Projects
![]()
No projects found
Investors
New
DAO Treasuries
DATs
Lend / Borrow APRs
Chains
Categories
Spaces
Nvidia’s $100B AI Bet, Google’s Data Grab, And Humanoid Hype!🎙️
Recorded: Sept. 30, 2025 Duration: 1:40:21Space Recording
Short Summary
In a week marked by significant developments in AI and blockchain, NVIDIA's $100 billion investment in infrastructure signals a new era of partnerships, while OpenAI's chat-based checkout innovation transforms retail. The rise of AI-driven commerce is reshaping user experiences, with major players like Amazon and Etsy adapting to these trends, indicating a robust growth trajectory in the crypto and tech sectors.
Full Transcription
Thank you. Thank you. Music Thank you. Thank you. Music Thank you. Music Thank you. I'm going to go to the next video. Thank you. Music Thank you. Thank you. all right let's get started a lot to cover today lewis how are you doing good good yeah
it's another busy seven days so you know you know, a billion here, a billion there, and you add up to 100 billion.
So lots happening.
Yeah, no kidding.
So NVIDIA with a $100 billion infrastructure bet.
What do you know about this?
It's interesting because you've got this circle occurring between NVIDIA, Oracle, and OpenAI with $100 billion
kind of floating between all three of them.
It's kind of bizarre.
You've got a data grab from Google.
You've got a whole bunch on the humanoid and robot news pumping up, people kind of throwing
stuff back and forth.
But the real big news that really happened was yesterday, and that was OpenAI.
They introduced chat as the new checkout. And I think this is a really key announcement and
really important. I mean, it turns ChatGPT into a native point of sale. And this is not another
sales channel sort of thing. It's really a control point because whoever becomes the default decision surface is going to set the economics of retail and they're going to shape which brands are going to be discoverable.
And they're going to decide how trust is earned and priced and enforced.
I mean, it's really key and it's a really slap against, I guess, to the most part against Google.
against in i guess to the most part against google um you know i mean so so having this ability to
in your chat you know buy stuff off etsy or buy stuff off other places is a very big move and i
think what we'll see next is that instead of a single item you'll start to see the full multi
item card you'll see bundle selections and suggestions. You'll see cross
merchant orders handled in one thread. Obviously, the competition is going to move really fast with
it. Google, Apple, Amazon are all going to bring native checkout into their assistants.
So there's going to be protocol skirmishes and assistant channels in every commerce platform
going forward. You'll see assistant media with closed-loop sales.
There'll be retail media budgets.
Again, shift to conversational placements,
where impressions, engagements, and conversations
live in really one flow.
You'll see new merchant scorecards.
We're going to have to have dashboards
that track assistant impressions, answer quality,
trust signals, refund latency,
and repeat purchases that
occur inside a chat.
You know, your lawyers are going to get their fingers into this stuff.
So you'll have to have clear rules for dispute child safety, like we just saw with OpenAI
introducing parental stuff.
And really some maybe dark patent prevention inside conversational commerce.
So I mean, it really is a huge change.
And in many ways, the shopping cart is dead and chat is now the new checkout.
What do you think about that?
Yeah, so one of the things I was wondering about was what happens to Etsy and Shopify and other marketplace UIs when these assistants are handling search, discovery, and payment.
I would imagine that just like me being able to pick up my phone and order anything I want at a whim through Amazon and have it come to my door, this is just going to give us a step up, right? Before I had to actually use my fingers and search for the specific product that I want.
And I'm sure there's still going to be a lot of that because when it comes to certain products,
there's nuance and you want to make sure you get the right one.
But I'm just curious to see what happens.
Are the UI is going to now be this thing where you just kind of speak into it
and you watch it do the searching for you?
Yeah.
And then, you know, you say,
well, this isn't the one I want.
Specifically, the one that I want has, like,
keratin in it and it can't get some shampoo.
And that's, you know, it's just one less point of friction when it comes to getting from point A to point B.
Yeah, I mean, you've got less clicks to do, right?
So, you know, you're in your open IR.
You're doing the chat GPT.
You say, hey, I want to buy a gift for a friend.
You describe it.
And then now you can obviously buy on Etsy.
And that's where their stock went up yesterday like crazy.
There was another merchant also.
It says if a merchant is already processing payments with Stripe, they can enable
eugenic payments in as little as one line of code.
So you're getting massive integration.
So you don't have to go to Etsy.
You don't have to go to Amazon.com.
You don't have to go to these stores.
Within ChatGPT now, you can search, order and buy and close and that's it.
Ryan, go ahead.
What do you think about this?
I think it's the next step towards replacing user interfaces with AI.
Yeah, absolutely.
Which I really think is the culmination of all of this is the website.
The website as we know it will go away. Web two, as we know it
will be antiquated and you will interface with an agent. And that's what chat to the T is slowly
becoming. It's becoming that interface agents that will build a UI for you on the fly. It will show
you the products on the fly and it'll let you purchase on the fly um so
it might be a boon for etsy short term but i think in the very near future uh etsy ebay like all these
you know marketplaces amazon will be toppled because they're all based around a user experience
they're all based around a ui and that's all going to be replaced with an agent building the experience for you on the fly.
And all we have to do is put a decentralized product listing system in place, like a Silk Road type listing that's all on smart contracts.
that's all on smart contracts.
And now you can have like a decentralized market
that's completely global,
completely peer-to-peer payments,
no processing systems.
And if you want to launch that on like,
say a circle layer one or a stripe layer one,
now we've completely upended
all of these shopping cart systems.
So I think we're seeing a step towards replacement.
Yeah.
I think we're seeing a collapsing of the shopping experience.
You know, marketplace brands like Etsy and Amazon and others, I mean, they're facing
a platform sandwich, right?
So they're going to benefit maybe with assistant traffic, but they're going to risk being essentially
disintermediated when buyers never visit their front door, right? And loyalty travels as metadata. So, you know, and I've said for a
long time now that AI is going to be the number one interface to all technology within the next
two to three years. You'll interface with nothing but AI. And like you said, websites, you know,
they'll be like, you know, it'll be like the,'ll be like the landline, right?
Some people will still have it, some people will still use it, but yeah, most of us aren't
going to do that anymore because it's just not fast.
It's not efficient.
It's not frictionless.
And it's like, hey, just tell the AI I want to buy this and find the cheapest price.
It'll be the equivalent of shopping through the JC penny
catalog yes going forwards by going backwards here yes and that that's the thing is remember
in the 80s and 90s catalog shopping was huge that was the user interface for shopping. And now we look at it thinking, wow,
like what is this really thick book that they constantly published?
Well, it's going to be the same thing with websites.
So we're going to look at them going, wow,
what was that antiquated software with all those pages that we created?
Right. I mean, the biggest effect could actually be Amazon. I mean,
because it really disintermediates Amazon and puts them under a layer, right? You know, Amazon
is a name that's a huge, obviously a huge marketplace. But now this really levels the
playing field, as you said, you know, and the only thing I'm seeing is ChatGPT.
And the only thing I'm seeing is chat GPT.
You know, it's funny.
Ten years ago, we looked at Amazon.
We looked at Google.
Actually, five years ago, we looked at Amazon.
We looked at Google, and we thought nothing could topple that.
Ten years ago, we looked at Facebook and thought nothing could topple that.
And now Facebook is scratching to stay relevant by acquiring all of the latest and greatest social networks.
They're building their own AI systems trying to stay relevant with Grok and all these others.
You know, X came along and ate everybody's lunch.
And OpenAI basically took out Google search within a couple months.
basically took out Google search within a couple months.
So it's,
it's crazy to see how when you have a new technology roll out,
there's a complete shift in just mindset and it completely topples these
before like insurmountable giants.
