
on the memory market as well, because they'll have to, right? Because everything right now is in extreme demand. I'm sure everybody has seen what's been happening with RAM, with just NVMe SSDs, all that kind of good stuff.
Micron, for instance, you know, they just decide, hey, fuck it, I guess we're going to give up all the consumer markets, and we're going to focus all on AI.
So one of the major things that I always disagreed with was the idea that the AI bubble would pop.
And the reason why I disagreed with it is not that I don't think it will pop, it's that I don't think it's a bubble, but it's more so a foam that's going to disappear if enough wind comes onto it, right?
Like we've all seen what happens when you like step out of the shower, and you're covered in a whole bunch of like, you know wind comes onto it, right? Like we've all seen what happens
when you like step out of the shower and you're covered in a whole bunch of like, you know, foam
and such, right? Or like you come out of your bath and it just kind of like disappears all the
time because all the little bubbles are popping. That's kind of how I see AI. It's kind of like a
foam. It's very stable actually because of this. And it explodes into many little tinier bubbles
that are each more difficult to pop than the last one that did. And so it just kind of will remain for quite a while until all the money runs out and the
wind just blows it away.
So until then, things will just kind of stay as they are, which is fine.
Like, I don't really see, I don't really see like that being a negative.
I don't really see that being a negative.
I think like if there's more demand on AI and perhaps like maybe compute, it's like it'll drive innovation both in like say America and in China because like
memory right now is a really short supply, dude. It's crazy. Like at the beginning of the year
with X revenue, I bought myself a computer. It's like it has a 5090 in it, has 128 gigabytes of RAM, and
yeah, it's got a nice CPU as well. I bought that all from the X revenue, which
was really nice, and yeah, it was a solid investment. It went up by like, what is
this, like 300% or some shit?
It was really good.
But yeah, it's pretty neat.
What the fuck just happened with a listener account?
Oh, hi, Elon.
Oh, no.
That explains everything.
I was just like, wait a minute.
I'm not that relevant.
There's no way.
Hang on, let me bring up Elliot real quick.
And I think...
Fuck it, I'll give you an invite as well if you'd like to join up.
Blythe, Elliot, can you hear me?
Absolutely.
Can you hear me?
Yes.
Yes.
We hear.
Got any compute thoughts yourself there, my man?
You're the one who's optimizing all the CUDA kernels.
I have a lot of takes on this, but it'll take a few minutes to get through that.
Sure, if I could go for it, man. I'd like a little bit of expansion.
What's up, guys?
Yeah, basically, I don't know. It's really tough to see.
I kind of now regret making all these resources and stuff on books
and teaching kernel programming,
because I don't even think it'll be relevant in a year.
By the time people are learning this stuff, it won't even make sense,
because through evolution or any other causes,
it'll simply just dissipate.
At least performance optimization tasks
are very objective to optimize for.
So I don't know.
This is something that I've kind of integrated into everything
that I do on a day-to-day basis is just like yeah don't learn any of this or optimize by hand it doesn't make sense
real you know what's up man a long time no see
or like here rather, because, you know, spaces.
Hello, other me.
Hi, other me.
That's a good one.
Yeah.
So I've seen your year has been quite the adventure.
There's been a lot.
Yeah, it's been quite a year. I think 2026 is going to be a real banger year.
Indeed, indeed.
I think it's going to be lots of narrative engineering, personally.
Because the midterms are coming up, and it's always a major thing right about that period.
Anything that's political, you've got tons of competition.
So there's lots of narrative engineering.
It should be really entertaining.
I'm going to stay away from it, largely i don't i don't deal with that as much uh also
because i got a ton of work now yeah what's your work sorry come again what work oh manufacturing
stuff okay it's what i've like done i kind of don't want to talk about it it's not entirely
relevant it's just like it's kind of like a luxury product
type thing. It's not that high up.
It's quite simple. I just don't want to talk
about it too much because I don't want to bring attention to
those
people out of
I don't want any harm to come to them
type of deal. You know what I mean? So I just don't
talk about that as much.
Okay. Doing some automation
stuff now. It's pretty fun.
Well, Tesla
should have
a widespread
robotaxi.
That'll be a big thing for Tesla.
Optimus 3
will launch, and then
hopefully SpaceX will achieve full reusability
with Starship. Those are the
pretty giant ones. Those are the pretty
giant ones.
I assume the first major
shipments with Starship
are just going to be like Starlink satellites, right?
Yeah.
Yep.
And then we're going to go to the moon.
Oh yeah, definitely.
The space computer
thing is like a really good accelerant, I think.
So you become like the...
Like SpaceX becomes the major delivery company of choice then, right?
Yeah.
Biggest, most reliable rocket.
I got to ask.
I got to ask.
Have you ever thought of what it's like to manufacture things on the Moon?
Because you can make different types of materials that you can't make on Earth in low-gravity
types of environments.
Well, I think the biggest option on the Moon is to actually make solar cells and radiators
so that you're manufacturing on the moon
anything that weighs a lot. Chips
maybe still come from Earth, but they
weigh very little.
And then you can use a mass driver to
put a billion tons of AI powered satellites
into orbit per year.
Mass driver basically being like
a kind of rail gun. I just like a rail gun.
Sounds cleaner.
Like, if you were on Dyson Spheres before, pivot to this.
Well, this will create a
Dyson swarm where essentially a bunch of satellites,
intelligence satellites around the sun.
Elon,
do you imagine we still need
gravity, even lower
Earth, Moon gravity for manufacturing
could this be just done in orbit?
you need mass
mass must come from somewhere
yes, the resources
you just need
you need a lot of tonnage
will there be a lot
of boring going on the moon,
you reckon? Or would it just be surface structures?
I'd say going underground would make more
sense. They have big lava caves down there.
Could be quite neat.
Sure.
We'll figure it out. Most important thing is
to get serious tonnage to the moon
in order to send even
way more serious tonnage to the moon in order to send even way more serious tonnage from the moon.
But you could scale to 100 terawatts of AI compute per year from the moon.
Indeed, indeed. Have you ever considered anything about magnets?
Big boy numbers.
Have you considered anything about magnet shields by any chance?
Because I've been looking into that a little bit, because for certain environments,
you'd want to, like, shield it
against, like, the
solar wind, basically. And so
there's, like, you can make super
conducting magnets, and that
would, like, shield against even, like, small objects
as well, like high-velocity objects.
No, it'd be fine.
You think so?
Okay. We already have 9,000
satellites in orbit,
so we know what it's like being in space.
I randomly saw your chat.
I've got to head back to Tesla work meetings.
Nice, man.
Well, thanks for coming, dude.
It's a pleasure as always.
Appreciate you so much.
See you, man.
Have fun.
Good luck out there dude alright bye
take care
big surprises eh
I did not expect that
nah man we need more than terawatts
real
I bet like everybody's notification
like notifications lit up
like a Christmas tree
it's all in the time too like notifications lit up like a Christmas tree. Yep. And that's why I saw so many people.
It's all in the time notes too.
Nice.
I'm going to bring a few people up.
Hey, Omar.
Hey, Alex.
How is it?
Hey, man.
Happy New Year.
Happy New Year to you too, man.
Well, it's almost New Year.
It's technically like in 12 hours we're going to do a crossover.
So it's going to be quite nice.
For me, everything wraps a little earlier.
