September 25, 2025 • 38 mins
In this week’s Market Maker Podcast, Anthony and Piers dive into three huge stories shaping markets, technology, and careers in finance:
Trump’s U-turn on Ukraine & the defense stock boom – What does his latest stance mean for NATO, Europe’s defense spending, and why has one German arms manufacturer surged 2,000% since the war began?
Nvidia’s $100bn OpenAI deal & the global AI race – From landmark partnerships with OpenAI, Alibaba, and Intel, to political maneuvering at the top table, we break down how Nvidia is cementing itself as the backbone of AI infrastructure.
HSBC’s quantum leap – The bank claims a 34% improvement in bond price predictions using quantum computing. We unpack what this really means, how it could reshape Wall Street, and why students and early-career professionals should pay attention.
Whether you’re curious about markets, fascinated by frontier tech, or exploring your future in finance, this episode connects the dots between geopolitics, AI, quantum, and the career skills that will matter most.
💡 Timestamps
(00:00) Intro & Topics
(02:04) Trump’s U-Turn & Defense Stocks
(08:55) Nvidia’s $100bn OpenAI Deal
(25:13) Nvidia’s Global Strategy & Politics
(30:45) Quantum Computing in Finance
👉 Useful Resources
AmplifyME free Markets Finance Accelerator
AmplifyME free Banking Finance Accelerator
📌 Student Training Programmes
5-day In-Person Bootcamp in London
3-Week Virtual Summer Analyst Programme
🤝 Connect
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
(00:00):
Hello, and welcome back to the Market Maker Podcast.
And we've got 3 topics we're going to discuss in this
conversation. Number one, Trump just flipped
his stance on Ukraine, saying that Kiev can retake all of its
land. Defense stocks, particularly in
Europe, have surged. Ryan Matal, the German arms
(00:20):
manufacturer, the 5th largest European arms manufacturer, in
fact is up nearly 2000% since the war between Russia and
Ukraine. So we're going to unpack why are
they such an outlier? What's happening to defence
names as NATO's pledge to boost defence spending continues?
Second story, of course, NVIDIA,they planned $100 billion
(00:42):
investment into open AI alongside its ongoing deal
making spree. We've also had news this week
about Alibaba last week, Intel signalling a strategy to cement
their dominance as the indispensable engine in this
global AI race. And then HSBC just claimed a
world first in quantum computing.
(01:03):
So look, everyone's talking AII want to talk a little quantum
for the for the back end of thisepisode.
And the reason why is HSBC, they're boosting bond
predictions by 34% accuracy, they said.
And in the world of finance, where 1% significantly moves the
dial they put out there in a demo test, 34%, it's
(01:25):
experimental. But even if part of this is
real, it could dramatically reshape finance competition on
Wall Street and Koreas in finance.
So I want to kind of loop it back into that side as well.
Quick shout out to Saab, our newpodcast researcher doing an
awesome job for us. So Saab Ertam, thank you very
(01:46):
much. Then also final one is we're on
821 Spotify ratings. I would very much love to get
that to 1000 by year end, year end target 1000 locked in.
Let's go So over to the first story then.
So Piers Trump bit of a surprise.
(02:06):
He has done well. Is it a surprise?
I don't know. He's done AU turn again.
Trump said Ukraine could retake all occupied land.
That obviously reverses earlier suggestions that Kiev should
give up territory. So first of all, what was the
market reaction to that? Yeah, well, is it a surprise?
I mean, I guess the thing about Trump is he just keeps flip
(02:28):
flopping and, you know, is proneto changing his mind seemingly
on the fly, you know, every day or two or whatever.
So, I mean, it's it's an unsurprising surprise if you see
what I mean. But it is a massive flip.
And U-turn on his whole stance around the the Ukraine
situation. Of course, he's been trying to
(02:48):
broker the ceasefire and the deal, right?
But Putin's seemingly not playing ball.
So it looks like, in my opinion at least, it's a part of this is
about him trying to, it's a, it's a kind of veiled threat to
Putin to say that, look, if you weren't prepared to come to the
table to broker a deal, well then, all right, game's back on,
(03:10):
right? But of course, Trump wants to
avoid getting kind of boots on the ground involved in this kind
of thing. And so he's kind of pointing
towards the Europeans and it's their responsibility and really
blaming the Europeans for not getting kind of doing their
share in this whole situation. And so what's the market
reaction to this? Well, the most stand out market
(03:33):
reaction was the European sort of defence companies and they've
kind of marched higher. You know, part of this is
because, you know, NATO's pledged to increase defence
spending to 3.5% of GDP. So we know that the Europeans
have been woefully off that sortof pledge for many years.
(03:54):
And you know, Trump has been kind of banging them for that
for a long time. And, and so commitments from
European governments to spend more on defence, of course, is
good news if you're in the business of supplying, you know,
military equipment and and defence items.
Right now, there's one stand outthough player who's benefited
(04:14):
more than anyone else from this.And this is a company, a German
company called Rhine Metal. And they're up.
They're actually since the Ukraine Russian war began,
they're now up 2000% their shareprice.
That yeah, I saw that and I, I saw a chart and it was a whole
cluster of these European defence names because it's quite
a few and they've all gone up dramatically.
(04:35):
But this German firm was up way over and above everyone else.
So why is that? Why are they?
Why are they up 2000 when everyone else is only up about
sort of 200 to 400%? Yeah, so this is where you've
got a, if you kind of just go below that first layer.
So the layer one is, ah, what? More defence spending, right?
(04:57):
Who are all the defence companies, right?
Let's buy all of them. That's kind of layer one, right?
If you then go one layer down, well, this is where you can find
some massive outperformance because what do Rhine metal
specialize in? They're Europe's biggest
ammunition producer. Now there's two interesting
things to that. Firstly, the EU and NATO have
(05:18):
named musicians resupply as their top short term defense
need number one. Number two, you can bang out
these munitions like just turn on the tap and it doesn't take
long to manufacture munitions. You know, if you're in the
business of building, you know, high tech kind of missiles or
(05:40):
let's say high tech airplanes, right?
I mean, these take a long time. You can't just turn the tap on
and off with that stuff. So Ryan metal are seen to be
here perfectly positioned to benefit in the, you know,
straight away short term from the, you know, defined kind of
top sort of priority to kind of re kind of load on the munition
(06:03):
side. So yeah, they're the ones that
are going to benefit for most and in the shortest term.
And so it's kind of an ammunition pureplay if you like.
And so they're top dog and that's why investors have been
buying those shares like they'regoing out of fashion and 2000%
return. I mean, that war has been going
(06:24):
on, unfortunately for like 4 years now.
So it's a 2000% return over a four year period.
And seems like that upside's still on the go.
Yeah. I, I guess one of the things is
the, the long term tailwinds here that they're going to
benefit from because when you would, you said that about 3
(06:44):
1/2% of GDP that NATO's kind of pledging, assuming though that
that pledge is several years. Well, I don't actually know what
the defined timeline is or is ituntil we decide to change again
in the future. So they'll continue doing that.
Well, there are risks to this. It's all very well and good that
your policy is right. We're going to aim to move
(07:06):
towards spending 3.5% of GDP on defence.
The problem is all of these governments, maybe with the
exception of Germany, To be fair, all of these governments
have got a massive debt problem.You know, they're trying to
figure out ways of spending less, not more.
So you know, that's obviously the key barrier here, the key
risk to all of this policy. It's kind of execution risk and
(07:30):
and whether. It's possible or not.
Yeah, it's interesting as well. Just on the final point on this
before we move on, I think the the Trump angle politically
where it's, you know, he went in, I think he's he was going
into some of these big confrontations like with China
or with Putin and he was trying to strong arm a deal.
But as you said, the Putin interactions have kind of
(07:51):
happened, have kind of been and gone and seemingly the war is
continuing at a pace that hasn'treally changed.