Like for example,
Apple right now is becoming very,
very lackluster when you
when you think of Apple as far as AI technology right now they just have a
corner on the devices and all it takes is grok to come out with a device or one
of these other you know big AI companies to come out with a device that just
completely blows everyone away with an operating system that just builds itself on the fly. Right. I mean, you brought a good
point around the social media stuff. I mean, that would be the logical next step that as you buy
this product, you can do a social media post which says, hey, just bought this. I think it's
really good or I could do my reviews i mean it i mean this chat
gpt experience could now really start to pull in a whole bunch of other things within that one frame
right so i'm talking to the ai i want to buy this i want to tell my friends about it
or maybe i don't because it's a present but i you know i want to post that so so it't because it's a present, but I want to post that. So it's not a huge leap to think that OpenAI could add social capabilities.
So what I was wondering about is, sure, this is going to improve the user experience,
but in the case of Amazon for example they still have the
infrastructure they still have the the delivering and shipping and the
employees and it's the spider web that makes Amazon so good I don't necessarily
think that is in trouble I think that if anything open AI will just be a front
end plug-in for folks to more seamlessly order the items that they want delivered to their door.
But even with the – I don't know.
I wasn't around for the Silk Road days.
But I think with that, you're reliant on the individual.
It's peer-to-peer, so the individual needs to ship it to you.
I don't think it would be as seamless as it is with Amazon, Next Day Prime and all the warehouses
they have in different states. I don't know if I
necessarily think that Amazon's in trouble.
I'm not familiar with Shopify. I haven't used them but I see those
infrastructures still being quite necessary to keep the experience.
You're touching on something
that's so incredibly important to point out is amazon's primary business and like their
chief revenue is not from their retail it's from aws it's from their servers. And not a lot of people realize that.
Like Netflix runs on top of Amazon.
A lot of these very, very large companies that we think of like behemoths, they run on top of Amazon infrastructure.
So there's a lot of underlying linking that people don't realize.
For example, and I'm just going to throw this out there, it's a little aside, but MicroStrategy. MicroStrategy was acquiring tons and tons of Bitcoin and everyone thought
MicroStrategy was a Bitcoin company. The reality is MicroStrategy was a logistics software company
that was actually running a lot of Amazon and Walmart's logistics. So MicroStrategy was doing
very well as a software company, as being kind of an underlying backbone before they even switched into Bitcoin stuff.
So with Amazon, I think you're exactly right, Noah, is Amazon's key competitive advantage right now is their AWS infrastructure and their fulfillment network.
The physical fulfillment network, until someone comes out with a better
fulfillment robot.
Amazon's got everybody beat.
So, so Amazon becomes an ingredient brand in the same way that Intel's an ingredient
brand for laptops and PCs.
It seems to be the case.
Amazon could quickly, I mean, they've already taken so much business away from UPS and FedEx.
It's crazy.
But they did it from a very kind of backwards, you know, back forward way where it's like they started out with the marketplace.
They built so much traction.
They needed better web services.
So they built their own server farms and then productize
that and then they built out a fulfillment and logistics network and just wait until they
productize that they have not productized their fulfillment logistics network yet and when they
do fedex and ups are gonna definitely have a run for their money yeah so i i think i still think
social engagement could be the next step for them because if
social engagement increases time in app, right, which in turn drives more commerce.
So, you know, if, if this is their first foray into doing this commercial front
end, then it makes sense for them to start to do things in terms of, you know,
what are others buying? You know, you know,
I want to book a trip with three of my friends. Can I do that? You know, so I can I can see some social aspects
popping into this also where you get AI enabled social utility rather than just another social
network. Right. I think we'll start to see some more layers pouring in on this over time. It's
just the beginning. But you're
right in many ways. Will it make Facebook go away? It probably won't. In the same way
that Amazon probably, well, Amazon's different. You're right because of the infrastructure.
That's a real powerhouse for them. So I think it's an amazing announcement. I think it's
a real earthquake shift in many ways.
And it kind of upset my whole notes and flow for this that happened yesterday.
But anyway, it's good to be on top of the news.
We mentioned about NVIDIA's $100 billion AI bet.
They committed to back open AI with $100 billion.
It's from an infrastructure trade.
Stargate is you know this whole
Stargate thing with cloud across the US I also did some research found there are two other cloud
deals Stargate deals one Stargate Norway and then there's another one in Europe as well it might be
the UK so this will be a global play even though most of the focus at the moment is on these hyperscale data centers in the US.
So that's a big, huge shift within the last seven days.
And then what else do we have?
Oh, Google's real world data play.
The real world data sets, they just made a whole bunch of real world data sets accessible through MCP, a bunch of great demographic
psychographic data, not only just for the US, but demographic stuff around the planet.
So if you're a global business, you now have access to global demographics, which, you
know, when you're doing consumer analysis is just a huge piece.
And that, you know, you can integrate the RA rag stuff into it and do a whole bunch of stuff.
And then the other story for the week
was on the humanoid hype cycle.
People throwing rocks at each other,
Rodney Brooks calling,
catch the assistants, pure fantasy,
saying the humanoid robots at Grok,
they're kind of streaming themselves,
but we'll see lots of activity there. know robotics really needs rigor and not just you know it's fancy videos um so it's a whole different
space so that's sort of some of the stories no did you see anything else or anything else in
that that you thought was interesting yeah sorry i didn't mean to unmute everyone. So one of the most expensive infrastructure projects in
tech history you know i wonder if we're being fast enough on and i actually not wonder i it
seems to me that we're not being fast enough on the building of nuclear reactors because you need
a lot of energy to run these models and china is is just full steam ahead, way ahead of us in terms of not just nuclear reactors,
but the rate at which they're building them and planning to continue to build them.
And so I feel like energy is going to be a big problem and it's going to be a bottleneck.
And it almost feels like whoever wins the energy race is going to win a big problem and it's going to be a bottleneck and it almost feels like whoever wins
the the energy race is going to win the ai race and i just wonder if i hope hope the government
and you know the the the people making the decisions are aware of i'm sure they're aware
of this but it just kind of seems like it's not something that's as discussed nuclear reactors yeah i mean
sam altman announced they needed 10 10 gigawatts it sounds like i'm in a back to future you know
i need gigawatts to get into time no he needs 10 gigawatts of compute activity capacity which is
essentially 10 large nuclear reactors or tens of thousands of wind turbines or renewable energy or something like that.
And, you know, so that's, I mean, it's a five-year, $500 billion target to kind of roll through on this stuff.
So it is huge.
But I think you're right.
I think the race for AGI and the race for AI in general really comes down to, you know, sovereign infrastructure.
And clearly China is in the lead in building that and moving it forward. Otherwise, you know,
it's like, you know, we're going to run out of electricity. I mean, then we've got the impact
on the other side, you know, people's electricity bills are going up. So it's more than just that. Yeah. So, you know, a lot of the
the modular nuclear reactors are coming from China, right? So they're they're building a lot
of the technology that we're deploying. Obviously, we have technology come from Germany, technology
coming from Japan. But we really attacked nuclear in this country for years and years and years.
My dad worked at a nuclear power plant for over 35 years.
I interned at the nuclear power plant for several years.
I got a firsthand look at what the infrastructure looks like in this country from nuclear.
And we've been on a declining ramp for nuclear for years because of Greenpeace and the whole not in
my backyard movement of where are we going to put the spent fuel.
So people just stopped talking about it.
And unfortunately, Bitcoin mining was not a big enough mover and shaker to reutilize
the nuclear systems.
But San Onofre had a nuclear plant in Southern California.
They decommissioned that.
The state legislature in California voted to decommission
Diabola Canyon, which was the power plant that I had worked at.
They were supposed to decommission, I think it was 2018.
And then they got PG&E to Pacific Gas and Electric to extend the life of the facility.
And now I think they were slated to close down, I think, in 2026, but now I think they're being reinitiated again to keep it open.
The Palace Virtus plant in in Arizona they produce about six
gigawatts of electricity. Microsoft came in with their Goodyear Center and they
wanted to take up half of that and you know I just think it's so fascinating
like that you know what we were coming full circle now on nuclear power and
people are looking at where
they can put these new plants. And I tweeted out last week, I just thought it was hilarious that
people don't realize the repercussions of putting in a power plant or putting in a data center.