So, like, yeah, I get to have the numbers a little early.
Got any outlooks yourself?
Got any outlooks yourself?
Yeah, I mean, that was pretty interesting what Elon was talking about.
I do think 2026 is going to be a crazy year.
AI is just taking off.
You know, the space-based data center thing is really interesting.
Scaling out robo-taxis.
Yeah, 2026, I think, is going to be a really transformational
year you know something that i've thought of is how insane it is to have something and i don't
i don't know it i don't associate models to the company really it's i'd like to de-associate those
but opus 4.5 is insane and that's not running on
that many watts in compute i'm just thinking like what is that world gonna look like when we have
like 10 or 100 or 1000 times as many watts going into coding right that that world like i don't
think anyone can actually understand what that looks like. Yeah, Opus 4.5 is really something.
I mean, who has the best model seems to be kind of moving around
like a game of musical chairs,
but I've definitely been using Opus 4.5 a lot.
And I mean, the way it just changes your workflow
where you can just type in an idea,
it just implements the whole thing
and you just deploy it and it just works. You get some error, you just
copy-paste the error into cloud code and it just fixes it.
You don't even have to know what the hell happened.
This is getting from, and right now it's still kind of at the stage where
you have to kind of know what you're doing to get the full power out of it.
But pretty soon I think it's going to be smart enough that it can just sort of run on its own.
And imagine what the GDP looks like when you don't have to be a billionaire to have a team of a thousand programmers working for you.
Anybody could have it for a trivial amount.
The economy is just going to go berserk.
Oh yeah, another little thing that I want to add
with the computer problem is,
also like why I think it's actually good
when the demand goes up,
is because it's going to become more expensive
to earn a computer.
And I don't think that that's necessarily a good thing
for a lot of people.
I get the pain that this brings.
But what I think it will also bring
is much needed software optimization. So like a lot of new software I get the pain that this brings. But what I think it will also bring is much needed
software optimization. So like a lot of new software that's coming out is like very, very
shittily optimized. For instance, like we're running basically upscaled versions of like old
game engines, like for instance, with Halo, or like we've got something like, like Call of Duty,
they all use like very, very old engines, and they're pivoting to something that's much more
new, like say, hey, look, say, Unreal Engine unreal engine for instance and one of those um like like like a lot of people have problems with unreal
engine games because they're largely used like the engines are used to generate slop
and those games are of course very very badly optimized but i've recently played an unreal
engine game that's exceedingly well optimized where like you're sitting there and with my computer and get like fucking 500 frames uh if i don't limit it and i don't run any other
background processes and that's our creators right i don't really like the game too much
but it isn't it's an extraction shooter you know it's what it is it's very gripping it's exceedingly
addicting i will say that um it's fun funny moments if you know what you're doing it's really
fucking hilarious um but it's very well optimized i'll say that uh and i think If you know what you're doing, it's really fucking hilarious. But it's very well optimized, I'll say that.
And I think that because things are going to get more expensive,
you'll see more optimization.
And if you can't do the optimization, you'll just kind of fail as a company.
And I think that's going to be Microsoft's greatest challenge, actually,
because on the front of optimization, things are being more lightweight.
Yeah, it's crazy what's happening to RAM prices.
Linux is really good.
You know, NVIDIA is just eating up all the RAM.
It's like $900 for a stick of RAM now.
It's pretty nuts.
But, you know, honestly, in the future, you might not need a device.
The user interface might just be like a phone line, right?
You can have a dumb device that just connects to this super powerful AI,
and the AI keeps getting smarter and more powerful,
and your device can just stay pretty dumb.
Like you kind of think about Pluribus.
I don't know if you've seen that show on Apple TV.
That's kind of like what super intelligence is like.
How does she communicate with them to get anything done?
She just kind of picks up her phone.
So maybe we don't need super powerful edge nodes anymore.
So I'm going to still do for the latency problem.
I don't want all compute to be cloud-based.
I'm not into that.
I think a lot of the compute still needs to happen on the ground
purely because of latency issues.
Maybe at some point we'll fix that too, but it's, I mean, have you ever like actually played on the cloud?
It's not really nice.
I've tried it myself.
It's very, it's very problematic.
It's easier to have my own compute.
That's nice.
But also, hey, actually, quick question.
Have you invested in Micron?
No, I don't own their stock.
You know, hardware's pretty volatile, I feel like.
It can be susceptible to natural disasters and shortages and things like that.
But they're an interesting company, for sure.
Dude, when I saw that Micron is the only one that actually,
not really the only ones, but Micron, SK Hynix, Samsung,
when I saw what was happening with
the ram because like elliot brought this brought this to our attention it's like hey guys everybody's
like sucking up all the ram and then he was like taking out his nvmes and shit too that's like hey
by the way we could use this as ram too and i'm like oh no we are no this is like we are you can
do this like but basically okay can i give an insane take on like
how you actually dodge the ram prices um you you basically do what grok is doing like this is this
is not the only reason nvidia acquired grok but you have these massive uh essentially roc with a q
yeah you you do the like the grok's rebrass thing you put a shit ton of astram on the chip
it's more expensive but it's very very, very fast. So economically, it makes sense. And your users are definitely going to pay for that because literally the
tokens come out at like thousands of tokens a second per user. It's insane as a user experiencing
that, like in your coding workflow or something, people are going to pay for this. And then when
it comes to loading very, very big mixture of experts models, right? Like Kimmy K2 thinking,
for example, 1 trillion parameters,
right? Only 32 billion of those are active, but you still need to load those onto the SRAM.
So the way around this is to basically use an adapter that says, oh, we're going to take
four different NVMe SSDs, which are very fast, and we're basically going to stripe them and parallelize across all of
them so if you have a very very high quality NVMe SSD that's like say 15 gigs a second um you put
four of those on an adapter you can get 60 right that's approximately the speed of DDR5 now keep
in mind that is for sequential reads and writes, not random
reads and writes. So this is an inference thing only. It's very important you understand what
type of memory access is actually happening when you're doing inference or else you don't get any
of that performance. So like if you're listening right now, don't go binge on SSDs. You actually
have to understand how inference works down to
the literal byte level to make this work properly this is a thing i've invested in to learn um
and to my surprise it's you know worked decently um of course it's hard to write these sort of
things from scratch and see but you can in fact pull it off. Small problem though is that the same,
well, very similar manufacturing lines
that are used for DRAM,
it's also like the NVMe product.
And so now they're converting a lot of those lines
into RAM as well, just for RAM fab.
So you're not really dodging the price
because the value is going to go up for that as well
because they're you know they're making these things and flash ships so it's really nice
i've gotten four terabytes of ram for some i don't know definitely not as much as a dodge ram
oh yeah no you you start you started early i saw your fucking post like you started early that's
why i was like okay we're cooked and that's that's why i like asked omar here i was like hey
why don't we just like what about what about like what about like micron i saw that
i was like hey let's back up on micron and i was i wasn't gonna like back up on silver as well
but it was just like you know what fuck it silver is like it's just kind of weird we don't understand
it we're very speculative about it and it's not really like a good thing it's like unless you
have like pure quantitative reasoning where you go like yes this is just great and this will just perform because there's
really little that will speak
against it, then it's like, well,
you don't go into it.
With Microns, it's like, this is the
logical path forward.