And so I just wonder whether thecalculations have been that,
yeah, this, now that he's got more into it, it's just avoiding
it becoming his war. Let's just be a strong
figurative person within the talks.
However, this is really Europe'sproblem, so that it doesn't
(08:14):
really come to get him back later down the line.
Right. And look, Putin, I mean,
apparently economic, the economic situation in Russia is
deteriorating and it seems like the speed of deterioration has
picked up a little bit. So that's obviously that's bad
news for Putin and his war, right?
And so he needs an off ramp where he can somehow get out of
(08:36):
this and still claim a victory. And maybe Putin's just now
threatening that by saying, look, maybe this war's still on.
Maybe the Ukraine can regain territory.
You know, that is a risk, Putin.So come to the table now and do
a deal. Interesting.
All right, well, look, let's move on.
Talking video because this was aa big deal and I I came a little
(09:00):
bit of as a surprise as well. So the headline was that NVIDIA
and open AI announced a landmarkpartnership with plans for
NVIDIA to invest up to 100 billion.
So I guess a couple of questionshere.
What exactly is the partnership?How are they, you know, putting
the funds in? How does this work in practice?
So this is a letter of intent, OK, has been signed.
(09:23):
So this is NVIDIA saying, well, an open AI of course, agreeing a
deal, you know, at a top level in principle, the deal is $100
billion of investment, OK. So this is NVIDIA investing 100
billion into open AI. But look, this is going to be
staged over time. This is one key part of this
(09:43):
deal. The overall idea is this is this
is to help Invidia, sorry, open AI, you know, build out 10
gigawatts of AI data centre capacity.
So obviously on one side, open AI growth ambitions, I mean, at
the moment they've got about 700million weekly active users on
(10:06):
ChatGPT. Obviously each person that's
using this that requires compute, right?
You need a data centre to be able to handle their questions
and then do the thinking and then fire back the answer,
right? So the more people you have on
the platform, the more compute power you need.
Also, as they're upgrading, they're going through the
(10:26):
generations. What are we on ChatGPT 5 now?
Each new version is better, meaning it needs more compute
power anyway. So you've got this exponential
compute demand as these Chat GPTS get more advanced and the
number of new users increases atthe same time, right?
So it's about how can we, how the hell can we meet this
(10:49):
demand? And so This is why you're
getting all of these kind of data centre build out deals
being done. Because don't forget, Open AI
has already done a $500 billion Stargate deal with SoftBank and
Oracle, OK, with the US government getting tied in.
So this is just another deal. Now the way it's structured,
(11:11):
their plan is to build out 10 gigawatts, right?
But they're going to kind of do it one GW at a time.
So initially it will be 10 billion of the 110 billion will
be upfront and land in open A isbank account on the signing of
the deal. So that's the first point.
It hasn't been signed yet. A letter of intent is right.
(11:33):
We've agreed in principle and apparently it was just Jensen
and the open AICEO. His name has just completely
escaped me. Chris Altman, Sam Altman.
Sorry. So they apparently they broke up
this deal together like literally CEO to CEO, no
bankers, no advisors. Like they've been talking for a
(11:54):
while. You.
You. Imagine these egos, the size of
these egos, and. They both went, they actually
both. Where they finalized the lessor
of intent was when they were here in the UK, because Trump
did a a state visit and he brought a load of his big guns
with him, including Jensen and Sam Altman.
And apparently on the out, you know, on the outskirts of all of
this stuff that's going on, theyfinalized this deal just the two
(12:16):
of them apparently. So look, stuff needs to happen.
Now. The lawyers get involved,
obviously M&A advisers get involved before actually get
signed. But if it gets signed, 10
billion in the bank account for open AI.
So that's number one, right? And then they're going to do
this over time, A1 GW sort of build out each, you know, a time
(12:38):
times 10 right now, apparently. Well, so here, what does NVIDIA
get? Well, they get an equity stake
in open AI. Now, how much for the 100
billion? Well, we don't know yet because
it's going to be in stages and open AI is going to be growing
and their valuations going to shift over time.
I guess what we do know is that right now they're worth about
(12:58):
500 billion based on their last funding round.
So 10 billion upfront would get NVIDIA a 2% equity stake.
OK. The beauty of this for NVIDIA,
though, this money's coming back.
It's like, here's here's 10 billion.
Oh, but by the way, a condition of the deal is you've got to
(13:20):
spend that ten billion on buyingmy chips.
So this this, this money is going to do in a bank turn and
temporarily be an open AIS account before it now comes back
into Nvidia's. We'll talk about this in a bit
more. It's kind of a circular deal.
Is this like double counting from a kind of revenue point of
(13:41):
view? So some people are being pretty
critical about this is a bit dodgy, but it's not dodgy from a
legal sense. But it's like, is it kind of
over inflating Nvidia's revenue growth if essentially they're
growing it with their own cash? So there's that kind of angle to
(14:01):
it. But look, what do they get out
of this? And this isn't just chips, by
the way. So this is open AI will be
spending this money on Nvidia's next Gen.
Vera Rubin platform. OK?
So this is isn't just their chips.
This is a wider platform for these data centres.
One risk is this Vera Rubin platform ain't finished yet and
(14:28):
it hasn't been shipped. They're still in the design
phase. It's due for delivery autumn
2026, but they're Blackwell shipthat was late when they shipped
that a few months after they were kind of saying they were
going to. So there are risks here in terms
(14:48):
of can they deliver and ship Vera Rubin in time.
But look, if that if if they do well, then fine, right?
This data center build out cracks on and each GW then will
release another 10 billion of transfer from NVIDIA to open AI.
(15:09):
So that's the kind of deal at the top level.
Oh, and by the way, just finallyjust on this first phase of the
conversation, it's not just 10 billions coming back, right,
Because NVIDIA have guided and this was earlier in the year,
right? For every one GW of data center
build out, it costs about 50 to $60 billion, OK.
(15:35):
That's for the total spend, about 35.
So just over half of that is typically spent on NVIDIA
hardware, OK. So if you add it all up, if
they're doing 10 gigawatts, thenthey're going to end up spending
about 400 billion on NVIDIA stuff.
(15:56):
So this is NVIDIA saying here's 100 billion, we will finance
your data center build out ambitions.
We'll take an equity stake for that.
Thanks very much. Oh and by the way, you've got to
buy our stuff and we're basically going to Forex our
investment in terms of direct revenue back, you know over this
(16:17):
10 GW build out cycle. Sounds like a, a cheeky little
PE deal going on here to some extent, but like what?
So I, I think NVIDIA, what they generated just over 70 billion
in free cash flow over the over the last four quarters.
I mean, which is just insane. So is this this, this way of
(16:39):
operating? Is this like a new paradigm now
where it's like actually this isquite unique in a scenario
where, yeah, they do have financial firepower.
This isn't about I'm going to buy you.
It's an invest play I. Mean, if you think about this
from there's a few angles here, right?
It's like, can they afford it? Is this, does this put at risk
(17:01):
Nvidia's business? Are they taking on too much risk
here? That's kind of 1 angle.
Another one is, well, what does regulator think about this?
Are there any antitrust risks and and so on, But they
definitely can afford it. I mean, as you're saying, they
are are just spewing out free cash flow and actually 72
(17:21):
billion over the last four quarters.
But if you look at this fiscal year we're in now of course
they're growing right, every quarter they're growing faster.
So in the fiscal year we're in now, they're expected to churn
out 100 billion of free cash flow.
So that kind of covers that kindof covers this like in one fell
swoop just in 12 months worth ofbusiness, right?
(17:42):
So they can definitely afford it.
Don't forget in parallel, NVIDIAissued a 50 billion share
buyback program last year. This year they've upped it to 60
billion. They got, they got so much cash,
they just don't know what to do with it.
OK. So they can definitely afford
this. On the antitrust thing, it's
(18:04):
quite clever because obviously the antitrust police kind of get
involved when the big guns are doing full acquisitions where
they are buying entire companies.