So a power plant uses an immense amount of water for cooling and a data center does the same thing.
So when you put in a power plant, it takes up other resources.
And water is a limited resource.
It's not this unlimited abundant thing.
And people forget about water being one of the most valuable resources on the face of the planet.
Across all societies, water is always a very, very important resource.
You can't live without it,
and most industries can't operate without it.
So I put together a tongue-in-cheek tweet
basically saying, well, because we're going to be putting,
you know, Facebook's going to be putting $600 billion
into data centers,
and Apple is going to be putting $600 billion
into data centers,
and NVIDIA is going to be putting, you know,
hundreds of billions of dollars into data centers. Well, that is going to be putting hundreds of billions of dollars into data centers.
Well, that's going to take up a whole lot of water.
And who competes for water?
Well, it's the farming industry.
And if you have more expensive water, then you're going to have more expensive grain
and you're going to have more expensive feed, which is going to make more expensive chicken
and more expensive beef.
So my thought was people should invest in
beef and chicken futures now as far as into the future as they can because that's where it's all
going. We're going to take a very limited resource and make it very expensive. So it's all pork
bellies and it's interesting because China builds reactors faster than any other country. They
currently have 27 reactors under construction and they've become a net exporter of nuclear technology that they're pushing out.
So, you know, I mean, it creates this weird geopolitical sort of choke point.
You know, I mean, energy sovereignty is really as critical as compute sovereignty.
And I think whoever secures both really controls the pace of AI progress.
And we'll see that.
I mean, it puts the US in a really weird position,
given the US-China rivalry.
I mean, the UK, France,
they've been trying to reduce their dependence
on Chinese made parts in the nuclear area and stuff
by spinning up different things,
but you just can't get away from it.
And now AI is kind of making everyone, you know, with run around with their hair on fire.
It's creating a really interesting set of dynamics.
There've been a few announcements on fusion, but it seems like fusion is always 20 years
away.
What are your thoughts on that, Ryan?
I think if anyone's going gonna crack the science behind that,
it's gonna be an AI model.
So we might be getting the point
where there are AI models that can do the math
and the science required.
Cause we couldn't get to chat GPT two,
unless we had a number one model
and we couldn't get to three or 3.5 without number
two 2.5 and we used all the data out of 3.5 to train four and 4.5 and five came from four uh we're
getting to the the point where the the only way we can train and build better models and better
scientific discoveries is with the previous models.
So we're already far beyond human capacity with this stuff.
We're already relying on AI to propagate AI and to train the next version of AI.
So we surpassed human capabilities a long time ago.
So when it comes to better and more ingenious power production I think
there's a lot of stuff to be done out there that we didn't have the
capabilities before so you know cold fusion or cold vision or whatever we're
gonna do you know I don't know much about the technology personally, but I got to imagine, you know, it could be solved using the next generation of AI models.
Yeah, I mean, Microsoft did sign a power purchase agreement with Talyon, which is a fusion power company to get, you know, start getting power by 2028, which is, I think, pretty freaking ambitious.
power by 2028, which is, I think, pretty freaking ambitious.
Obviously, China and the EU with the ITA project in France,
they've really focused on fusion.
There was a pop recently about someone,
one of the companies actually getting a really good pilot version running,
but again, who knows?
But I can't see any other way of solving it.
I mean, you know, I think Fusion stuff solves some of the cooling issues.
But it seems that we're now in a position where in order to really, you know, advance in the AI space, we have to have a stable and extremely large source of power.
Otherwise it could stall. And the problem with that is it's infrastructure. It's not like software.
We can spin it up and, you know, in an hour.
So yes. Yes. No. So if you think about it, like if we just look at Bitcoin mining, for example, we started out with CPUs, right?
And that took a lot of power. And then we found that we could optimize the mining if we used GPUs.
And the jump from CPU to GPU for hashing was huge, right?
Like to get the amount of compute out of the GPUs
with CPUs was exponential.
And then we moved to FPGAs or field programmable gate arrays,
which are just like programmable chips
where you could build circuitry inside these chips.
And that was exponentially faster than GPUs. And from there, we moved to ASICs.
And ASICs were so much faster than GPUs. You could run one ASIC versus thousands of GPUs to get the
same performance. If you think of it like that, we are still on the CPU level for AI training.
We're possibly on the GPU level now
because we're building very, very specific GPUs
for AI training, right?
NVIDIA completely swapped out.
So the GPUs that you use for AI training
and AI inference are very different
than the GPUs you're going to use for gaming
or general purpose GPUs, right?
So we're slowly iterating on technology.
We haven't gotten to the ASIC world of models yet, right?
Basically the integrated circuits that can do the field or the vector,
Tens of processing calculations. Yeah. Yeah. Well,
we have tensor processing units and people are building those. to do the vector multiplication, you're simply running an electron
through a physical LLM.
And we haven't gotten there yet.
But once we get there,
this power race is going to subside
or alleviate greatly.
Now, it doesn't solve the model training part yet.
We'll still have hardware
to optimize model training and we'll i think we'll get to the point where we're going to have
uh specialized integrated circuits specifically for uh adding to these vector fields um there's
a whole area of mathematics that have has been re-energized from this. But once we start getting into the physical circuitry of LLMs,
the amount of power needed for these things,
that's when we're going to have real humanoid robots
with small battery packs or running off solar panels.
And they're not going to be these big monstrous things
where the entire torso is one big battery
to run the software inside
the head.
Yeah, you mentioned the positronic brain last time we chatted.
Do you think?
I mean, China's-
That's all right.
iRobot.
Sorry?
Yeah, that's from iRobot.
Yep, the positronic brain.
Yeah, I mean, China's been doing stuff around artificial neurons, right?
They're building brains that have like, you know,
100 billion synapses with custom neuromorphic chips.
I mean, are we starting to see the very beginning of Asimov's robot brain?
Oh, absolutely.
And once again, we couldn't get there without LLMs.
So, you know, all these internal models that people are running on Grok and OpenAI,
they're being used to build the next generation of advanced circuitry.
NVIDIA, they already came out and said, yeah, the current chips and the GPUs that they're building
and their current product offering was designed by AI.
This is far beyond human engineering capability at this point.
Yeah, it's pretty crazy.
I mean, we're seeing such fast advances.
And you're right.
It's like the AI is now, it's like, you know, the Oregon, the snake eating its own tail,
where the AI feeds itself, increases
itself, then builds a better AI, and then we move on from there.
And I can only think that that's going to happen faster and faster and faster.
One thing that also happened in the last seven days was a whole bunch of stuff on, we had
the chat to announce yesterday, but organic commerce and instant checkout stuff happening um that's accelerating visa was making announcements on a new platform
today that was in the newsletter um lots of stuff happening around a genic commerce as well um and
and i think that's going to really create some interesting dynamics in retail.
And we're still seeing security issues around that. I mean, Ryan, you know, we've kind of thrown around
a lot of stuff around security,
but it seems like we're building this,
a lot of these pieces so fast and implementing them
that it, to me, it feels like we've got to start seeing
some cracks form around security.
You know, there was before
the Mexican sales are using LLMs to make their networks more efficient.
I mean, surely we'll start to see a lot more around that.
Well, I have an article here that someone posted in one of our chats of Cloudflare
introduces NetDollar to support a new business model for AI driven internet so welcome cloudflare
to the the cryptocurrency revolution with your net dollar a little late to the
game but they think they're providing a internet dollar for agents to be able to
quickly pay for commerce in a secure way.
So I think you're exactly right.
Everyone's kind of narrowing in on this,
but netdollar.cloudflare.com apparently is their big move into web payments.
So I thought that was very entertaining.
Yeah.