So, yeah. And it worked
well. I had better performance than fucking silver
when
the assessment was made. So, winnings
are still all around,
and now you're basically kind of part of it.
And I think it's just going to go up a little more
until eventually, like, stabilizes.
But we haven't even gotten to the point
of launching computers into space yet,
so I don't really see that decreasing over time
as much as people would think.
Hence more, like, the skepticism I have on the bubble idea.
I don't like the whole bubble idea as much. It's more like foam so oh you do not have to raise your hands just like interject you don't
let me interject man you just talk you can in fact interject dude you just gotta learn how to do it
okay it's like when it's like when you shuffle a deck of cards you know the card flips in for
uh hallmark's catalog or omar as as adrian, I didn't actually know your name. Um, nice to meet you. I'm Elliot. Um, I think we followed each other back when I posted like the
Tesla Waymo thing. And you know, you were posted that and grew my account a ton. So I appreciate
you for that. Um, but you were mentioning that you use Opus 4.5 earlier. I didn't know you were
I know you were a coder, but like fill me in, like, what do you use this stuff for?
a coder, but like, fill me in. Like, what do you use this stuff for?
Yeah, so lately I've been doing some different projects.
Like I've got this autonomous car database.
So we're tracking telemetry from autonomous cars.
And basically they'll upload kind of where they are, how much they're driving, that kind of stuff.
where they are, how much they're driving, that kind of stuff.
So it's some pretty serious back-end, front-end work across the stack
using Kafka, Postgres, PostGIS, that kind of thing.
And I mean, for the last few years, I've been kind of writing,
making YouTube videos, doing social media stuff.
Before that, I was doing a lot of software.
Me and my freshman year college professor started this software company and sold it.
Then we had this family software business we sold to a private equity firm a few years ago.
But I really just enjoy writing software.
It just feels good and uh i want to
build some more apps but i've always been kind of you know too lazy to follow through with stuff
the llms really change that because it just makes the gratification so much more instant
you're not kind of working through a bunch of
maintained problem, you know, mundane problems, trying to eventually get to a working product.
You can just kind of put an idea out into the world and, you know, like an hour. So I think
you're going to see a lot more people like me who were not necessarily doing software development work actively,
who, thanks to LLMs, it can even be a part-time thing.
You could be a doctor or somebody else,
somebody who has a different day job
and still put out products on the side
that are generating real revenue.
I think I've probably had a few hundred people
sign up for my app now at $99 each.
So it's kind of a great thing when you just see those Stripe notifications coming in and people are giving you money.
Oh, you're already on that. Okay. Nice.
Yeah. And here's the thing. You can go get a job where you work and they pay you for how many hours you work. But then there are other jobs where you just build something and you put it out there and it just keeps producing money for you. Right. Like I want to just get one or two apps that are making like, you know, a few hundred thousand dollars ARR, a million dollars ARR. I think it's totally doable i'm just too lazy or but thankfully
i got caught or you just be an engineer pick up on the signals because i mean why would you write
any code manually it's just like signal and context management nowadays make sure you don't
bloat the context window so it performs well but it's like this is what i do like day to day right
you you don't have one session the models aren't fast enough to just have one session what i do like day to day right you you you don't have one session the models aren't fast
enough to just have one session what you do is you be a complete psychopath you go through a ton
of pain for a month basically burning out your brain and you get used to having 10 clods at the
same time and you work 10 different day jobs you do this so humphrect basically uh it works um like you it's not that stressful to to do 10 different projects
or like different working i don't know daily yeah it's it's feasible now maybe three months ago it
wasn't it doesn't take very many like stripe subscriptions to pay off a mac mini so if you
do all of your accounting
and how many Mac Minis are you accumulating,
then you just get hyper
hockey stick growth.
It's the year of the Mac Mini.
Yeah, and it's interesting because
the models are still kind of stupid.
You still need someone who knows
if what's coming out of the model
is dumb or not, right?
Or else you're going to get hacked.
What was that app that was posting people's dating profiles or something
and they got everyone hacked?
You don't want to have some vibe-coded security flaw
and everyone loses their data.
So the models are still kind of stupid in that way,
but they're just getting
better all the time. Like every month, the best model is going to be better than the best model
was last month. And in a year, it's like just an insane difference. So that's going to continue to
be, you know, a wind in the sail of developers, I think for sure. But right now they can,
they still need help, right? That's kind of the
interesting thing about AI, right? They're kind of completing our sentences. So if a smart person
talks to an AI model, they can get it to do amazing things if they ask the right questions,
if they say the right prompt. But if a stupid person is talking to the model,
they probably really, you know, can't get that much out of it
or they'll get an app that doesn't work well
or has some huge security flaws.
But yeah, give it a year or two more,
even a stupid person will be able to create
a billion-dollar business
that's just run entirely by these models.
Someone quote that.
It's like the meme with the IQ points.
So if you're anywhere in the middle of the bell curve,
then you're kind of fucked.
But if you're right at the high end or you're right at the low end,
then you're basically on the same playing field.
Yeah, I'm definitely on the low end.
Have you guys heard of the handicap principle?
Fill us in.
Do explain.
Basically, it involves all these things like costly signaling and so forth.
But if you think about what's a handicap,
and so on one end of it, you're basically showing off.
And on the other end, you're sort of strength training.
So you can either, if you're a smart person, you can use AI to show off.
But, and actually, I mean, you can use it at any,
so if you're a smart person, you can use it to show off so if you're a smart person, you can use it to show off.
If you're a dumb person, you can use it to show off.
But if you're a dumb person, you can also use it to learn.
So you can do strength training.
And if you're a smart person, you can also use it to learn
to do strength training.
If you are anywhere in the middle,
then you're just getting, like, beaten around by guys
who are way, you know, way down bottom tier to you.
And then simultaneously, you know,
anybody else around is going to be like the tippy top.
Like if you play Quake, then most people suck at Quake
and then there's just a dude who's like running around on 45 to zero.
Then that's what AI is like. And the issue is that inherently,
if you're using like an AI thing,
it's like a tool-assisted.
You know, you're like running a hack in a way.
And you're using that to plug in a gap in your intelligence
and you're using your compute to do that.
So you're having to make a trade-off between
like the available system resources that you have to do processing
for your like cognitive cognitive load right and? And so the ideal, the most efficient,
most intelligent extreme
is that the AI model isn't doing anything.
It's just doing processing,
but you have to find your way to get there.
And at the other end, you're just, you know,
turning off the computer.
Yeah, I tend to do that a lot more these days.
Like I use AI for a lot of things,
sure, but not as much as I used to. Like I can't remember the last time I actually generated an
image or anything like that. I use it for like mainly Google search because Google is like so
bloated right now that you can't really like get much of anything out of it. And it's just like,
hey, I can go through all of this bureaucratic mess and rot my brain, or I can just have the
computer do that for
me and then basically just read through the result of that and then do research based
on the details of the actual valuable components of a piece of text, for instance.
Or I just use it as a means to correct grammar.
But then again, if you wanted to correct your writing or speech, just read The Economist,
bro.
They use actual english properly
i only made it till about the end of q1 this year on ai stuff because i i was on like the 200
chat gpd subscription and then uh o1 came out and deep research and everything and i just ran into my context limit where I basically just, you know,
DDoSed myself and like the context complexity was just amplifying
up to the point where the overhead of like doing stupid AI shit
and just canceling all my subscriptions.