And then the regulator's like, whoa, whoa, whoa, hang on.
You're getting far too much control here of a particular
situation. And this isn't good for
competition or for price points to consumers and all the rest of
(18:27):
it. But here, now this is NVIDIA
just chipping away buying a initially a 2% stake, right?
That's not that's not a big enough stake for NVIDIA to be
able to control anything that open AI does.
And open AI have got plenty of other investors, Microsoft,
Oracle and so on, right. So I think this is an
interesting play where it kind of falls under the radar, I
(18:49):
would say, at least for now, thesort of antitrust police.
And really it's more of a kind of, I guess it's a compute land
grab for NVIDIA. You know, they're saying, well,
open AI, you know, they're, they're, they're a different
vertical to us. You know, NVIDIA, they are the
they, they're, they're the kind of infrastructure build out for
(19:12):
AI, right? Open AI.
Well, they're the product that'sthen the app that sits on the
top of this stack. They're the consumer facing
entity, right? At least for now, they're the
biggest. So NVIDIA are saying, look,
let's lock them in to our vertical infrastructure stack.
So that we're kind of making sure we've got we, we've got a
(19:35):
continued majority share in thatcompute land grab that's kind of
going on all over the world and we'll come on to it.
But NVIDIA, they're not just investing in open AI, right?
But there's deals going on all over the place here.
And it's all about trying to maintain their kind of dominant
market share on that compute site.
(19:56):
Yeah. I just find, you know, working
in the strategic department of NVIDIA at the moment must be
such a such a like a brain box of like intelligence going on
and how to maximize this becausethey announced that partnership
with Alibaba. To bring its physical AI
development tools, so thinking things like autonomous vehicles,
(20:17):
robotics, smart spaces and into Alibaba Cloud.
And, and again, it's, I guess like you were saying, it's this,
you know, you think about China,this is a major global
distribution channel for your GP, us, your AI software stack.
So it's all kind of, it's all sort of feeding into this same
theme, isn't it? Yeah, that so that, that AI is
(20:40):
global, right. I mean, all right, we spend most
of our time talking and thinkingabout the US and fine, they are
leading this race, but obviouslyit's a global thing.
And so if NVIDIA can get a big slice of some of the action in
in the China market via the Alibaba tie up, well then
obviously that that kind of makes a lot of sense from that
(21:02):
sort of global perspective. Yeah, Yeah.
Just going to say like a bit of detail there, like locking in
the timing is interesting because on the Alibaba side,
there's this Quinn 3 Max model roll out there in the midst of
the moment and it claims to outperform rivals.
And this is like we'll see a keyuse case of AI.
It's encoding in in autonomous agents tasks.
(21:26):
And so if they can get in now and dominate, what is China's
top AI players? A secure critical role
empowering their cloud systems. And it's a stroke of genius.
Yeah, unlike they've also taken up a 500 million or they're in
talks of doing a 500 million, taking a 500 million stake in
WAVY, which is AUK based kind ofsort of car basically a break.
(21:53):
Like it's like rivaling Tesla's kind of autonomous vehicle sort
of brain, if you like. And they're taking a stake here
in the UK. So again, that's like just
trying to diversify outside of obviously Nvidia's big customers
are the Super scalers. They're the Microsofts and the
(22:14):
Amazons and the metas and and the Googles, right.
So it's just trying to diversifythemselves away from just those
kind of big tech super scalers and and like be a bigger wider
part of this broader AI future. That's obviously sprouting out
in many, many directions. One of the big things here, the
risk factors you said about on the regulatory front.
(22:37):
So also what I find is really fascinating about Nvidia's
approach is they invested 5 billion into Intel.
So they struck A partnership to Co develop some custom GPUs that
could easily integrate into Nvidia's GPUs.
Now, what was, I think it's really clever about this, is
that the deal? So the context, the US
(22:58):
government said it's going to take a 10% stake in Intel.
This was like a month ago, that deal converted nearly 9 billion
in promise grants and equity andmade the Commerce Department the
Intel's largest shareholder. So this latest agreement,
Nvidia's done. I think it just shows how a lot
of these executive leadership teams, they're viewing efforts
(23:21):
to help Intel as a little bit ofa let me help you, Trump.
You've got objectives and ambitions.
Let me align with you. And you know, one of Jensen
Wang's top priorities, of course, is he wants to export
Blackwell chips to China. And so how does he get
favourable terms? Well, look, if I scratch your
back, you scratch mine. So again, it makes a lot of.
(23:42):
Sense 100%. This is a big, you know, Jensen
is playing at the top political table.
You know, he was part of Trump'sparty on this state visit here
over to the UK. He is 100% playing a political
game here. And that China export sort of
ban. I mean, basically they're doing
(24:03):
it that they're negotiating. But basically the what Jensen's
hoping for is Trump to agree that they can start shipping
Blackwell chips, but not not full, not Full Monty Blackwell.
Apparently they they need to be 30% less capable from the Full
Monty Blackwell, I think this isthe number that's being
negotiated. And Trump's going to say, OK, if
(24:23):
you if you kind of, you know, dumb these down a touch, fine,
you can send them. And of course, don't forget also
Trump's done a deal with Jensen where they're going to get, I
can't remember the figure. Was it 30% of revenue from chips
getting sold to China goes to the US government.
So this is getting crazy political.
(24:44):
And look, we could have a whole conversation on a tangent here
about the US government's role in essentially what are private
and public companies and whetheror not that is a good direction
for the US government to be taking.
That's a whole different conversation.
But right now, Jensen's got to play the politics game.
And you know what? Looks like he's playing it
(25:07):
pretty damn well to be fair. He's at the top table.
So the, well, maybe just to pullall of this together, then
there's kind of five components of it this that this particular
story to summarize because it isa really fascinating 1.
So #1 is, is the partnerships. Many of these different firms,
we said, so NVIDIA investing andpartners like Open AI Alibaba to
(25:31):
secure chip demand, reinforce its role with the backbone of
this build out this infrastructure for AI #2 compute
power. That's obviously the priority
here. You've, you said this in nearly
every episode. You just can't keep up pace at
this point. It's the new battleground.
It's kind of massive energy financial costs shaping the, the
competition landscape, the circular loop.
(25:52):
I guess there's pros and cons. Here's there's the the funding
customers who reinvest in your demand makes a lot of sense.
But also I can see the point. I'd love to get people's opinion
in the comments section about capital recycling rather than
true underlying growth and value.
And then #4 the political leverswe've just been talking about,
(26:13):
you know, the Intel state, UK investments alignments, USUK
government. Why do they do this?
Well, they want to win favour. They want to reduce kind of
regulatory friction headwinds. And then finally, interestingly
for the investment bankers listening is NVIDIA avoids large
scale acquisitions. So one of the stats I saw here
(26:34):
and a shout out again to to Saabour our researcher, he said the
2020 acquisition of Mellanox is so far Nvidia's only M and a
deal valued over $1 billion. Yeah, I mean, we're talking here
with this latest pop this week. Nvidia's a four and a half
trillion dollar company and we've been used to seeing big
(26:56):
tech like Facebook back in the day gobbling up the Instagrams,
the Whatsapps. These guys have only spent
evaded on one deal over 1 billion and here we are it's.
Amazing. A four and a half trillion
dollar company that spews out 100 billion of free cash flow
every 12 months, and yet they'veonly ever spent a billion or
more on one acquisition and thatwas 5 1/2 years ago, which was
(27:17):
before this AI kind of accelerated boom even began.
Of course, this is pre ChatGPT. I'm going to put two other for
thoughts on this to kind of wrap.
Number one is this and I was reading the Lex column in the FT
which is I'd recommend. Let me guess, they're going to
(27:39):
be critical. Let me guess.
Oh, they're going. To be critical, well, that's it.
So you're going to understand their flavour, you got to
understand that they they need to.
Look at sun shining. I know it's cold, but it's it's
not raining. Come on guys.