And then on the bleeding edge of hardware, there's a bunch of stories around quantum, you know, the hypes around quantum and quantum supply. I mean, qubits, there was announcement on a 6000 qubit commercial system, 6100 actually, qubit processor shattering the record.
record you know so it's interesting that we're also seeing events in that area which might also
overlay at some point into the ai space um so yeah this this stuff is must just moving really fast
and for me it brings up stuff around governance california um just put out a frontier transparency
bill um we're seeing regulatory stuff come out of the u.r the eu
um and interesting there's a lot of a lot of interesting articles coming out of china
on ai regulations and pieces around that um but one of the biggest stories around the whole thing
that kind of pulls it together is is that it they can't keep. The technology is moving so fast and the implementation is moving so fast,
you know, in areas like, you know,
companion, AI companions
and all the issues around that.
Do you think that it'll never catch up,
you know, Ryan, or I mean,
is governance now basically
always going to be a lagging thing?
Well governance is reactionary just inherently. Very very few leaders or politicians nowadays
lead from in front of an issue. They often are very reactionary and they do it as a way of preservation to stay in power.
Very, very few, you know, the ones that aren't reactionary, the ones that are typically at the forefront and are moving before something are typically like the CIA operatives or, you know, military officials are the ones that you know our
task was staying ahead of things that's why I think I think governance
inherently is always trying to catch up it's always reactive it's very very few
systems nowadays are proactive in governance right and you know we're
starting to see a lot of politics start to pour into the whole
chip hardware stuff as well i saw an interesting article where the u.s is pressuring taiwan
to move production to the u.s otherwise it won't support it which i think is really
a kind of a critical strategic decoupling as they they say. You know, it sounds like China's pushing the U.S.
And so the U.S. is trying to find some way to go, yeah, Taiwan,
maybe you should join China.
You know, at least we're starting to see this starting to bubble,
you know, with the whole move with AI, chips, power,
and all the other drivers that seem to be making
a real push around the planet.
Yeah, I thought the moment Putin went into Ukraine, I thought China was going to make
a move on Taiwan.
They seem to have been positioning themselves towards taking over certain regions around the Chinese borders.
So taking over parts of Nepal, Tibet, taking back Hong Kong.
They have control of the Panama Canal.
They have, you know, moving into the South China Sea, you know, that was kind of threatening to Vietnam.
They've really kind of been expanding.
And a lot of people in those different regions are very much aware of China expanding.
But I feel like as a global community and in global news, it's not really focused on very much.
in global news, it's not really focused on very much.
So, you know, China taking over Taiwan, you know,
they don't necessarily need to take it over.
I think in their minds, Taiwan already belongs to them.
I think in the minds of the CCP, I'm sorry, CCP,
they believe that Taiwan is China.
So it's like they're just waiting.
So I think U.S. pulling chip manufacturing back mainland is very strategic.
It is in that regard a more of a military proactive play
where it's like let's sure up our chip manufacturing and let's sure up our
our ability to continue in the ai race what's interesting is we're still getting as we said
earlier the the nuclear reactors and a lot of the a lot of the technology for power production, like solar, is coming from China still.
So until we can solve that.
Yeah, yeah.
One little interesting piece, you mentioned Tibet.
One of the reasons China went into Tibet
is because they're often called the third pole.
It's got glaciers in the plateau,
hold the largest supply of fresh water
outside of the Arctic and the Antarctic.
And so the reason they took over Tibet, I believe in many ways, was to support the Chinese continent with water.
And, you know, you look at the major rivers, the Yangtze, the Yellow, the Mekong, the Indus, the Ganges, Brahmaputra, they all originate in Tibet.
And so Tibet really was um an infrastructure play
for water you know wild wild that's wild I mean I really that's why I think they really took it
over I mean otherwise it's like why there was no reason for them to do that yeah and so it was water
absolutely water so it was a really I think it was a really strategic, you know, domestic sort of focus to get them, you know, and it's been part of the South North Water Transfer Project.
There's massive hydropower happening up there at the moment, which is really annoying India and Pakistan because they're looking at damming it to generate massive amounts of energy.
So not only water, hydropower, you know, that's also getting
booming up there as well. So that's and I you're probably right about Taiwan. I think
maybe that's more just, you know, more political than, you know, infrastructure resource, you
know, type stuff. But I just thought it was interesting, the recent push from the US, basically pressuring them into moving 50% of the chip production or loose protection.
That's a really kind of bold statement.
And so it's interesting watching the dominoes kind of ping pong and fall around the planet.
Yeah.
You mentioned the Ganges River coming from the glaciers up there. So a fun aside,
there is a legend that the goddess Ganges blesses the Ganges River and that if you travel to the
mouth of the Ganges River where it actually comes out from underneath the glacier, it actually comes
out of like milk white. It's like glacier milk and the the legend is that if you drink from the
mouth of the ganges River or that's right the base the source of the
Ganges River that you'll live to be a hundred years old so so there's some
folklore for you if you ever want to put something on your bucket list uh maybe uh travel to the source of the
ganges river um so you know with all this stuff it's clearly sovereignty you know we're seeing
the rise of sovereign ai the rise of sovereign resources and the rise of sovereign power
becoming absolutely important you know from canada u, UAE even, stories pop up across Southeast Asia.
Lots of public sector pilots are getting built and sovereign builds.
There was a recent SAP partnering to build a sovereign cloud AI for Germany.
You know, so Nation as a platform suddenly is kind of popping onto the whole thing.
And, you know, when I heard the initial stories about sovereign AI, I was like, really?
But now you can see it, you know, it's starting to become real in terms of the infrastructure builds that are occurring.
And maybe Stargate is just a way to overlay on top of that.
I don't know.
It's going to be interesting where nationalism is kind of competing across, you know.
And the conspiracy theories are going to run rampant.
Yeah.
Oh, yeah, if not already.
I mean, people are bitching about Grok AI being too woke, you know. Yeah. Oh, yeah, if not already. I mean, people are bitching about Grok AI being too woke.
You know, sovereign AI is just going to be a hyper extension of that.
Israel was complaining about AI and trying to work on AI to be more positive towards
them.
So, you know, it's where politics is suddenly going to go against AI LLMs.
That's going to be an interesting mix.
And it's going to be who owns truth and where does the source of truth come from?
Right. 1994 all over again, you know?
Yeah. But real quick, I want to jump in a different direction, more towards vibe coding.
Real quick, I want to jump in a different direction, more towards vibe coding.
So I find it hilarious with all the new Claude models that are coming out and people using
Cursor and people using Lovable and using all these different tools now for creating code.
There was announcement around this new Droid system that could one shot software.
And I saw this guy post around Claude 4.5 Sonnet.
It was so great.
It was Claude 4.5 Sonnet just refactored my entire code base in one call,
25 tool invocations, 3,000 lines 12 brand new files it modularized everything
broke up monoliths cleaned up spaghetti none of it worked but boy was it beautiful that's it
pretty yeah yeah the comments the comment sections are just stellar, you know? Yeah.
So, you know, Vibe Coding and everyone with an app idea
can create apps faster than ever.
And it essentially is the end of the junior engineer.
That's the interesting thing is the junior entry-level web developers have been
replaced with a lot of the software. I think the senior-level architects and senior-level
developers are still safe for now because they can go in and clean up and fix a lot of, you know,
this vibe coding stuff that doesn't work. But it's catching up very quickly. You know, this vibe coding stuff that doesn't work, but it's, it's catching up very quickly.
You know, eventually there will be, you know, systems that can test with even senior engineers.
You know, we got to remove two levels of engineers inside of a year. So it's going to be interesting.
Yeah, but we talked about it last week that it's one thing to write code.
It's another thing to build systems architecture.
You know, the lack of AI and systems architecture and support for that means I can build an app.
And then, you know, if people want a whole bunch of changes, it could just break it.
So, you know, how long do you think it'll be before we actually have good systems architecture,
both front end and back end?
I think two years.
I think within the next year, we're going to have replaced our senior developers.
I think two years, we're going to be replacing the architects and the scalable systems guys.
I think within two, two and a half years, we may be even, you know,
replacing all the principal engineers and the highest level of engineering that we could have.
Really? You think it's going to happen that quickly, Ryan?