So I was going up like the sales tier.
It's really good that there wasn't a $2 thousand dollar tier because i probably would have subscribed to it and um and then i
just canceled it and then like took six months not using any ai it's interesting when you like
not use these types of systems for a little bit that you realize how little you're capable of but
then after you've like gone through that period and you come back to the system you go like oh
wow i'm actually smarter than i was before it makes a lot of sense to like do a little bit and then go away for a day and then come back
and then do like one prompt in the morning especially if you're on the on the free tier
of claude like if you're on the free tier of claude it's pretty smart and that it can actually
write really good c code but you have to like think scarcely about what three prompts do I want to use today for like
what C code I want to generate and then like use the, use the C code that you got out of it for
the rest of the day and then come back and go, okay, now I want this feature. I kind of go by
like the Mike Metzler thing where it's like you do progressive overload where you say you try to,
you try to basically like put your mind through the paces of something
that's useful you do that as much as possible and ai is like this thing that disrupts your ability
to do that but also enables your ability to like increase your knowledge base so like you kind of
introduce that uh over time so you like spread it out say like you know once a week how many times
do you actually use it or once a day how many times you actually use it and what do you use it
for and then just like keep a tally of that as much as possible you're trying to remove the
it does the thinking for me bit because otherwise you're going to lose your ability to think right
like that one of the interesting things that i've seen that greatly trouble me actually is also one
of those things that will be interesting for the coming year uh is that people just like outsource
uh pretty much all their thinking to any of the systems that are available.
You get to the point where eventually you're going to be at a store,
you're going to be at some fucking bodega or whatever,
and you're about to get a pack of cigarettes over the counter,
and you ask ChatGPT what on earth you're supposed to say.
Guy says, how are you doing?
And then you go, hey ChatGPT, the guy said, how am I doing?
What do I say?
And it gives you some sort of like weird sycophant-like response
where you just say, Grokina tells you,
how about you just get on with the fucking thing?
This is just the perfect example of like how AI exacerbates autism though
because the actual chance that someone is actually in that situation and doing that for real
versus sitting on x and a high you know uh imagining that what what would i say if i was
in that situation and they're never actually going to be in that situation ever because they're they
wouldn't even like walk up to a bodega or go outside. That's true, yeah.
But they'll still ask ChatGPT what they should say in that situation.
Mark, do you have LLM psychosis?
No, I don't.
Okay.
How can you have LLM psychosis if you don't use them?
I'm just curious.
Also, it's general psychosis, I'd say.
I think it's like LLM psychosis is kind of like a...
Thunderbolt psychosis.
Yeah, yeah.
Thunderbolt psychosis.
It's just one way to decide.
It's just a condition.
There's just general psychosis.
And one thing just sets it off or makes it really obvious.
I think everybody's actually psychotic.
Well, not everybody.
I think a lot of people are actually psychotic.
They just don't know it.
Because they live in a very lulled, packaged, padded world.
And they don't actually encounter any of the sharp bits or the weird corners that look not as straight.
And then they just look at it for the first time in their lives and they just completely crash out and you have rioting.
That's kind of how I see this. I think that psychosis is everywhere.
It's just not... There just isn't a means to like express it.
And I think AI was one of the easiest means to not only express it, but also, uh, it's,
it's very abstract at the moment.
Octra, Octra psychosis, fully homomorphic encryption psychosis.
Yeah, pretty much.
Yeah, pretty much. Like everybody has something. Everybody has something.
Everybody has something.
I have quantitative analysis
going for me for the rest of
2026. That's going to be my whole thing.
Tons of research, lots of learning.
I'm going to see if I can actually condense information.
One of the things I figured out that's really useful
is you can either bloat data or de-bloat it.
And it's like, how much of that can you do
with your own brain?
And how well can you take the bits that you've created
that are de-bloated and stitch them together
into a fully comprehensive take on a specific
situation such that you don't
need to sit there for hours and mull over it
you just basically already have the answer and you can interface
it with any of the other problems
and answers that you have been able to mine
Tesla Rhodes
pre-order psychosis
we have to all just like take the night off
he's mining words he's mining words with a mechanical pencil and some grid paper
and just really nail out and think critically what type of psychosis you want yourself to have in
2026 you have to you dude it's too late you can't do it tomorrow's too late you're going to be with
family you have to do it today pull out the late. You can't do it. Tomorrow's too late. You're going to be with family.
You have to do it today.
Pull out the grid paper, mechanical pencil.
What type of psychosis do you want?
Think very carefully.
You have to psyop yourself into what type of psychosis.
It's going to be the most high-specific.
What specific, like, geometry grid paper?
Are we talking isometric psychosis?
I don't know. Just any grid paper, man man i don't care he's over complicating lines
i don't know i personally have tensor psychosis so i like to use grid
that makes sense
or oh wait actually an insane bet, though.
I don't know.
I don't know if y'all have, like, a couple hundred bucks laying around,
but, like, actually get a whiteboard, dude.
Whiteboard is, like, the best financial decision I've ever made.
It's just, like, right next to me at all times.
Any time Claude messes my idea up, I'm just like, okay,
we're going to write the entire architecture.md on a whiteboard instead of Markdown.
That's what we're going to do.
You should, like, you should take it. You should take the phone and put it ind on a whiteboard instead of markdown. That's what we're going to do. You should
take the phone and put
it in front of the whiteboard and then have it talk to the
whiteboard and make it reason with itself and then figure out
what the actual situation is.
I don't trust multimodality.
No, it's not there.
Elliot, do you have whiteboard psychosis?
Yes.
Unfortunately, I don't use it enough.
It's just like it's just painful to have to like write things by hand when i could just like speak them into existence what if you sold your computer monitor and got one of those really cheap
projectors and then project your screen onto your whiteboard that's's too many... No, no, no, that's too much.
Not just a white...
Dude, I want to feel the freaking ink
rubbing and screeching on the whiteboard, man.
It has to be authentic.
You can still do that.
Just get yourself one of those laser projectors from China,
and then you can just write shit on the whiteboard still
and not have any interference,
or at least minimal interference,
because of the light and the shading.
You ever seen one of those?
Why are we overcomplicating this?
It's not overcomplicated, man.
Aztecs as instruments, they're the ones who made
the little chips that make that possible.
Yeah, that's right.
They're inside of Chinese products.
How much RAM do you guys think they're going to
put in the Tesla Roadster?
Ooh, good question. I don't know. Maybe 300 megabytes?
Maybe less.
What do you think?
Well, what is the self-driving?
It's like, what, a transformer of some kind,
probably like a few hundred billion parameters,
and it's going to be doing...
I mean, that's a VRAM problem, really.
You just do the software thing, hold them on SSD,
load them into...
I don't know, man.
Clearly, they've optimized this enough
to the point where they don't need DRAM.
You still need to do DDR3.
We need a 10ense Torrent chip.
Yeah.
Well, I don't know.
SRAM.
Can they really produce that many chips?
It depends on if Tesla buys them or not.
I mean, Tesla is buying from NVIDIA anywhere, right?
For the moment.
They're not making their own chips.
They sort of have this XAI thing now.
XAI is also buying from NVIDIA, isn't it?
Yeah, yeah. There's a secret third option to this,
and I don't think I can comment on it.