They're basically saying, look, this circular thing as in open
AI here's 100 billion. Oh, you got to give it give it
us back though, and that'll be Walmart that up as revenue.
They're like, come on, they're basically calling this, and I'll
(28:02):
quote this is partly theatre to keep the fly, the flywheels
spinning. So to keep this whole kind of
story about the AI revolution going, obviously NVIDIA needs to
carry on growing. So how are they going to achieve
that? They're going to fund other
people to then self funds their revenue growth.
(28:26):
So that's one, I'd say that's one angle.
Here, welcome to something that's been happening through
the history of time. It just happens to be in the
context of AII mean. I love it when market
traditionalists, it's kind of like market traditionalists sit
here, the crypto crew and NFT groups sit here.
This AI one's kind of somewhere in the middle.
And it's like as soon as it's threatening traditional
(28:48):
landscape of finance, they're like, Oh no, it's all bad.
But yeah, go on, you have a second point.
Yeah, boy, you're right. This isn't you because Lex were
referencing A Nortel Lucent dealin the early 2000s.
I won't go into it now, but they're basically saying it's
the same again, this is good, this isn't good.
The other part is, is really outof there.
Well, it's kind of out of Nvidia's control to a degree for
(29:12):
all these. You can, you can build all the
data centres you want, right? But is there enough power?
Is there enough electricity? Number one, is there enough
energy and #2 is there enough infrastructure to get the energy
to the data centre? So there are there are big
infant like power and energy infrastructure challenges here.
(29:36):
So again and partially this is about the government and
partially this is about the regulatory environment around
getting permits to a increase energy kind of production, but
also permits to kind of transition this power through
through the grid and through power lines.
So all of this stuff is a definite risk to this huge
(29:59):
ambition of like building fine 10 gigawatts of data centres,
yeah, go for it. But is it going to even be
possible from a power point of view?
So that's another risk to factoring.
And just to round up your man Sam Altman and his, his
relationship as chair just so happens to be chairman of OCLO.
(30:20):
That's that, that that nuclear technology, right fusion power
company. So he, he is got this all worked
out for such such as the Albans and the Musks of the world.
They're certainly have got all bases covered here to to meet
your second point. But look, just quickly to wrap
in about 5 minutes, just a storythat I thought I didn't want to
(30:42):
go into it too much, but just asa headline, HSBC, they've
achieved a breakthrough in deploying quantum computing in
financial markets. And I thought it was interesting
because it felt like quantum wascoming up and then the AI just
kind of superseded and it's goneback a little bit.
So this was an interesting development because I remember
(31:03):
watching a documentary about HSBC&IBMIBM being one of the
forefront kind of, well, at least U.S. companies in the the
quantum space. And this was years ago.
They were talking about it, but basically IB Ms. Heron quantum
processor and they're using it to improve bond price
predictions. So one of the things here, and
(31:25):
the reason why it stood out thisweek is the bank applied quantum
processing to an anonymized set of European bond trading data
and they found it can significantly enhance the
efficiency of the market with a 34% improvement in predicting
how likely a bond will trade at a given price.
Right now, I've always struggledto understand conceptually what
(31:49):
is quantum computing. So this is the best effort that
I can do for anyone who's still was in that camp I was in not so
long ago. So a classical computer works in
bits. Each one is either A0 or A1 and
it tests solutions at A at a time in a sequence.
A quantum computer uses qubits which thanks to superposition
(32:13):
can be 0 to one at the same time.
Now my first question was what the hell is a superposition?
So the easiest analogy I could come up with is think about if
you flick a coin. Now a normal computer is like
flipping a coin and you have to wait for the coin to land right
before you can then determine, oh, it's a heads or a tails.
(32:35):
So you have to flick it, it spins and it lands and then you
pick it up and flick it and it'sone sequence after another.
But each event has to occur in in a in a structure in a quantum
computer. Think about it again, me
flicking the coin, but while it's in the air, it's at at any
point in time, it's rotating through both heads and tails at
(32:56):
the same time. So this in between state, if you
like, is the superposition right?
You're still with me, Leonardo DiCaprio.
I. Mean I like.
It I was just thinking best. Description I've heard.
Carry on. I'm thinking inception here.
Quantum computers can can basically use that state the
(33:17):
superposition, to explore many possibilities at once instead of
waiting for the coin to land. So it means that you can solve
problems exponentially faster. Like when I mean fast, I mean,
you know, dramatic. Volt.
You same Volt fast. So tying this back then to
markets what you know when I wasreading this, I was like, OK, so
(33:39):
so So what does this mean for financial markets?
You mentioned the bond pricing. So first of all, who's looking
at this? HSBC has been a bit of a front
runner. They've been doing this for
several years. JP Morgan, Goldman's, they are
actually all investors in quantum themselves in house.
So the ways to think about this are risk management.
So think about if you were testing scenarios, but on an
(34:02):
instant basis, imagine if a piece of news broke and you
could run it into your system and automatically then get
outputs of it might mean this orthat correlated effect or
second, third order impacts and they happen immediately as
trading signals, for example, inan automated way, price
(34:22):
prediction accuracy obviously could that where this is what
this data were was saying, but Iknow you had a third one as well
that some of these banks look at.
What with regards to what? What as in price prediction?
Or as in fraud, wasn't it like? Fraud.
Oh, I see. Yeah, well, fraud detection and
security, yeah. I mean, analysing patterns and
(34:43):
kind of building new encryption and just just kind of massively
upping the game with regards to fraud detection, which of course
can save them a lot of money. So yeah, that's another angle to
it. Yeah.
And and so when I was thinking of this, I was thinking, OK, so
thinking about a lot of our audience, Well, what, what does
(35:03):
this mean? Like, why should a student care
about this? And it's this kind of
intersection with tech and finance.
And I think most students have got got a finger on the pulse.
You know, the world's moving in a way where concentrated
researchers, analysts, these areall quite key.
So typically programming skills,advanced mathematics,
(35:23):
statistics, leaning towards AI, machine learning, these types of
principles. I thought myself when I read
this, I was like, Oh my God, like what about people like me
who don't have any of these skills because I'm not a
mathematician by trade. And So what I thought was
interesting and just to finish and share, I was just, you know,
talking to AI going, you know, this is the scenario like, yeah,
(35:45):
what type of roles can I do thenwithin a financial institution
that yes, we've talked about open AI and we're talking about
all these, it's all very sophisticated, like frontier
technology. But what about where does that
leave then the humans element? And I thought this was
interesting from from 5 different roles that you could
(36:07):
think of. The first one is products and
strategy. So there's one thing building
this stuff out, but like you said, the Huang and Altman
mindset of OK, how do I engineersituations, the strategic
elements of deal making and strategy creation, like that's
(36:28):
very much still led by experience and and organizing
teams getting input, stuff like that.
So, you know, deciding where quantum should be applied, where
it should be applied is just as important as being sure it can
actually work and do it's thing.So strategic priorities, client
facing sales roles. So, you know, I won't mention
(36:51):
who, but I'm doing some actual coaching on, on public speaking
and how to communicate with a fairly sophisticated group of,
of, of high end quants because they're struggling to really
articulate to PMS in a way that the PM can understand very
quickly. So again, being able to
understand this stuff, maybe notbe a technical person who's, you
(37:12):
know, an engineer, for example, but enough where you can
articulate it to other people and also to clients, obviously
external. Then there's regulatory risk
oversight. That stuff's only going to get
more onerous as technology becomes more influential.
Project management, you know, coordinating all these people
and then communication and education, you know, what about
(37:35):
these bits about internal comms,your staff training, you know,
these translators. I, I think it's going to become
a really important skill as well.
So look, overall, I, I don't want people to be spooked too
much. So they've got to go away and do
a PhD in maths that for sure is going to lock you in on that
Open AI or what was it that $100million salary at Facebook?
(37:55):
Sure, but that's real a place for, for, for me in this world.