Oh, yeah. Yeah. Because remember, this is like thought process. This is,
because remember this is like thought process this is it builds on itself and
it accelerates so we will see these systems accelerate as far as how fast
they're able to develop how fast they're able to scale how fast they're able to
monitor things we're gonna have DevOps agents you know know, because remember, all of our web services, even through AWS and Google Cloud and Azure, they are all, you know, program, programmatically propagated now where you can build out scripts through Terraform and Docker containers, and you can write code that builds servers that deploy systems.
You can write code that builds servers, that deploys systems.
So all of this is codified now, if that's a word.
So if it's codified and you can do it with code, then an agent can take it over.
So we have all the tools in place to automate everything.
Now it just comes down to agent systems that have been trained on monitoring, building, scaling, and changing or adapting data schemas in real time as needed.
Yeah, I think you're right.
I think what we're seeing is a bubbling up effect where like on the front end, LLM's design tools are working and doing stuff on the back end.
Code agents are doing stuff you know with
databases stuff the real thing is the integration complexity so while i think we'll see the ai
handling these separate bits at some point they'll start to handle that complexity of the overall
design how the overall services the apis the databases, the front end, everything starts to fit together. And two years is probably about, yeah,
I think we'll start to see that.
And I think certainly within three to five,
but the humans then, what do humans end up doing?
Are they just becoming architects?
Is there any, I mean, Ryan,
what do you see the human component as these things bubble
Where are they left behind?
Do they just become, you know, conductors of a vast audience or what?
I believe the humans will no longer be developers, but they'll be composers.
They are no longer playing the instruments in the orchestra, but they can direct the instruments of the orchestra.
Now, it doesn't mean that you can't have an artificial composer as well, like a lead agent doing everything that you just interact with. But I think humans,
by their nature, still like being creative. They still like imagining something and seeing it come
to fruition under their direct guidance. So I think there's a lot of humans that will view it
as an art form or view it as a means of self-expression to curate and to compose with these AI systems.
Yeah, I think we'll see AI conductors, you know, like a bunch of musicians.
You'll see, you know, instead of junior developers or senior developers,
we'll see AI junior conductors, AI conductors, you know,
the full stack people will turn into systems, you know, AI conductors.
And then basically you'll get business tech strategists on top of that saying, OK, here's the overall city I want to build.
And, you know, these are the needs of it. Now go build it.
And then this AI orchestra people will start to do the needs of it. Now go build it. And then this AI orchestra, people will start
to do that sort of stuff. So you're right. I think we'll still have human involvement, but it'll be
again, a level of abstraction on top of what we're doing now.
Yep. Interesting stuff. Yeah. And also a bunch of stories about people who did a whole bunch
of vibe coding. I mentioned this last week, people who did a bunch of vibe coding,
and then they couldn't fix it or it broke or they couldn't add features.
And so they went to the, quote, real developers to get help.
You know, so, you know, again, if it generated spaghetti code,
which I was pretty good at at my time, lots of spaghetti code,
you know, it could almost be impossible to pull the
whole thing apart um yeah interesting interesting stuff a bunch of stories on the military occurring
getting involved in ai this week um it's interesting of all the news categories i tracked
it's one of the only ones where, surprisingly, has the least amount
of PR.
So it's, I mean, seriously, it's really clear that there's a ton of, when you go searching
for it, you can find it, but there's a ton of activity occurring in the defense and military
side.
You know, Netanyahu, the Israeli PM, in washington this week and he had a whole bunch of
meetings with with the ai tech people which wasn't really kind of made obvious but it went on and
there's there's tons of stuff i mean we see ai enabled drones we're seeing ai enabled military
applications but there is a ton of activity occurring in it with different major corporations
getting involved around it so it is an interesting area to track if anyone has any comments around it
or anything like that please reach out to me because there's stuff happening in that area
and like I said surprisingly no one wants to talk about it because obviously it's a huge sovereignty slash national security
advantage of people around that. Mind your biz. No, any comments on that, Ryan? I mean,
have you had any interaction with the complex? Yeah, I can either. Yeah, i go by my new business thanks for inviting me to comment on
this list so as far as uh as far as me having direct any kind of direct involvement with the
military industrial complex i think the standard answer to that is always supposed to be i can
either confirm or deny but no for me it's a solid no um yeah unless Unless they're utilizing something like yourself where you're
publishing meaningful
commentary and meaningful digest
of some of these kinds
of thoughts, you could have a lot
of your work skimmed and not know
that you've been made useful
by some of these
groups. I'm always
cognizant of that, especially
Noah and I have joked about this in the past.
I've really focused a lot over the last five years in particular on the privacy aspect of
cryptocurrencies, various mixers, mixing technologies, ways to make cryptocurrency
tokens more fungible, which in a circulating economy, right, in a native crypto economy,
is essential. Provenance is important for certain things,
but most of the rest of the time for certain types of interactions, you want something to
be fungible and as close to cash as possible. And I think that we're going to see that that's
a reality in AI-based economies where agents are just swapping tokens in order to be able to power
transactions where value is exchanged. So I'm cognizant of the fact that
some of the commentary that I've made over the last five years about privacy, that a lot of the
digests that I've written or the media appearances that I've done, the commentary that I've made
publicly, any of this is up for grabs from an OSINT perspective to just scrape, aggregate,
and then turn into some kind of a bio or a dossier on, you know,
what is this guy's thoughts on these topics? So any one of us, for all we know, is already
working with the military industrial complex. We just don't know it yet. I'll save that to add a
little bit, wrap a couple layers of tinfoil onto the conversation. But as far as there being an application for defense, of course, intelligence
at the lowest possible price is always something that's going to interest both industry and
government. So yeah, there's got to be the absolute bleeding edge has got to be in a bunker somewhere
that is, you know, that's in an EMF proof container.
So they hope it can't escape.
Well, remember, you know, the Internet was a military project.
I mean, DARPA started the Internet.
So decentralization really came about because of them.
You know, so there's no single point of failure when you're in a war that's, I guess, super important. But I think what we're seeing with AI is, you know, and they talk about in the military, they talk about asymmetric advantage. You're talking about clever AI can essentially allow smaller nation states to, you know, punch above their weight, as they say, you know,, I mean, we saw in the Ukraine, I mean, you know,
You know, so there's no single point of failure when you're in a war that's, I guess, super important.
this small, tiny country is beating the crap out of what was supposedly, you know, one of the greatest military forces on the planet,
though now we kind of maybe know better.
I think AI really maybe gives excitement to those smaller nation states that,
hey, maybe with AI, maybe we can either get a significant advantage or do stuff.
So it's no wonder there's a ton of stuff happening in it.
I remember I had a relative of mine who was a colonel in the U.S. military
who his job was logistics.
And, you know, we know AI, you know, you throw it at logistics
and it's going to make it
far more you know efficient transparent frictionless so lots of advantages i think across
every application for them um but it's it's just interesting as i look at the headlines i kind of
sometimes i have to go outside and pull some stuff in because it's just not getting reported on
um i think um mind your biz, if I remember rightly,
I think you and I talked, I made the comment that anything in the military,
typically from a technology point of view,
is about 20 years ahead of anything we can really see.
So it would be super interesting to kind of peek into it
and to see what's happening.
Yeah, let me volunteer this little tidbit.
I suspect that based on what you're saying, Lewis,
and I totally agree, it's an objective fact
that anything the government is currently working with,
and not just the Skunk Works at Lockheed,
but genuinely, internally, that they already have adopted
for internal use, that it's so many years ahead that that to us it would look like magic i don't think it's beyond
the pale to to think that um that i mean we have things that they're already working with things
like quantum teleportation and uh and some amount of limited time peering if not time travel
there's stuff that would blow lines so when you you add to that, the fact that, yeah,
we're just now becoming publicly aware of there being non-human intelligence
in the form of silicon.
Yeah.
Yeah, there's probably something that to the average person today
would look like AGI or would look like ASI
and would just cause us to never sleep again.
Yeah.