Those GPUs that were like, Tesla'ss, that they ordered from NVIDIA that went to XAI.
The old Jensen package.
Gotta love Jensen, though.
Bro, Jensen is...
Dude, I want to meet Jensen, dude.
The other day, I woke up from a dream.
You haven't met Jensen?
No, I haven't met Jensen yet.
But the other day, I had a dream, and Jensen was there.
It was at my old elementary school, and Jensen was just there.
I was like, why the fuck is Jensen here?
Anyways, I asked him.
I said, Jensen, do you know who I am?
He goes, no. Do you know of my cuda course 12 hours it's on youtube he goes no then i woke up and got sad i think
it's time for you to turn off the wi-fi in your room before you go to bed
okay like real shit
okay like real
definitely there's no way
his engineers know my cuda course they follow me bro of course he does
maybe not him
by the way if you want to learn cuda and how to optimize gpu deep learning kernels
uh you should look at my pinned post
no one else has done anything like this bro like
it's it's a banger i spent an entire summer making that course it's pretty good
but didn't you say that you regretted putting out stuff for free uh well for free um no never for free um but
like putting out a paid textbook is kind of like a waste of time this is another thing that jensen
said he's like did you shelf did you shelf the whole project did you like not do the textbook
oh no no no um it's just like publishers are slow and they've been and they've not really
been respectful of my time and energy and resources,
and so I'm just, like, kind of pissed off at them.
Wow.
Yeah, no, no, no. I'm never going to publish a book ever again. Terrible idea.
NVIDIA doesn't actually want people writing CUDA.
They just want to get you trained up on CUDA and then hire you onto their internal proprietary.
That's where you learn the actual Knights, yes.
Well, what you learn is like, oh, this is how basic a Kublaz S-Gem call is.
Okay, now just make everything wrappers around that and optimize the S-Gem itself with evolution or something.
Basically, what he's saying is they give you the ability to absorb the brain rot so that it restructures you in such a way that it's like highlights whatever potential you have so that NVIDIA can look at you and go, yes, we can integrate this thing into our Borg collective consciousness.
useful skill um will you actually be programming them yourself no but do you actually understand
when you're able to look at gpu code and understand that this is what this is doing
that's a useful skill like for example the other day you know i've taught a fair bit of cuda in my
life and recently i took a an implementation on an archive paper and i said opus go implement this
training forward and backward pass CUDA kernel for this
layer in the neural network. And it one-shoted it and made it 5x faster. The neural network trained
5x faster in one minute after writing these kernels. And I did not understand shit. I looked
at it, I was like, this looks very optimized. I don't know what the hell's going on.
I don't know what the hell's going on.
This is what I mean about the bell curve.
So if you knew how to write CUDA,
then you could just program the GPU directly
and the efficiency wouldn't include the overhead of you.
Well, not just that.
You can also find all kinds of security issues with CUDA
and how the C and C++ works inside it.
Could be fun.
What are you talking about security issues?
It's a matmul.
Okay, here's the thing.
There can be some interesting things done inside the GPU.
Like there are some unintended processes you can run and stuff.
Nope.
There's, like,
maybe for Blackwell, because there's still
undocumented features there, but everything up to
Hopper, it's, like, pretty clear on what the
sort of case is.
Really?
Yeah.
Yeah, like, Ampere and Hopper
are solved. Blackwell is
like foreign shit, different
programming model, like like rewire your
brain type thing he see he just wants a job at nvidia where's your badge i'm gonna do the linus
torvalds thing everybody wants to do the line they're not gonna hire you except for linus
torvalds bro i don't want to work at nvidia i, I already have jobs. They're not gonna hire you, man. Fine!
Wait, what's wrong with NVIDIA though?
I think it's a pretty good company, I think.
Would you really try to say it's a really good software company that's also a good chip architecture company?
But like TSMC is the chip company.
It's a good stock. Bro, the issue with NVIDIA is that
if you were doing them, they would make me too rich.
I am a happy camper.
You already are corrupt.
That makes no fucking sense.
I just had to say it.
The main reason NVIDIA is so successful
is that they ray-trace their keynote presentations.
And nobody else in the entire world knows how to ray-trace their keynote presentations.
Ding, ding, ding.
Damn.
And if Tesla ray-traced their keynote presentations, then they would out-seat NVIDIA.
But it's too hard.
Also, their hardware doesn't have ray-tracing tracing they'd have to buy that from nvidia we were talking about this before but like basically everything will be okay i'm jumping
a little bit like i'm jumping like 10 steps in the future here but when we do have this like
dyson swarm close to the sun or maybe in Earth orbit, literally everything will be a bitstream.
You won't write software.
LLMs will be the compilers, right?
Everything will be a bitstream to your edge nodes, which are just phones that are taking
the AI bitstream and just displaying it as pixels, like the raw outputs, essentially.
If James Cameron made a computer company and ray traced it with avatar graphics.
That would be worth like $100 trillion.
No, no, no, no, no.
We won't have ray tracing.
We'll have AI bit streams.
The ray tracing will be simulated via image and video diffusion.
You sound like a garbled LLM that's like misspeaking words.
Ray tracing. I don't think you understand how many billions of tokens I've used with Opus today. LLM that's like misspeaking words Ray Tracer
I don't think you understand how many billions of tokens
I've used with Opus today
How many selfies with Jensen Huang
do you have?
I've got two selfies
with Jensen Huang
How many selfies with Jensen Huang do you have?
Damn, he's
mocking you, man
It's so over
wait what is going on i will say that what he says is kind of true
you know the lm problem it's like if you if you use it if you use it too much
like if you use it too much it does take up you start you start talking about
about llamas and stuff bro yeah bro llamas talking about llamas
i think i actually kind of agree with what elliot said in one of the other spaces um maybe a week
or so ago i think he said it will amplify what we already have so i think that's that's i i can
agree with that can we be like a little bit more vague, please?
No, that's not vague.
It's actually quite direct.
Amplify what you already have.
If you're retarded, you'll be more retarded at a greater scale. And if you're smart, you'll be just hyperscaling your smartness, I suppose,
if you can actually keep it without becoming retarded.
Wait, so if I'm both very smart and very retarded,
will it cancel out and I'm 10x1 normal?
No, because you are super moody all the time,
which doesn't make any sense for someone of your biology.
What ends up happening is both will be amplified
and it will look like they're canceled out,
but you're just going to oscillate between the two
and your brain just doesn't know the difference.
So you think you're multitasking
when in actuality you're just oscillating
and you're just like living in constant delusion.
So you're going to do the like...
The answer is yes
if if you have two computers oh no and on one of them you don't have any ai and on the other
you just let the computer so so life is like a constant increasing frequency thing of like
it's over we're so back it's over we're so back and it just yeah it's a period between those
increases every day yeah it's like when you it's like when we're so back. And it just, like, there's a period between those increases every day.
Yeah, it's like when you tell someone
that we have, like, a problem
and we're going to turn the entire problem
into a triangle
and you just, like, circle it
and go around and around and around
until you finally, like, just say,
okay, well, that's just...
That's like a...
The Chinese based all of their philosophy on this.
It's like a black and white yin-yang thing.
Yes.
You are going to burn out
Him or me
All of us
Bro if
Not me
You do it's so over we're so back
If you're doing this
What do you think a computer is doing
Have you heard of binary logic
This argument makes no sense.