I'm, I'm thankful to say, but sowe'll wrap it up there.
Piers, pleasure as always. Thank you very much.
And any comments that people have, just let us know.
We'd love to know, and we'll seeyou next week.
Thanks a lot, catch you later.
Hello, and welcome back to the Market Maker Podcast.
And we've got 3 topics we're going to discuss in this
conversation. Number one, Trump just flipped
his stance on Ukraine, saying that Kiev can retake all of its
land. Defense stocks, particularly in
Europe, have surged. Ryan Matal, the German arms
(00:20):
manufacturer, the 5th largest European arms manufacturer, in
fact is up nearly 2000% since the war between Russia and
Ukraine. So we're going to unpack why are
they such an outlier? What's happening to defence
names as NATO's pledge to boost defence spending continues?
Second story, of course, NVIDIA,they planned $100 billion
(00:42):
investment into open AI alongside its ongoing deal
making spree. We've also had news this week
about Alibaba last week, Intel signalling a strategy to cement
their dominance as the indispensable engine in this
global AI race. And then HSBC just claimed a
world first in quantum computing.
(01:03):
So look, everyone's talking AII want to talk a little quantum
for the for the back end of thisepisode.
And the reason why is HSBC, they're boosting bond
predictions by 34% accuracy, they said.
And in the world of finance, where 1% significantly moves the
dial they put out there in a demo test, 34%, it's
(01:25):
experimental. But even if part of this is
real, it could dramatically reshape finance competition on
Wall Street and Koreas in finance.
So I want to kind of loop it back into that side as well.
Quick shout out to Saab, our newpodcast researcher doing an
awesome job for us. So Saab Ertam, thank you very
(01:46):
much. Then also final one is we're on
821 Spotify ratings. I would very much love to get
that to 1000 by year end, year end target 1000 locked in.
Let's go So over to the first story then.
So Piers Trump bit of a surprise.
(02:06):
He has done well. Is it a surprise?
I don't know. He's done AU turn again.
Trump said Ukraine could retake all occupied land.
That obviously reverses earlier suggestions that Kiev should
give up territory. So first of all, what was the
market reaction to that? Yeah, well, is it a surprise?
I mean, I guess the thing about Trump is he just keeps flip
(02:28):
flopping and, you know, is proneto changing his mind seemingly
on the fly, you know, every day or two or whatever.
So, I mean, it's it's an unsurprising surprise if you see
what I mean. But it is a massive flip.
And U-turn on his whole stance around the the Ukraine
situation. Of course, he's been trying to
(02:48):
broker the ceasefire and the deal, right?
But Putin's seemingly not playing ball.
So it looks like, in my opinion at least, it's a part of this is
about him trying to, it's a, it's a kind of veiled threat to
Putin to say that, look, if you weren't prepared to come to the
table to broker a deal, well then, all right, game's back on,
(03:10):
right? But of course, Trump wants to
avoid getting kind of boots on the ground involved in this kind
of thing. And so he's kind of pointing
towards the Europeans and it's their responsibility and really
blaming the Europeans for not getting kind of doing their
share in this whole situation. And so what's the market
reaction to this? Well, the most stand out market
(03:33):
reaction was the European sort of defence companies and they've
kind of marched higher. You know, part of this is
because, you know, NATO's pledged to increase defence
spending to 3.5% of GDP. So we know that the Europeans
have been woefully off that sortof pledge for many years.
(03:54):
And you know, Trump has been kind of banging them for that
for a long time. And, and so commitments from
European governments to spend more on defence, of course, is
good news if you're in the business of supplying, you know,
military equipment and and defence items.
Right now, there's one stand outthough player who's benefited
(04:14):
more than anyone else from this.And this is a company, a German
company called Rhine Metal. And they're up.
They're actually since the Ukraine Russian war began,
they're now up 2000% their shareprice.
That yeah, I saw that and I, I saw a chart and it was a whole
cluster of these European defence names because it's quite
a few and they've all gone up dramatically.
(04:35):
But this German firm was up way over and above everyone else.
So why is that? Why are they?
Why are they up 2000 when everyone else is only up about
sort of 200 to 400%? Yeah, so this is where you've
got a, if you kind of just go below that first layer.
So the layer one is, ah, what? More defence spending, right?
(04:57):
Who are all the defence companies, right?
Let's buy all of them. That's kind of layer one, right?
If you then go one layer down, well, this is where you can find
some massive outperformance because what do Rhine metal
specialize in? They're Europe's biggest
ammunition producer. Now there's two interesting
things to that. Firstly, the EU and NATO have
(05:18):
named musicians resupply as their top short term defense
need number one. Number two, you can bang out
these munitions like just turn on the tap and it doesn't take
long to manufacture munitions. You know, if you're in the
business of building, you know, high tech kind of missiles or
(05:40):
let's say high tech airplanes, right?
I mean, these take a long time. You can't just turn the tap on
and off with that stuff. So Ryan metal are seen to be
here perfectly positioned to benefit in the, you know,
straight away short term from the, you know, defined kind of
top sort of priority to kind of re kind of load on the munition
(06:03):
side. So yeah, they're the ones that
are going to benefit for most and in the shortest term.
And so it's kind of an ammunition pureplay if you like.
And so they're top dog and that's why investors have been
buying those shares like they'regoing out of fashion and 2000%
return. I mean, that war has been going
(06:24):
on, unfortunately for like 4 years now.
So it's a 2000% return over a four year period.
And seems like that upside's still on the go.
Yeah. I, I guess one of the things is
the, the long term tailwinds here that they're going to
benefit from because when you would, you said that about 3
(06:44):
1/2% of GDP that NATO's kind of pledging, assuming though that
that pledge is several years. Well, I don't actually know what
the defined timeline is or is ituntil we decide to change again
in the future. So they'll continue doing that.
Well, there are risks to this. It's all very well and good that
your policy is right. We're going to aim to move
(07:06):
towards spending 3.5% of GDP on defence.
The problem is all of these governments, maybe with the
exception of Germany, To be fair, all of these governments
have got a massive debt problem.You know, they're trying to
figure out ways of spending less, not more.
So you know, that's obviously the key barrier here, the key
risk to all of this policy. It's kind of execution risk and
(07:30):
and whether. It's possible or not.
Yeah, it's interesting as well. Just on the final point on this
before we move on, I think the the Trump angle politically
where it's, you know, he went in, I think he's he was going
into some of these big confrontations like with China
or with Putin and he was trying to strong arm a deal.
But as you said, the Putin interactions have kind of
(07:51):
happened, have kind of been and gone and seemingly the war is
continuing at a pace that hasn'treally changed.
And so I just wonder whether thecalculations have been that,
yeah, this, now that he's got more into it, it's just avoiding
it becoming his war. Let's just be a strong
figurative person within the talks.
However, this is really Europe'sproblem, so that it doesn't
(08:14):
really come to get him back later down the line.
Right. And look, Putin, I mean,
apparently economic, the economic situation in Russia is
deteriorating and it seems like the speed of deterioration has
picked up a little bit. So that's obviously that's bad
news for Putin and his war, right?
And so he needs an off ramp where he can somehow get out of
(08:36):
this and still claim a victory. And maybe Putin's just now
threatening that by saying, look, maybe this war's still on.
Maybe the Ukraine can regain territory.
You know, that is a risk, Putin.So come to the table now and do
a deal. Interesting.
All right, well, look, let's move on.
Talking video because this was aa big deal and I I came a little
(09:00):
bit of as a surprise as well. So the headline was that NVIDIA
and open AI announced a landmarkpartnership with plans for
NVIDIA to invest up to 100 billion.
So I guess a couple of questionshere.
What exactly is the partnership?How are they, you know, putting
the funds in? How does this work in practice?
So this is a letter of intent, OK, has been signed.