Yeah, I mean, there's been – in some ways, maybe, you know,
the advancements with AI has allowed the, you know,
commercial side to maybe in some ways catch up with some
of those advanced technologies or have the ability in a medium
to short term to catch up to some of those
because you know if we can apply advanced AI to commercial problems we can do some leapfrogging
so maybe there's concerns on that side of it that you know uh the all the secrets they
have you know in terms of some of their technologies is now we're now, you know, at the level of commercial enterprises within
arm's length of potentially reaching into that. That could raise some interesting stuff
around it. There's been a bunch of stories around big AI firms pumping money into world
models as well, which I think is a really interesting aspect. You know, world models where you're actually,
which actually would end up supporting robotics
and logistics planning,
I think is also another huge area
of potential commercialization
and, you know, opportunities for enterprises.
I've said before, I believe that robotics,
I think, I believe we'll have, you know,
robots in the home in the next two to three years. I really think that's going to start that. We'll
start to see that, especially with what we're seeing in China and their developments in robotics
and some of the other advancements that are occurring. It's just happening so fast.
Go ahead, Ryan. Back to our good old positronic brain stuff.
Yeah, no, the idea of the military industrial machine is fascinating to me. So two notes on that.
One, I actually have a startup that I'm working on that I'm applying to be a DoD contractor for R&D.
And when you're actually filling out the forms
and they have all the different basically sectors and services
that you could be applying towards,
it gives you a little bit of insight on the different types of technologies
that they're focused on.
And I just had no idea.
It's like everything from like nano biological warfare
is like a, you know, that's an actual category
just in their application, you know?
But there's so many, you know,
quantum warfare, you know,
everything from, you know, quantum cooling logistics to,
you know, next generation space propulsion. There are so many categories just on the DoD
contractor application that kind of, it tells you what's on their radar. And you know, if it's on
just the general application, you know that there's a lot
of people already working on this stuff so uh it's fascinating just from that view and then
my second point and second note is that uh consumer facing products are typically
are typically five and sometimes even 10 years out after they existed.
So, you know, we're seeing, for example, that the TV that you have in your living room,
the technology for that TV has existed for probably well over 10 years,
but you're just now able to buy it in a consumer format.
but you're just now able to buy it in a consumer format.
A lot of the technology that we see in cars,
a lot of the technology we see in aircraft,
going into a consumer format,
the technology has existed for a long time,
it's just now being released.
So if you ever go to CES, which is a consumer show in Vegas,
they were showing technology show in Vegas,
they were showing technology back in 2013, 2014
that were just now seen being sold at Costco.
And even some of it were not,
we still haven't seen it.
They had transparent displays for windows
that turned into screens for over a decade now
where you see a grass front cabinet in your living room.
This was actually in one of the Panasonic movies. You walk into this living room mock-up and it has all this beautiful glass front cabinetry and then all of a sudden all those glass front cabinets turn into screens.
And very high resolution screens.
Yep. And that was technology.
They have
robots and
crazy drone technology.
Like, you know,
three-dimensional displays, waterproofing
technology, all this stuff.
You're just breaking up a little bit.
Oh, yeah.
So, all that to say,
technology.
Is it just me or is Ryan
Ryan rugging bad?
He sounds robotic.
The U.S. government doesn't like
what you have to say right now.
He said too much.
Ryan, they're on to you.
I mean,
it's like
I mentioned it before,
this bubble up effect.
You know, we're not seeing, you know, massive jumps on stuff like, you know, superhuman AI occurring and stuff like that.
What we're seeing is AI getting applied and there's incremental gains, whether it's through workflow, you know, handling low hanging tasks, boosting capacity.
It's kind of bubbling across that.
And then as they get better and better at that,
then it's starting to kind of link together stuff.
And so I think, but that whole effect of handling all these different low hanging fruits,
it's that impact and effect is,
I just see that speeding up faster and faster.
Go ahead, Ryan, your hand's still up. Or are we going to bubble again? Oh, I think I just see that speeding up faster and faster. Go ahead, Ryan, your hand slip.
Are we going to bubble again?
I think I just put it down.
Yeah.
The point is there's so much technology by the time it hits a general retail
consumers.
It's,
it's already antiquated from a research and development perspective.
You see that as soon as he stopped talking about the stuff he wasn't supposed to be
talking about, his voice got. Yeah, we're getting, yeah, yeah, they're kind of interfering with the
network. Actually, funny story. I had a, I was doing some events at the White House and it was in the very, very early days where I had a cellular modem
and I had my laptop and we were setting up the network.
So I was downloading network software and these guys bust in
and, you know, with the big dark glasses and the jackets that were too big
because they were hiding various military things underneath it
and told me to stop.
And I went, I can't because I'm setting
up the network for this event. They said, you can't do that. You're running a separate network
in the White House. And I said, look, and luckily the White House webmaster, who also funnily enough
was part of the military, came over and got in their face and said, no, no, we really need to
do this. You guys need to go away. And they said, you have two hours and walked
out and literally two hours to the minute I'm working my laptop. And all of a sudden it literally
the screen flashed and, and it went and it started to try and reboot, but it couldn't reboot.
And so, yeah, you hear stuff like that, but that really happened to me. Luckily I had a backup
laptop and I just went to that and then did a landline and put some stuff down.
But yeah, weird things happen when you start talking about military and other things like that.
Let's see.
Another good story was on clinical AI, lots of stuff happening in healthcare.
Like I said, a lot of it's on the, you know, logging in fruit stuff. And they're doing things around making, you know, appointments more
efficient, making doctors more efficient. There are some interesting emerging clinical AI trials
that are popping up that are on the radar with interesting things around that. So I think, again, we're seeing,
while we're seeing sort of the flashy hype AI stuff happening
in healthcare and medical stuff,
there's a lot of groundswell activity as well,
where like, for example, Mass General Brigham's using generative AI
to speed up screening and enrollments for heart failure clinical trials.
So lots of stuff happening on screening on clinical trials,
providing faster, better documentation for clinicians and patients.
We're seeing tremendous stuff happening on that.
Specific stuff around diabetic stuff popping through the whole thing
and things on liver disease trials, who would have thought?
So, yeah, you're starting to see a lot of interesting applications
of AI kind of filling in the cracks.
It's almost like it's becoming the cement between all the bricks
that we currently have and replacing a lot of the grunt work.
Anyone have any – I think it was about two tuesdays ago we actually had a doctor who
was building uh apps with ai um anyone else heard of any other uh ai and health care or even just
anything else you want to chuck into the mix that we're chatting about today i think a lot oh yeah
go ahead yeah there's plenty of anecdotes all over the place of,
of doctors just saying that right now they,
in some cases it's kind of a love hate thing where general practitioners,
they're hearing a lot of their patients come in and say, Hey,
like I just spoke to chat GBT and research this issue that I had.
And can you answer these questions?
And they're being stymied.
They're actually, research this issue they had and can you answer these questions and they're being stymied they're
actually they're actually being completely stumped by their own patients walking in and having done
more research in a short amount of time than they can possibly have done on a given specialty so i
mean they're general practitioners so they shouldn't be expected to to become specialists in every
aspect of medicine overnight but with tools that, you know, like search-enabled LLMs, almost anyone can, right?
It will affect, but on steroids, you know, on horse steroids plus crack cocaine, right?
It's like, it's not, it's really difficult to, you know, to properly quantify just how
much farther ahead the average person can dive into certain types of
research, especially when they properly follow the breadcrumbs left there. So if a person knows how to
read an abstract on PubMed, you're already light years ahead of the average Joe that doesn't know
how to just read research when it comes to any kind of medical trials or when it comes to a some kind of a
controlled study of certain types of medical intervention so now imagine you have even
something as pedestrian as chat gp providing you 10 links to relevant trials that are on pub men
where you can properly go and vet the research and say, oh, yeah, well, Chachi didn't
hallucinate this and it didn't lie. It just gave me breadcrumbs to go do my own research.