Yes, it does.
Somebody doesn't know how computers work.
Somebody doesn't understand
that computers don't even know what
it's or so back even means.
It's just a vector of floats.
Okay?
It's imitating the understanding.
It's not actually understanding.
Yeah, but what the question is,
what is it using? It's a binary signal. Yeah, but what the question is, what is it using?
Yeah.
What's a what?
It's a binary signal
is what he said.
Everything is a binary signal.
I mean, it technically is.
Except Christ.
What is he asking me?
I think he's losing his mind
no he's about to tell you about the octonian thing and the holy trinity
unless you want to
triangle consulting but that might work too yeah exactly i mean you have a triangle in your profile
so clearly like the triangles are winning well Well, this, you see, if you actually change...
Actually, I can't argue against that.
Did you know that GPUs run into triangles?
GPUs what?
They're actually triangle accelerators.
Well, what if I replace all of my CUDA cores with Tensor cores?
Didn't they try to do the non-triangle thing and nearly kill the company?
Adrian Auto is here.
Why do you have so many listeners?
Elon was here.
It was the whole thing, man.
Don't worry about it.
I see.
Don't worry.
It was fun.
I was just starting this space.
I didn't think it was going to be anything like that.
And all of a sudden, boop, bro appears.
I'm like, oh, nice. This is good.
I'm looking down 194, 673.
What the fuck?
Yeah, it's just like it's a constant expect.
It's the year of the mass driver.
Yes.
Wait, one second, real quick.
Elliot, have you heard of leftover locals?
Oh, no. Send the repo link. Yeah, already second, real quick. Elliot, have you heard of leftover locals? Oh, no.
Send the repo link.
Yeah, already did in the chat.
So they actually have a really good POC for GPU memory leaks.
You can check it out.
I can probably post it on the comments also.
I want to, like, wrap this up in a little bit.
We're still going to keep going here. Yeah, we're going to, like, move this over to up in a little bit. We're still going to keep going here.
Just, yeah, we're going to like move this over to Discord in a little bit.
By the way, for those of you who are wondering,
make sure to join the link down below.
I pin it up in the reply thread.
So yeah, actually, let me pin it to the top of the space.
I do some like after sessions in there as well for my subscribers.
And yeah, it's just kind of like community Discord. We still, of course, you know, do most of the content. I do some like after sessions in there as well for my subscribers. And yeah,
it's just kind of like community discord. We still, of course, you know, do most of the content
creation on X as it always is. We live stream everywhere and all these kinds of things. But
like if you want more of a direct kind of like connection, then this is one of the best ways to
do it. Because like there's this big problem of like having a parasocial relationship and like
social media is actually not social media.
It's parasocial media, right?
But then through something like that,
you actually get more social,
like you get more direct basically is what the thing is,
you know, because right now,
like I see 2000 something people,
not every single one of them are like,
I can actually connect to in any meaningful way
beyond just like them hearing me
and maybe me catching a glimpse of their profile picture.
So it's like very parasocial relationship.
So, yeah.
Also, what the fuck is this?
Why is there a prostitute in here?
What is this, man?
I sometimes look through, like, the listeners just to, like, see what's there.
And it's, like, sometimes I see accounts that just, like, pop in that are completely unrelated to every discussion that they're in.
And just, like, try to boost their numbers by, like, appearing in a space.
And it's just fucking prostitutes. It funny as fuck wait which one what do you mean which one
what do you mean which one what is it with you and the specificity of those details exactly like
you're really not beating the allegations my man like i'm the guy who comes up with the ridiculous
metaphors you're the one who just executes on him i'm like this is not what that was supposed to do but sure
what the hell i've i've like a hundred people have followed me in the last 10 minutes
don't worry about it man i gotta go block them all
hey you know the fun bit is once you like hit a certain level of
acceleration you put it to like
the mass driver of x where you like put it put into
a suggestion section so that every
single time a person like creates a new account and they have to
like follow three to four accounts they'll just like
follow whatever they see and you're just like
inundated and you go like
okay so 7000 of those people are not
going to engage with my stuff but maybe in the future
they're going to see things and be like, oh, this is interesting.
You're kind of getting a taste of that.
I think this is kind of like why you privated your account earlier, didn't you?
Did I say privacy is the next
mode or something?
Andres and Horowitz?
Real.
More like
90s, early 2000s
internet is like the mode.
You're talking about a world
that no longer exists.
Just let it go.
We're going to make it real again. we're going to make it real again.
We're going to make it real again.
Just in a different way, in a more complex, annoying way.
So it's going to feel just like what it did back then,
but just for us in a new and annoying way.
I'm trying to get the Octra guys to bring back America Online
and AIM messaging.
Bridge the AIM messenger to Discord for you.
Nice.
Just, like,
have it all, like, go over audio
modem, like, handshake,
with weird fucking sounds.
Like, bring all that back, just for funnies.
We're actually getting there again.
It was really funny when I saw this, like,
agentic AI last year,
where they were communicating via
this thing called gibberlink it's like oh hey we're calling each other and we're on the phone
and then it just like does weird fucking sounds to communicate with itself and people like oh my
god this is so cool it's so new i'm like have y'all just like missed out on the fact that modem
handshakes are a thing this is like inferior by comparison what do you what like why are people
so impressed by this we need to to use AI to make really complicated
new custom modem handshake sounds.
Yeah, lots of infrasound shit. We can do a lot of good stuff with infrasound. You wouldn't even notice this happening.
Do you think we could do encrypted audio over
X-Spaces and then you need to run an AI model on your
phone to decrypt it and everybody
who's like some pharma BTC holder from Nigeria that follows me now has to buy a subscription
to get the key code.
Are you suggesting a crypto miner or what?
No, no, that's not what he's saying.
No, no, what he's saying is that you could disseminate a signal.
The signal is then taken by an LLM or something.
And then that's turned into like actions because most people are automated.
If you play the right, you know, like dial-up tone,
then everyone stops saying llama, octa, mass driver.
They're keep repeating
certain things.
And then they start talking normal.
You can start repeating certain words
and people have already made themselves
super dumb in the crypto scene.
So what they do is they then take words
and they try to turn them into money.
And the next iteration of this
was instead of scamming people
and turning it into potential SEC violations
that will later on fuck your ass,
what ends up happening is you have a thing
now that's called a betting market
and you just do like
you know, you do
Polymarket or Calcio or any of the others
that may pop up over time
so it's like now you can say all kinds of things
you can say BTC, LLM
Facebook
Twitter
Planetary Defense Planetary defense.
Vibe.
Vine.
Computers.
Space.
Moon.
Rocket.
There's going to be a few other ones, but I'm not saying those.
Rail guns.
We were told we weren't allowed to say that word.
I said Railgun.
I like Railgun.
It's clean.
Mass Driver is cool, too.
Luna Mass Driver.
Luna Mass Driver is nicer.
Would be a beautiful name for a girl.
Bruh.
I like the name, too.
Oh, no no What?
Just like one word, like Lunar Mass Driver
Like all together
It's very German of you
Can you translate it into German?
It's not very German of him.
It's not long enough to be German.
Yeah, it gets close.
They tend to take words and they put them all together
and then turn it into a singular word,
which is really interesting.