(09:23):
So this is NVIDIA saying, well, an open AI of course, agreeing a
deal, you know, at a top level in principle, the deal is $100
billion of investment, OK. So this is NVIDIA investing 100
billion into open AI. But look, this is going to be
staged over time. This is one key part of this
(09:43):
deal. The overall idea is this is this
is to help Invidia, sorry, open AI, you know, build out 10
gigawatts of AI data centre capacity.
So obviously on one side, open AI growth ambitions, I mean, at
the moment they've got about 700million weekly active users on
(10:06):
ChatGPT. Obviously each person that's
using this that requires compute, right?
You need a data centre to be able to handle their questions
and then do the thinking and then fire back the answer,
right? So the more people you have on
the platform, the more compute power you need.
Also, as they're upgrading, they're going through the
(10:26):
generations. What are we on ChatGPT 5 now?
Each new version is better, meaning it needs more compute
power anyway. So you've got this exponential
compute demand as these Chat GPTS get more advanced and the
number of new users increases atthe same time, right?
So it's about how can we, how the hell can we meet this
(10:49):
demand? And so This is why you're
getting all of these kind of data centre build out deals
being done. Because don't forget, Open AI
has already done a $500 billion Stargate deal with SoftBank and
Oracle, OK, with the US government getting tied in.
So this is just another deal. Now the way it's structured,
(11:11):
their plan is to build out 10 gigawatts, right?
But they're going to kind of do it one GW at a time.
So initially it will be 10 billion of the 110 billion will
be upfront and land in open A isbank account on the signing of
the deal. So that's the first point.
It hasn't been signed yet. A letter of intent is right.
(11:33):
We've agreed in principle and apparently it was just Jensen
and the open AICEO. His name has just completely
escaped me. Chris Altman, Sam Altman.
Sorry. So they apparently they broke up
this deal together like literally CEO to CEO, no
bankers, no advisors. Like they've been talking for a
(11:54):
while. You.
You. Imagine these egos, the size of
these egos, and. They both went, they actually
both. Where they finalized the lessor
of intent was when they were here in the UK, because Trump
did a a state visit and he brought a load of his big guns
with him, including Jensen and Sam Altman.
And apparently on the out, you know, on the outskirts of all of
this stuff that's going on, theyfinalized this deal just the two
(12:16):
of them apparently. So look, stuff needs to happen.
Now. The lawyers get involved,
obviously M&A advisers get involved before actually get
signed. But if it gets signed, 10
billion in the bank account for open AI.
So that's number one, right? And then they're going to do
this over time, A1 GW sort of build out each, you know, a time
(12:38):
times 10 right now, apparently. Well, so here, what does NVIDIA
get? Well, they get an equity stake
in open AI. Now, how much for the 100
billion? Well, we don't know yet because
it's going to be in stages and open AI is going to be growing
and their valuations going to shift over time.
I guess what we do know is that right now they're worth about
(12:58):
500 billion based on their last funding round.
So 10 billion upfront would get NVIDIA a 2% equity stake.
OK. The beauty of this for NVIDIA,
though, this money's coming back.
It's like, here's here's 10 billion.
Oh, but by the way, a condition of the deal is you've got to
(13:20):
spend that ten billion on buyingmy chips.
So this this, this money is going to do in a bank turn and
temporarily be an open AIS account before it now comes back
into Nvidia's. We'll talk about this in a bit
more. It's kind of a circular deal.
Is this like double counting from a kind of revenue point of
(13:41):
view? So some people are being pretty
critical about this is a bit dodgy, but it's not dodgy from a
legal sense. But it's like, is it kind of
over inflating Nvidia's revenue growth if essentially they're
growing it with their own cash? So there's that kind of angle to
(14:01):
it. But look, what do they get out
of this? And this isn't just chips, by
the way. So this is open AI will be
spending this money on Nvidia's next Gen.
Vera Rubin platform. OK?
So this is isn't just their chips.
This is a wider platform for these data centres.
One risk is this Vera Rubin platform ain't finished yet and
(14:28):
it hasn't been shipped. They're still in the design
phase. It's due for delivery autumn
2026, but they're Blackwell shipthat was late when they shipped
that a few months after they were kind of saying they were
going to. So there are risks here in terms
(14:48):
of can they deliver and ship Vera Rubin in time.
But look, if that if if they do well, then fine, right?
This data center build out cracks on and each GW then will
release another 10 billion of transfer from NVIDIA to open AI.
(15:09):
So that's the kind of deal at the top level.
Oh, and by the way, just finallyjust on this first phase of the
conversation, it's not just 10 billions coming back, right,
Because NVIDIA have guided and this was earlier in the year,
right? For every one GW of data center
build out, it costs about 50 to $60 billion, OK.
(15:35):
That's for the total spend, about 35.
So just over half of that is typically spent on NVIDIA
hardware, OK. So if you add it all up, if
they're doing 10 gigawatts, thenthey're going to end up spending
about 400 billion on NVIDIA stuff.
(15:56):
So this is NVIDIA saying here's 100 billion, we will finance
your data center build out ambitions.
We'll take an equity stake for that.
Thanks very much. Oh and by the way, you've got to
buy our stuff and we're basically going to Forex our
investment in terms of direct revenue back, you know over this
(16:17):
10 GW build out cycle. Sounds like a, a cheeky little
PE deal going on here to some extent, but like what?
So I, I think NVIDIA, what they generated just over 70 billion
in free cash flow over the over the last four quarters.
I mean, which is just insane. So is this this, this way of
(16:39):
operating? Is this like a new paradigm now
where it's like actually this isquite unique in a scenario
where, yeah, they do have financial firepower.
This isn't about I'm going to buy you.
It's an invest play I. Mean, if you think about this
from there's a few angles here, right?
It's like, can they afford it? Is this, does this put at risk
(17:01):
Nvidia's business? Are they taking on too much risk
here? That's kind of 1 angle.
Another one is, well, what does regulator think about this?
Are there any antitrust risks and and so on, But they
definitely can afford it. I mean, as you're saying, they
are are just spewing out free cash flow and actually 72
(17:21):
billion over the last four quarters.
But if you look at this fiscal year we're in now of course
they're growing right, every quarter they're growing faster.
So in the fiscal year we're in now, they're expected to churn
out 100 billion of free cash flow.
So that kind of covers that kindof covers this like in one fell
swoop just in 12 months worth ofbusiness, right?
(17:42):
So they can definitely afford it.
Don't forget in parallel, NVIDIAissued a 50 billion share
buyback program last year. This year they've upped it to 60
billion. They got, they got so much cash,
they just don't know what to do with it.
OK. So they can definitely afford
this. On the antitrust thing, it's
(18:04):
quite clever because obviously the antitrust police kind of get
involved when the big guns are doing full acquisitions where
they are buying entire companies.
And then the regulator's like, whoa, whoa, whoa, hang on.
You're getting far too much control here of a particular
situation. And this isn't good for
competition or for price points to consumers and all the rest of
(18:27):
it. But here, now this is NVIDIA
just chipping away buying a initially a 2% stake, right?
That's not that's not a big enough stake for NVIDIA to be
able to control anything that open AI does.
And open AI have got plenty of other investors, Microsoft,
Oracle and so on, right. So I think this is an
interesting play where it kind of falls under the radar, I
(18:49):
would say, at least for now, thesort of antitrust police.
And really it's more of a kind of, I guess it's a compute land
grab for NVIDIA. You know, they're saying, well,
open AI, you know, they're, they're, they're a different
vertical to us. You know, NVIDIA, they are the
they, they're, they're the kind of infrastructure build out for
(19:12):
AI, right? Open AI.
Well, they're the product that'sthen the app that sits on the
top of this stack. They're the consumer facing
entity, right? At least for now, they're the
biggest. So NVIDIA are saying, look,
let's lock them in to our vertical infrastructure stack.
So that we're kind of making sure we've got we, we've got a
(19:35):
continued majority share in thatcompute land grab that's kind of
going on all over the world and we'll come on to it.
But NVIDIA, they're not just investing in open AI, right?
But there's deals going on all over the place here.