So you have these people coming in with way more targeted research at their disposal. And so,
yeah, you have these doctors, in particular, again, general practitioners saying like, yeah,
I'm really uncomfortable now because every other person who comes in is more of an expert on their symptoms than I could ever hope to be. So it's been disruptive.
Totally. I mean, yeah. And the thing is they're feeding their lab reports into ChatGPT. There's
been a bunch of stories around that. You know, in 2020, there was a paper that came out of JAMA
saying basically any general practitioner would need 25 hours per day just to keep up.
And specific to the medical knowledge, it's doubling every 73 days.
You know, so basically medical practitioners have no hope of trying to keep up.
But you're right, it creates this interesting dynamic now where, you know, if you want to be your own advocate and you can follow a breadcrumb trail through and you can reasonably put two things together, it really creates interesting dynamics between patients and doctors.
And whether, you know, where that's going to go, I'm not so sure.
I mean, maybe we do end up with an AI doctor, you know,
or at least certainly an AI assistant doctor
that then becomes a dialogue between the three with the AI.
I'm hoping along those lines, just to interject a little bit there on that note,
I'm hoping along those lines that before long, in the same way that we're currently seeing a bunch of those real-time, I think, 120, 140 language, real-time translation earbuds that exist now, AI-enabled earbuds, that will have a system like that where end users, like the average person, can just plug into their, whatever, their first aid kit,
have attached to their first aid kit a very simple first aid coaching program that's AI-enabled or AI-enhanced.
So that, you know, as we, let's face it, millennials, they were raised in cotton wool.
They don't have the rough and tumble scars Gen X does.
And I'm not saying that to, you know, to cash fade.
It's just a fact, right?
Gen X has more scars, more battle wounds.
And I think it's probably a function of whatever our parents' insurance is being better during that time.
We were beneficiaries of a flush economy.
I won't say it's anything wrong,
you know, intergenerationally, just, you know, that's just how it panned out. But all these,
you know, anyone younger than, say, millennial age, generally speaking, they don't know what
it's like to reset a broken finger at home. They don't know what it's like to potentially,
you know, apply sutures at home
or to deal with a head wound that looks really dangerous, that looks really scary and ominous
because it bleeds profusely, but actually is not remotely life-threatening. So to have
an AI assistant in the home that is purely trained for first aid is going to vastly decrease
the number of emergency room visits that are necessary
when it comes to those sort of fringe cases where people, you know, today,
they actually, they kind of need some financial relief from medical bills.
So AI really could be a path forward for that as well.
Yeah, I mean, you know, a long time ago, we saw the introduction of automatic electronic defibrillators you know home home versions and you know and then that
technology became cheaper you know and it's not only a problem with you know
senior citizens it's a it's problem with young people you know because it's not a
plumbing problem it's an electrical problem so you get a baseball whack into
the chest it can stop the heart and the he is the only thing that's going to kick it back. So maybe the Star Trek AI tricorder, you know,
might be the next thing to get introduced into homes. I mean, your watch can measure your,
you know, ECG, your oxygen rate, a whole bunch of stuff. There's no reason to believe that,
you know, three, four years from now, you know, that same technology on your wrist can not only diagnose stuff, but maybe I don't know if it'll jolt you.
But certainly it could take your blood.
It could do, you know, blood analysis, analyze any other.
Those precious bodily fluids and come up with results so that at least from a diagnostic and,
you know, thinking ahead sort of view, create, you know, hopefully make us healthier and live
longer. So maybe we could be, you know, we don't need to go to the base of the Ganges, Ryan, to
drink the Soma and live to 100. You know, maybe AI is going to help us to do that. But I think,
I think Chachi would advise you against drinking anything out of the Ganges
today.
Yeah, that's true. Anything lower Ganges, definitely true.
All right. You want, you want another fun fact about the Ganges river?
It's a fun fact. As long as it's AI related.
All right. Well, it could be, I guess.
So, Ganges River, don't ask me why I know this stuff.
I just read so much random stuff thanks to AI now.
But apparently, when ship captains were trying to circumnavigate the globe, and they were going from London all the way down around the Cape of Africa, all the way to Australia and back, they would always stop off at the Ganges River to refill all their barrels of water because the barrels of water from the Ganges River was the only water that would last the trip there and back.
And they could never figure it out. But the folklore was the goddess Ganges actually purified herself so some
microbiologists came across this and they could never figure out why there
was never a mass outbreak of disease or you know any type of you know like
serious ailments from the Ganges River because so many people swim in the Ganges
River. So many people dump sewage in the Ganges River. It's like something like a billion people
live along the banks of the Ganges River. And yet there's never been a massive pandemic outbreak in
India from it. So microbiologists actually studied the Ganges River water and they found there's actually a microorganism in the water that will multiply based on the food source.
And they live off of bacteria.
So the more the bacteria goes into the water, the faster this microorganism propagates.
It eats up all the bacteria and then it dies off again from the lack of food supply.
So it literally purifies itself.
So the Ganges River water somehow, because of this microorganism,
is actually one of the safer bodies of water on Earth.
That's OK. That is a great fact.
I also just found that India launched an AI-powered initiative
under the national mission for clean ganja,
using machine learning and computer vision to monitor pollution levels, illegal dumping and flow dynamics of
the river in real time. And actually, there's a whole bunch of rivers that are doing that now.
We're seeing a massive upswing because of the environmental issues with AI. We're also seeing
AI getting applied to environmental and climate stuff. And if you look in the environmental section of the newsletter that I push out,
you'll see lots of interesting stories bouncing back and forth. One of it's using too much
electricity, it's using too much water. And then on the other side, people are using it for
conservation, using it for reducing pollution, detecting pollution, using satellite information
to detect pollution. So the environmental stuff with AI is really super interesting because it's
a lot of conflicting back and forth around its implementation and its usage. You know,
will it help us to get to a cleaner planet? I think in many ways it
will because it allows us to kind of do a lot of really deep problem solving around
it. But hopefully in implementing it from an infrastructure point of view, it won't
suck up all the water and use all the electricity. Interesting competing dynamics there not sure how it might break out but you know it's obviously
going to impact all the sovereign ai trends that we're also seeing noah anything else you you've
been interested in this week and seeing popping up that we can i was just thinking of the ganges
river a bunch of the guys were up in Varanasi. Yeah.
It was like 2013.
And they went and they bathed in the Ganges River,
and everyone thought they were crazy,
and everyone was speculating on how sick they would get,
and they were completely fine, which I never looked into why,
despite all the dead bodies being dumped in there and the ashes being dumped in there and the trash and the people and why people could go in there and swim and be completely fine, as Ryan was saying.
I didn't realize there's research done and there's this microorganism that basically functions as a filter. It absorbs all the bacteria and then it functions as a filter.
It absorbs all the bacteria and then it functions as a filter for the river.
Yeah. I got to take back.
When my, when my pet dies, I gotta, I've got to go dump him in the Ganges.
Now I am, I'm morally obligated not to know better guys.
I just posted a link in our AI chat.
I don't know if you guys saw that.
So OpenAI just dropped just like, what, 40 minutes ago?
A teaser for their new Sora app where they're using the Sora 2 video generation and audio generation.
video generation and audio generation, it's insane.
It's insane.
Like everyone listening, if you guys get a chance,
after this is over, go check out the new OpenAI
Sora announcement.
Their video spot that they dropped on X
is all generated by Sora 2.
And it's quite ridiculous.
You could never tell that it's not real yeah I mean yeah there's there's there was another thing on
YouTube similar to that where open source is now at the point with video
where you can very I mean basically you put up a video of yourself talking and
then you have a photo say of I don, I don't know, Taylor Swift.
You can throw that up and it will take the photo and replace you with her voice talking in the video in real time.
And it does it very quickly and it doesn't take much effort.
Um, so yeah, I mean, it's crazy what's happening both with the proprietary AI stuff and the open source AI stuff.
Um, you know, I think people have kind of said, you know, they see a battle between, I don't see a battle between open source and proprietary.
They're both going to exist.