So instead of hyphenations,
they just remove the spaces and jumble it up all together
and jumble it up all together
and just like have you work out the phonetics
and just have you work out the phonetics along the way.
along the way.
Like how
ship
what is it? Shipping?
Not shipping. What was the
ship fart is like three Fs
which is like
ship
travel
whatever fart is. Fart is like travel
going something, moving, bit, you know.
So it's like it's three Fs and you don't know that.
You have to know that to get why it's such a long F.
Languages are so weird.
Especially English.
There's so many hidden Easter eggs in everything.
And we use a lot of words.
Completely incorrectly.
I figured out the missing word.
It would be the Volks Lunar Mass Driver.
How is that?
Like for the German one.
You know, the peoples.
The peoples Lunar Mass Driver?
Yeah, yeah, yeah.
Is that what China's going to make?
Is China going to make the peoples Lunar Mass Driver? Yeah, yeah, yeah. Is that what China's going to make? Is China going to make the People's Lunar Mass Driver?
That'd be pretty good, wouldn't it?
How are you doing, by the way?
Didn't think you were going to show.
Me too.
I just got finished with something I had to do.
Oh boy.
I'm treating it.
I'm going to pin it.
I was like, what the fuck
is going on? Why does it have so many listeners?
Yeah. Yeah.
It was nice.
I wanted this to be an hour feature,
and it really is an hour feature, so, yeah.
I think we're kind of like,
is there anything else that we need to still cover?
Anything that could be of interest or value?
Anything like that?
Did they remove the pin tweeting thing?
How do you pin things?
Oh, I can't do it unless you're the host.
Yeah, I'm sending you the... Okay, actually you can. Remove the pin tweeting thing. How do you pin things? I can't do it unless you're the host.
I'm sending you the... Next year is going to define the world.
Next year is going to define the world.
If you drop the wall, let's see what happens.
Yes.
You have to see what happens with the fat.
You have to see what happens with the midterms.
Even other countries.
And the fat people too.
What? What do you mean what
do you not know they're gonna fucking like
hyperscale more of the fucking GLP1
stuff dude look at how fat everybody is
man it's a huge market like for years
you've been like put into denial being like oh that's
totally healthy of course it's not
you mean the GLP1 inhibitors
yeah GLP1 inhibitors
yeah
yeah the GLP-1 inhibitors. Yeah, GLP-1 inhibitors, yeah. Yeah.
Gotsmonde mesentriba.
The samaglutides.
I mean, there's always some security news if someone is interested in discussing.
Outer, have you been following
MongoDB CV?
I don't know. Have you been?
I heard they had like a hack that's like affects 10 years of MongoDB or some shit.
Yeah.
Yeah.
They recently had a CVSS 8.7, which is super high.
Yeah.
And they even had an active exploit, which is a huge problem.
MongoDB doesn't even have transactional consistency.
It's an eventual consistency
database. Just using MongoDB
as a security vulnerability.
There are four groups that are involved in
the CS... Not the CS, for fuck's sake.
I've never used
MongoDB.
Is it a NoSQL or an SQL
database?
Yes, NoSQL.
Oh, okay.
Fair enough.
And they don't provide any guarantees
that your operations are going to be
applied in the right order.
Well, can anyone provide guarantees, though?
What?
Give me the data.
Give the data to me and I'll handle it.
Don't worry about it. Yeah, give him the data and give me the money. We data to me and I'll handle it Don't worry about it
Yeah, give him the data and give me the money
We're all gonna be good
Don't ask too many questions
Exactly
Don't worry, just give us all the power
Give us access to everything
We're totally not gonna do anything bad with it, we swear
I'm gonna put all the data in a fucking orb
And then you're gonna be asking
What happened to the data And I'm gonna be like, data in a fucking orb and then you're gonna be asking what the fuck is what happened to the
data and I'm gonna be like it's all the triangles
I'm pretty sure all of the people in this space are fake because I
posted a tweet and I pinned it and
Even though I've gotten like over a hundred followers in the last 10 minutes
I've only got two likes and 56 views on the tweet that I pinned.
So maybe they all just have it in their pockets or something.
It's a meme thing.
It's a meme thing.
I got a third one.
There's a problem with getting followers from viral hits.
It's that most of the people that came from viral hits don't fucking interact with you.
Yeah, it's always going to be sharp moments of progress.
understand about like the it's we are so back it's over kind of like circle is that you kind of the
goal with it being so over is that your next over is higher than your prior back so like if you if
you've been to a point where it's like oh we are so back right the next time that you're so back
you have to like so you have to like really eclipse that so that the next time it's over, it's never below your last back.
This is also kind of, you know PewDiePie, right?
The guy that used to be the biggest YouTube channel.
This is one of the good things that happened with him.
Is that he never had any super viral videos.
So he grew his channel kind of consistently.
Whereas a lot of other channels, they had a viral hit and that's how they got like a million
subscribers and then it would just keep like and then now there's like 30
30 AI reply people who are insisting that they're real in my comments wait
PewDiePie did a really good thing he switched to Linux I think that's the
best thing he did not roasting PewDiePie what a really good thing he switched to Linux I think that's the best thing we're not roasting PewDiePie
no he's not
I'm saying it's a good thing
yeah but we're not talking about that
stop changing the subject
I'm just talking about the algorithm
and how it works
because when you
have a consistent audience that's better than
a million when you have like a hundred
thousand consistent followers it's much better than a million... When you have like 100,000 consistent followers,
it's much better than a million inconsistent followers.
Yes.
It also depends on the nature of who those followers are
and what kind of content you create.
Say, for instance, in here, a large average number would be about 400 people uh this is of course
massively inflated now because elon's attention on the situation right undeniable of course um but
then you go into something like the noetic order discord server and you have like peaks of 50 maybe
60 people who sit in the advanced membership section with me right and like those people each
pay like fucking 15 bucks to be there or actually actually 10 depending on whether or not you
subscribe on the browser or on the phone you should always subscribe on browser because otherwise
apple or android takes a cut of your uh thing and just makes the price higher for no fucking reason
in case of android it makes sense in case of apple there's literally you don't deserve
shit but you know you buy super expensive phone i don it. I mean, both of them are 30% extra, so I don't think both deserve anything.
No, and Google...
Google makes sense.
Why?
Discord, for whatever reason, doesn't accept the Google Play Store.
They send you to a browser that you need to pay through.
you need to pay through.
Oh, that's interesting.
Oh, that's interesting.
In general, I don't have so many problems
with fees for something like
Google or Android.
Because their products are different.
It actually makes sense.
I'm not saying that they're good, those fees.
I'm just saying that I have less of a problem.
It's a dilemma, basically.
Yeah, Google basically
doesn't make any money on the phones themselves
that make money from your data
and Google Play services.
Whereas Apple makes a shit ton of money from the phone.
And then they're like, hey.
Yeah, but then they recycle those profits
into doubling the RAM on the Mac Mini.
So as long as you buy Mac manis, then it works good.
You know, it is actually like, I think the M chips are the best product out there, especially right now with the unified memory, right?
But also, a lot of people used to give Apple shit for the MacBooks before they switched to the M chips.
And those fucking people don't know what the fuck they're talking about.
And those fucking people don't know what the fuck they're talking about.
Because it used to be that Apple was fixing all of the issues on Intel chips.
Apple was finding as many issues on Intel chips as Intel themselves were finding on their internal fucking testing.