And it's all about trying to maintain their kind of dominant
market share on that compute site.
(19:56):
Yeah. I just find, you know, working
in the strategic department of NVIDIA at the moment must be
such a such a like a brain box of like intelligence going on
and how to maximize this becausethey announced that partnership
with Alibaba. To bring its physical AI
development tools, so thinking things like autonomous vehicles,
(20:17):
robotics, smart spaces and into Alibaba Cloud.
And, and again, it's, I guess like you were saying, it's this,
you know, you think about China,this is a major global
distribution channel for your GP, us, your AI software stack.
So it's all kind of, it's all sort of feeding into this same
theme, isn't it? Yeah, that so that, that AI is
(20:40):
global, right. I mean, all right, we spend most
of our time talking and thinkingabout the US and fine, they are
leading this race, but obviouslyit's a global thing.
And so if NVIDIA can get a big slice of some of the action in
in the China market via the Alibaba tie up, well then
obviously that that kind of makes a lot of sense from that
(21:02):
sort of global perspective. Yeah, Yeah.
Just going to say like a bit of detail there, like locking in
the timing is interesting because on the Alibaba side,
there's this Quinn 3 Max model roll out there in the midst of
the moment and it claims to outperform rivals.
And this is like we'll see a keyuse case of AI.
It's encoding in in autonomous agents tasks.
(21:26):
And so if they can get in now and dominate, what is China's
top AI players? A secure critical role
empowering their cloud systems. And it's a stroke of genius.
Yeah, unlike they've also taken up a 500 million or they're in
talks of doing a 500 million, taking a 500 million stake in
WAVY, which is AUK based kind ofsort of car basically a break.
(21:53):
Like it's like rivaling Tesla's kind of autonomous vehicle sort
of brain, if you like. And they're taking a stake here
in the UK. So again, that's like just
trying to diversify outside of obviously Nvidia's big customers
are the Super scalers. They're the Microsofts and the
(22:14):
Amazons and the metas and and the Googles, right.
So it's just trying to diversifythemselves away from just those
kind of big tech super scalers and and like be a bigger wider
part of this broader AI future. That's obviously sprouting out
in many, many directions. One of the big things here, the
risk factors you said about on the regulatory front.
(22:37):
So also what I find is really fascinating about Nvidia's
approach is they invested 5 billion into Intel.
So they struck A partnership to Co develop some custom GPUs that
could easily integrate into Nvidia's GPUs.
Now, what was, I think it's really clever about this, is
that the deal? So the context, the US
(22:58):
government said it's going to take a 10% stake in Intel.
This was like a month ago, that deal converted nearly 9 billion
in promise grants and equity andmade the Commerce Department the
Intel's largest shareholder. So this latest agreement,
Nvidia's done. I think it just shows how a lot
of these executive leadership teams, they're viewing efforts
(23:21):
to help Intel as a little bit ofa let me help you, Trump.
You've got objectives and ambitions.
Let me align with you. And you know, one of Jensen
Wang's top priorities, of course, is he wants to export
Blackwell chips to China. And so how does he get
favourable terms? Well, look, if I scratch your
back, you scratch mine. So again, it makes a lot of.
(23:42):
Sense 100%. This is a big, you know, Jensen
is playing at the top political table.
You know, he was part of Trump'sparty on this state visit here
over to the UK. He is 100% playing a political
game here. And that China export sort of
ban. I mean, basically they're doing
(24:03):
it that they're negotiating. But basically the what Jensen's
hoping for is Trump to agree that they can start shipping
Blackwell chips, but not not full, not Full Monty Blackwell.
Apparently they they need to be 30% less capable from the Full
Monty Blackwell, I think this isthe number that's being
negotiated. And Trump's going to say, OK, if
(24:23):
you if you kind of, you know, dumb these down a touch, fine,
you can send them. And of course, don't forget also
Trump's done a deal with Jensen where they're going to get, I
can't remember the figure. Was it 30% of revenue from chips
getting sold to China goes to the US government.
So this is getting crazy political.
(24:44):
And look, we could have a whole conversation on a tangent here
about the US government's role in essentially what are private
and public companies and whetheror not that is a good direction
for the US government to be taking.
That's a whole different conversation.
But right now, Jensen's got to play the politics game.
And you know what? Looks like he's playing it
(25:07):
pretty damn well to be fair. He's at the top table.
So the, well, maybe just to pullall of this together, then
there's kind of five components of it this that this particular
story to summarize because it isa really fascinating 1.
So #1 is, is the partnerships. Many of these different firms,
we said, so NVIDIA investing andpartners like Open AI Alibaba to
(25:31):
secure chip demand, reinforce its role with the backbone of
this build out this infrastructure for AI #2 compute
power. That's obviously the priority
here. You've, you said this in nearly
every episode. You just can't keep up pace at
this point. It's the new battleground.
It's kind of massive energy financial costs shaping the, the
competition landscape, the circular loop.
(25:52):
I guess there's pros and cons. Here's there's the the funding
customers who reinvest in your demand makes a lot of sense.
But also I can see the point. I'd love to get people's opinion
in the comments section about capital recycling rather than
true underlying growth and value.
And then #4 the political leverswe've just been talking about,
(26:13):
you know, the Intel state, UK investments alignments, USUK
government. Why do they do this?
Well, they want to win favour. They want to reduce kind of
regulatory friction headwinds. And then finally, interestingly
for the investment bankers listening is NVIDIA avoids large
scale acquisitions. So one of the stats I saw here
(26:34):
and a shout out again to to Saabour our researcher, he said the
2020 acquisition of Mellanox is so far Nvidia's only M and a
deal valued over $1 billion. Yeah, I mean, we're talking here
with this latest pop this week. Nvidia's a four and a half
trillion dollar company and we've been used to seeing big
(26:56):
tech like Facebook back in the day gobbling up the Instagrams,
the Whatsapps. These guys have only spent
evaded on one deal over 1 billion and here we are it's.
Amazing. A four and a half trillion
dollar company that spews out 100 billion of free cash flow
every 12 months, and yet they'veonly ever spent a billion or
more on one acquisition and thatwas 5 1/2 years ago, which was
(27:17):
before this AI kind of accelerated boom even began.
Of course, this is pre ChatGPT. I'm going to put two other for
thoughts on this to kind of wrap.
Number one is this and I was reading the Lex column in the FT
which is I'd recommend. Let me guess, they're going to
(27:39):
be critical. Let me guess.
Oh, they're going. To be critical, well, that's it.
So you're going to understand their flavour, you got to
understand that they they need to.
Look at sun shining. I know it's cold, but it's it's
not raining. Come on guys.
They're basically saying, look, this circular thing as in open
AI here's 100 billion. Oh, you got to give it give it
us back though, and that'll be Walmart that up as revenue.
They're like, come on, they're basically calling this, and I'll
(28:02):
quote this is partly theatre to keep the fly, the flywheels
spinning. So to keep this whole kind of
story about the AI revolution going, obviously NVIDIA needs to
carry on growing. So how are they going to achieve
that? They're going to fund other
people to then self funds their revenue growth.
(28:26):
So that's one, I'd say that's one angle.
Here, welcome to something that's been happening through
the history of time. It just happens to be in the
context of AII mean. I love it when market
traditionalists, it's kind of like market traditionalists sit
here, the crypto crew and NFT groups sit here.
This AI one's kind of somewhere in the middle.
And it's like as soon as it's threatening traditional
(28:48):
landscape of finance, they're like, Oh no, it's all bad.
But yeah, go on, you have a second point.
Yeah, boy, you're right. This isn't you because Lex were
referencing A Nortel Lucent dealin the early 2000s.
I won't go into it now, but they're basically saying it's
the same again, this is good, this isn't good.
The other part is, is really outof there.
Well, it's kind of out of Nvidia's control to a degree for
(29:12):
all these. You can, you can build all the
data centres you want, right? But is there enough power?