They're both going to go forward.
They're both going to exist.
You know, they have their own reasons.
Um, but it's interesting seeing the race at least at the video level on ai between the two of them um and the
things that are popping up the asura 2 is seen as a competitive move versus google's recent vo3 model
which already offers video plus audio so maybe it's while it sounds like it's pretty amazing, maybe it's more of a
competitive shift. What do you think, Ryan? Is it more trying to meet the competition
with Google? Is OpenAI going up against them on the video stuff?
Ryan Duffy It seems like it. You know, what's interesting is OpenAI seems to be trying to leverage their technology to take on so many different industries all at the same time.
Because I think they realize their internal models are really, really good and they're very disruptive.
And they thought they had kind of a secret sauce and they had more time than they actually had because you know sam altman was uh alluding
to the power of chat tpt5 uh like a year ago or maybe it was a year and a half a year and a half
ago right when like four or 4.5 came out they were talking about how five was so disruptive
that they're going to roll it out in pieces. What they did not anticipate is Elon coming along
and building Grok so quickly or China data scraping, uh, chat TPT and these other AIs
to come out with their models. Um, so it does seem like open AI is in a race against all these
other companies in a bunch of different, uh, categories. So, I mean, I don't know. I don't know how you keep up the breakneck
speed of scaling like they're doing, but they, I mean, they have billions of dollars in revenue
and hundreds of millions of users. So they're becoming a rightful juggernaut and one of
the world's largest companies very, very quickly.
I can only see this having a profound impact on Hollywood I mean Hollywood
studios had that you know the historical advantage you know they had the
expensive cameras the actors and all this other stuff I can see the rise of
micro studios coming up and this and basically their scale advantage is just
totally collapsing I mean you can now have synthetic extras, right?
You don't have to have the big, you know,
go around locations, you know,
background economies now collapse,
you know, storytelling formats collapse,
you know, our attention spans short anyway.
Maybe, you know, we'll start seeing more micro cinema stuff
occurring on YouTube, Instagram and so on.
So it's, and you know, with the ability to essentially create fakes so fast,
you know, are we seeing sort of potential synthetic piracy?
There's a bunch of stories in the entertainment section of the newsletter
on this new AI star, and people are bitching the hell out of it.
But, I mean, it's just going to be the beginning.
We'll start with AI stars, which are a composite of others.
So it'll reduce, I guess, maybe copyright, trademark stuff issues.
We thought vibe coding was big, right?
Where people are creating code and apps where once they create it,
they have no
knowledge of how to tweak it or fix it behind the scenes so they're doing complex engineering using
these models think about how much faster they're going to adapt to doing you know, whatever. It's, yeah, I think Hollywood, you know, and actors, OnlyFans, porn,
like all this stuff is going to be so disruptive because, I mean,
it lowered the bar for human creativity and expression.
Yeah, I think we're seeing the creation of what I'm calling
narrative architects, right, world builders, right?
So, I mean, look at J.K. Rowling.
Look at George R.R. Martin.
Look at James Cameron, right?
They no longer need Hollywood.
They don't need studios anymore.
They can now, with this technology,
easily in the next two to three years, I mean,
imagine a J.K. Rowling type of fantasy world
that can get built, produced, and launched
with all the richness and, you know,
the storytelling that's coming out of their brain.
I mean, one person created all that stuff.
You've got to remember.
Now with this technology,
they can now create a whole world
that they can implement and
launch really quickly.
Video games.
Sorry, go ahead.
I was just saying video games.
Yeah, no, I agree with that too.
Just as a side comment, video games are, depending on the complexity of the games, the customization of them is going to go so much faster with AI enabled coding tools or completion tools, I should say. things that we used 10 years ago to make things faster, there's still a really good call for using
templates, I think, and using AI systems for just faster customization and faster iteration.
In the same way that, like you're saying, somebody like JK Rowling, we want to credit her as this
auteur where she imagined the entire world at Harry Potter. And that's true, but she didn't
imagine it all at once.
And it wasn't like a fever dream where, you know, whatever, she caught cold one day and the
following weekend, she had all of Harry Potter fully ideated and written down, right, across
multiple notebooks. And then she had a perfect manuscript the week following. It was an intense
process, right? And AI, depending on the types of generative AI that you use, they can accelerate
some of those processes, but you still have to do error correction as you go. And even as someone
with a vision, you still need to guide it. So there's definitely going to be a need for frameworks,
even just storytelling frameworks. And it's hilarious because for me, my undergrad degree,
I got something that I got an undergrad degree in something that I thought was going to be entirely
useless. And that I haven't used as such since then.
It's not remotely technical.
It was much more theoretical.
And now, the deeper we get into AI and the automation of all work, the more grateful I am that I got a degree in advanced theory for certain types of things,
advanced theory for certain types of things, precisely because it taught me frameworks
for how to think, not how to do stuff, how to think. And that puts me a step ahead of AI,
especially when I wanted something creative. So going back to the example of JK Rowling,
there are frameworks for coming up with stories. There are frameworks for coming up with character
development. There are frameworks and scaffolds that writers use to keep track of their world building as they're filling up dozens, dozens of notebooks full of notes, handwritten, scribbled notes off to the side and in the margins.
I remember I visited the home of an author once who was in the middle of writing his first trilogy.
who was in the middle of writing his first trilogy.
And he didn't even want to release the first book
until he had finished completely building the world
and the characters developed.
And he had something like two dozen,
I think like 500 page condensed notes journals
where he had gone back through and collated his notes,
gone back through and cross-indexed his notes.
And he still had dozens of books of notes, only notes, just notes, on the various different kinds of idiosyncrasies that he
built into these characters so that when we read these stories that he had written, they don't
fall apart right upon. In my mind, then, creating a fictional character that when we question their motive, we realize,
oh, there's not really much substance to that one, is there? Good writing and good art,
it bears a little bit closer scrutiny where if we were to ask, hey, if I were to zoom in,
punch in a little bit on this one character, could they actually handle
receiving a spinoff piece of media, either a spinoff book or movie or TV series? Like Better
Call Saul, right? Interesting, quirky character from Breaking Bad, who was interesting enough
that we actually wanted to see more of him. So being able to write characters that level of depth,
So being able to write characters that level of depth, that's not nothing.
And AI systems, generated AI systems, unless directed to do so, they don't automatically think to that level of depth.
They don't go into, if they do generate second order considerations, they don't often generate third and fourth order considerations.
they don't often generate third and fourth considerations. They're great at chess in
terms of having a fully memorized canon of moves for chess. But when it comes to writing characters
who then have multiple additional dimensions, they're not really great at 4D chess yet,
we'll just say. So it'll be interesting to see how that develops in the next little while.
And clearly, as soon as that's developed into a system,
AI will kick our ass at that too.
But for now, it is nice to know that we can be somewhat in control
or feel like we're in control of directing their general development.
I mean, to paraphrase Homer Simpson,
they're not as good as we are yet.
And that seems to be the phrase that we keep saying,
well, they're not really good as that yet.
I see a lot of that in some of the news stuff with people commenting.
As you get experts in different fields, they're saying,
oh, it's terrible at this.
But I keep coming back, well, when anything's a baby,
it's always kind of crappy.
So who knows? One one day maybe it'll
start doing even more and more and i expect it really well i want to thank everyone for listening
today it's been a really interesting set we've talked about robots and military and the ganges
and drinking clean water at the mouth hole and you know different politics sovereign ai we bounced
around on nVIDIA's
hundred billion dollar bet that seemed to bounce from Oracle back into open AI and around in
circles. Lots of stuff happening. You know, the Sora 2 announcement, you know, ChatGPT,
you can now buy stuff directly. So the increasing use of AI as your front end as we go forward in
the world. And we'll wind this up.
And thanks for everyone.
And we'll see you again next week on another great AI podcast.
Talk then.
And if you have any comments, post it.
Otherwise, read the newsletter every day because there's lots of stuff happening.
And we'll talk again soon.
Bye for now.