You know what the end game is with the M series though, right?
Servers.
AI servers.
They're going to be fabbed by Intel.
What?
I hope not.
I hope not.
Oh, it's coming.
Two years.
It's probably going to happen because the government...
Two years for the announcement.
The government's going to force them to do it.
But I don't...
NVIDIA just bought like 5 billion and more Intel stock.
I know NVIDIA just bought 5 billion dollars of Intel stock.
But they also tried Intel's 18A process.
And they were like, nah, bro, this is shit.
We're going to go back to...
So like the only reason Apple has to make their own CPUs is because Intel doesn't know how to do their job.
Yes, pretty much.
And so they're basically just making the CPU that they are forced to make.
And then as soon as somebody else can do their job better, then they'll just go and outsource that, right?
Yeah, Intel has so many issues when you bought a pc when you bought a pc you need to
understand the reason that your intel chips were better than apple's uh for the reason apple looked
so overpriced was because intel was based intel basically had so many issues that you were basically
subsidizing you were being subsidized by Apple for your PC.
Because Apple was finding all the
fucking Intel chip issues.
Now they just subsidize the
RAM on the Mac Mini directly.
Yeah.
It's literally cheaper than the PC.
Apple has long-term contracts
with the memory suppliers
like Micron, SK Hynix, and Samsung.
So does NVIDIA, but like...
Yeah, but NVIDIA, the difference is that NVIDIA,
they were already selling...
NVIDIA doesn't know how to make a computer,
is the way to say.
NVIDIA was selling their chips to server,
to large companies like Facebook and Google and OpenAI and Oracle without the memory anyway because the companies wanted to customize their stuff anyway
the memory
configs
it was mostly
the ones with the memory
were already for smaller companies
anyway
2028 is going to be the year of the
memoryless process
talking about like using NVMe
no that's memory talk about like using NVMe no
that's memory
I mean it's NAND storage
Autor did you hear about the
booking of 2nm chips
from TSMC by Apple
no I didn't
do you know what the
M and NVMe stands for?
Yeah, memory.
But usually when you say memory,
people are thinking of RAM.
Did you know that you can randomly access non-volatile memory
yeah I mean
in Linux you can set up
you can set up a swap partition
that effectively acts as RAM
Intel actually made this
memory tier called Optane
a few years ago
which basically was
in between RAM and
NAND
I hear it
perform like shit
compared to
yeah it was like bad memory and bad RAM
bad flash and bad RAM
yeah
yeah bad memory and bad RAM. Bad flash and bad RAM.
Their GPUs are kind of shit too.
They're getting
beat by the Chinese on their fucking
GPUs.
Yeah, they've been getting really
good too.
The one thing I don't like about china right now is
the fact that they're getting so aggressive you know it's going to be like really it's going to
be such an eerie cold war feeling i said this earlier in the space as well with like whatever
china is doing because like they're just becoming annoying in a sense you know what i mean like it's
with this latest stuff with taiwan and all this you know, I wrote in the general chat yesterday,
I was like,
hey,
this is just kind of signaling to the US more than anything.
It's like,
hey guys,
we're going to do this.
You know,
we can do this.
They are doing us great open weight models.
Yes,
but these are all good things because this is part of the asymmetric warfare part,
right?
This is what they've been doing for quite a while, which is a really good thing.
It was a really good strategy.
This is why I don't understand the whole Taiwan thing.
It's just like, okay, are they trying to basically show, yes, they have the power to do a thing
and to make it really annoying, I suppose?
Perhaps that's what they're trying to do.
Do they actually take the US defense thing seriously, or do they not take it seriously?
I don't know the answer to that.
Either way, what they're doing is just kind of ballsy, right? Unnecessarily so, I might add, because it just adds
way too much attention to a hand that I don't think needs to be played just now. Or maybe they
really are confident of what they're doing, and maybe this is as good a time as any to play that
hand, because you're going to need more of like a transactional, quite reactionary response
as Trump usually does it. He's very
transactional in a sense, more so than he has
like these
deep long-term plans. He's very transactional
and somewhat instinct
driven from what I've been able to see.
And so I think this is actually
maybe one of the reasons why I play it this way,
because you can kind of instill within a person
a set of choices, because of the pressure
that you put on in this way,
that create a certain type of instinctual reaction
from Trump, perhaps, I guess. I don't know.
Or the rest of the world, whatever goes on.
Yeah.
What do you guys think?
War Games, 2026.
Lots of narrative engineering.
Should be fun.
Don't get confused.
Yeah, don't get confused yeah don't get confused and start
outside off for the war
I just hope that the
wall doesn't have to switch to intel chips
yeah
did you see the
space force trading
cards
were they renamed all the concepts?
No, no
No, I did not see that
I don't really see that much political stuff
On my feed anymore
Because I decided to unfollow
Like 80% of the
Government people that I used to follow
You cannot imagine how much that cleaned up my feed.
It's actually much nicer now. Now all I see is stuff about silver,
stuff about chips, different types of materials,
new experiments with 3D-printed rocket engines,
things like that. Occasion occasionally like a cat or something
that's pretty good
orbital warfare
Norse gods
cyber to mythological
creatures
EM to serpents, navigation to
sharks, sitcom
constellations
yeah to Sharks set com constellations yeah oh yeah
also part of that
cyber warfare
that's another aspect
this will become
this will become
very apparent soon
I think
because you're gonna
again part of narrative warfare
is like you have to
like Again, part of narrative warfare is like you have to...
Like... It's more than just physical warfare a lot of the time.
At first, everything's done on paper and through memes.
And like these memes could take on any kind of shape more.
Like say, you have very legacy forms of this
where like politicians are beefing with each other
or they're trying to say get the population to side with a certain set of politicians and their
ideals all this kind of stuff so it's kind of like cyber warfare in a sense
I think it's just a the whole cyber thing is just the advertising campaign for packet processing.
Please buy my overpriced server to do packet processing.
I have a firewall.
Shit.
By the way, fun fact.
The HTTP certificates might be expired.
You need the packet process
what do you think cyber attack what do you think the chances that like elon's going to reach a
trillion this like 2026 by the end of 2026 i mean just look at the usd
chart at 100 years and then just like zoom out to however many trillions you want.
I know what you were going to say.
Fair enough.
I don't think we should wrap it here, not going to lie.
Sounds good.
I still think James Cameron,
if he ray traced Tesla's thing,
that would get him to 10 truly faster.
And I hear that he's looking for a funding model
for Avatars 4 and 5.
So he's looking for a sponsor.
So if he did a deal to like ray trace the tesla deck also that would uh bring forward elon's
compensation roadmap just an idea see ya speculation all right see you man all right we're gonna we're
gonna move into the uh we're gonna move into the noetic order now, where I do the daily thing that I usually do.
The link is pinned at the top, so you can go check that out.
Join me in there, this is the daily subscriber thing that I do.
So yeah, thank you all for coming.
Hope you enjoyed the session, and perhaps maybe you see some value in joining the advanced members.
Not forcing anybody to do that, but also make sure not to spam into the Discord server, because I will ban
you without hesitation. Like, don't
test my patience, I don't have that much of it.
So, yeah.
Alright everyone, I'm going to head over there.
See you all in a little bit.
See you there. Thank you.
Insights
Project L
P