Is there enough electricity? Number one, is there enough
energy and #2 is there enough infrastructure to get the energy
to the data centre? So there are there are big
infant like power and energy infrastructure challenges here.
(29:36):
So again and partially this is about the government and
partially this is about the regulatory environment around
getting permits to a increase energy kind of production, but
also permits to kind of transition this power through
through the grid and through power lines.
So all of this stuff is a definite risk to this huge
(29:59):
ambition of like building fine 10 gigawatts of data centres,
yeah, go for it. But is it going to even be
possible from a power point of view?
So that's another risk to factoring.
And just to round up your man Sam Altman and his, his
relationship as chair just so happens to be chairman of OCLO.
(30:20):
That's that, that that nuclear technology, right fusion power
company. So he, he is got this all worked
out for such such as the Albans and the Musks of the world.
They're certainly have got all bases covered here to to meet
your second point. But look, just quickly to wrap
in about 5 minutes, just a storythat I thought I didn't want to
(30:42):
go into it too much, but just asa headline, HSBC, they've
achieved a breakthrough in deploying quantum computing in
financial markets. And I thought it was interesting
because it felt like quantum wascoming up and then the AI just
kind of superseded and it's goneback a little bit.
So this was an interesting development because I remember
(31:03):
watching a documentary about HSBC&IBMIBM being one of the
forefront kind of, well, at least U.S. companies in the the
quantum space. And this was years ago.
They were talking about it, but basically IB Ms. Heron quantum
processor and they're using it to improve bond price
predictions. So one of the things here, and
(31:25):
the reason why it stood out thisweek is the bank applied quantum
processing to an anonymized set of European bond trading data
and they found it can significantly enhance the
efficiency of the market with a 34% improvement in predicting
how likely a bond will trade at a given price.
Right now, I've always struggledto understand conceptually what
(31:49):
is quantum computing. So this is the best effort that
I can do for anyone who's still was in that camp I was in not so
long ago. So a classical computer works in
bits. Each one is either A0 or A1 and
it tests solutions at A at a time in a sequence.
A quantum computer uses qubits which thanks to superposition
(32:13):
can be 0 to one at the same time.
Now my first question was what the hell is a superposition?
So the easiest analogy I could come up with is think about if
you flick a coin. Now a normal computer is like
flipping a coin and you have to wait for the coin to land right
before you can then determine, oh, it's a heads or a tails.
(32:35):
So you have to flick it, it spins and it lands and then you
pick it up and flick it and it'sone sequence after another.
But each event has to occur in in a in a structure in a quantum
computer. Think about it again, me
flicking the coin, but while it's in the air, it's at at any
point in time, it's rotating through both heads and tails at
(32:56):
the same time. So this in between state, if you
like, is the superposition right?
You're still with me, Leonardo DiCaprio.
I. Mean I like.
It I was just thinking best. Description I've heard.
Carry on. I'm thinking inception here.
Quantum computers can can basically use that state the
(33:17):
superposition, to explore many possibilities at once instead of
waiting for the coin to land. So it means that you can solve
problems exponentially faster. Like when I mean fast, I mean,
you know, dramatic. Volt.
You same Volt fast. So tying this back then to
markets what you know when I wasreading this, I was like, OK, so
(33:39):
so So what does this mean for financial markets?
You mentioned the bond pricing. So first of all, who's looking
at this? HSBC has been a bit of a front
runner. They've been doing this for
several years. JP Morgan, Goldman's, they are
actually all investors in quantum themselves in house.
So the ways to think about this are risk management.
So think about if you were testing scenarios, but on an
(34:02):
instant basis, imagine if a piece of news broke and you
could run it into your system and automatically then get
outputs of it might mean this orthat correlated effect or
second, third order impacts and they happen immediately as
trading signals, for example, inan automated way, price
(34:22):
prediction accuracy obviously could that where this is what
this data were was saying, but Iknow you had a third one as well
that some of these banks look at.
What with regards to what? What as in price prediction?
Or as in fraud, wasn't it like? Fraud.
Oh, I see. Yeah, well, fraud detection and
security, yeah. I mean, analysing patterns and
(34:43):
kind of building new encryption and just just kind of massively
upping the game with regards to fraud detection, which of course
can save them a lot of money. So yeah, that's another angle to
it. Yeah.
And and so when I was thinking of this, I was thinking, OK, so
thinking about a lot of our audience, Well, what, what does
(35:03):
this mean? Like, why should a student care
about this? And it's this kind of
intersection with tech and finance.
And I think most students have got got a finger on the pulse.
You know, the world's moving in a way where concentrated
researchers, analysts, these areall quite key.
So typically programming skills,advanced mathematics,
(35:23):
statistics, leaning towards AI, machine learning, these types of
principles. I thought myself when I read
this, I was like, Oh my God, like what about people like me
who don't have any of these skills because I'm not a
mathematician by trade. And So what I thought was
interesting and just to finish and share, I was just, you know,
talking to AI going, you know, this is the scenario like, yeah,
(35:45):
what type of roles can I do thenwithin a financial institution
that yes, we've talked about open AI and we're talking about
all these, it's all very sophisticated, like frontier
technology. But what about where does that
leave then the humans element? And I thought this was
interesting from from 5 different roles that you could
(36:07):
think of. The first one is products and
strategy. So there's one thing building
this stuff out, but like you said, the Huang and Altman
mindset of OK, how do I engineersituations, the strategic
elements of deal making and strategy creation, like that's
(36:28):
very much still led by experience and and organizing
teams getting input, stuff like that.
So, you know, deciding where quantum should be applied, where
it should be applied is just as important as being sure it can
actually work and do it's thing.So strategic priorities, client
facing sales roles. So, you know, I won't mention
(36:51):
who, but I'm doing some actual coaching on, on public speaking
and how to communicate with a fairly sophisticated group of,
of, of high end quants because they're struggling to really
articulate to PMS in a way that the PM can understand very
quickly. So again, being able to
understand this stuff, maybe notbe a technical person who's, you
(37:12):
know, an engineer, for example, but enough where you can
articulate it to other people and also to clients, obviously
external. Then there's regulatory risk
oversight. That stuff's only going to get
more onerous as technology becomes more influential.
Project management, you know, coordinating all these people
and then communication and education, you know, what about
(37:35):
these bits about internal comms,your staff training, you know,
these translators. I, I think it's going to become
a really important skill as well.
So look, overall, I, I don't want people to be spooked too
much. So they've got to go away and do
a PhD in maths that for sure is going to lock you in on that
Open AI or what was it that $100million salary at Facebook?
(37:55):
Sure, but that's real a place for, for, for me in this world.
I'm, I'm thankful to say, but sowe'll wrap it up there.
Piers, pleasure as always. Thank you very much.
And any comments that people have, just let us know.
We'd love to know, and we'll seeyou next week.
Thanks a lot, catch you later.
Popular Podcasts
My Favorite Murder with Karen Kilgariff and Georgia Hardstark
My Favorite Murder is a true crime comedy podcast hosted by Karen Kilgariff and Georgia Hardstark. Each week, Karen and Georgia share compelling true crimes and hometown stories from friends and listeners. Since MFM launched in January of 2016, Karen and Georgia have shared their lifelong interest in true crime and have covered stories of infamous serial killers like the Night Stalker, mysterious cold cases, captivating cults, incredible survivor stories and important events from history like the Tulsa race massacre of 1921. My Favorite Murder is part of the Exactly Right podcast network that provides a platform for bold, creative voices to bring to life provocative, entertaining and relatable stories for audiences everywhere. The Exactly Right roster of podcasts covers a variety of topics including historic true crime, comedic interviews and news, science, pop culture and more. Podcasts on the network include Buried Bones with Kate Winkler Dawson and Paul Holes, That's Messed Up: An SVU Podcast, This Podcast Will Kill You, Bananas and more.
24/7 News: The Latest
The latest news in 4 minutes updated every hour, every day.
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com