Subscribe
Share
Share
Embed
Around the Desk: This is where we at Avory think out loud, challenge narratives, and look for signal through the noise. Each episode, the Avory & Co. team dives into what’s moving markets, how companies are performing, and where opportunities may be forming. We break down earnings, macro trends, and investor sentiment — all from the lens of a concentrated, high-conviction portfolio.
*** The views expressed on Avory Podcast: Around the Desk are those of the hosts and guests and do not constitute investment advice. This podcast is for informational purposes only and should not be relied upon to make investment decisions. All investments involve risk, including potential loss of capital. Avory & Co. may hold positions in the companies discussed.
Sean: All right.
Uh, I'm just making sure it's doing
what it's supposed to be doing.
Say something
Luis Alvarez: Yes
Sean: All right, cool.
All right.
Are we in a bubble?
Is the rest of the market
about to participate?
That rotation we have been talking
about here at Avory Are the highest
quality compounders being mispriced?
Is identity, identity the most
underappreciated layer of the AI stack?
Maybe, maybe not.
We think so.
And is real-time data finally pulling
policy towards reality, something we
have been waiting for, for a long time?
Those are the five questions we are
working through today as we talk about
our Avory second half views, a review
of the first half, and that is that
Luis Alvarez: Welcome back
to Avory Around the Desk.
I'm Luis Alvarez, co-founder
of Avory and Company.
Each episode, I sit around the desk with
my co-founder and chief investment officer
of Avory we work through one important
investment question using data, first
principles, and long-term thinking.
My job in these conversations is
to question every thesis, push
on the assumptions, and make
sure the framework holds up.
Today, we're gonna walk through Sean's
and Avory's mid twenty twenty-six macro
view, it's been a volatile year, and look
at the news that mattered last quarter and
what it means for how we are positioned
Sean: And before we go further,
you know, a quick disclosure.
This discussion is for educational
purposes only, should not be
considered investment advice.
Opinions reflect our views as
of the recording th-- uh, of
this date, and those may change.
Uh, you know, please do your own research,
consult, you know, a professional before
making any investment decisions whatsoever
Luis Alvarez: John,
let's start at the top.
When you look at the market
today, what do you actually see?
Sean: Yeah, it's a good question.
You know, it's a broad
market, a lot going on.
You know, I see a market that
continues to be very bifurcated.
You know, Larry Robbins here recently,
I think it was, uh, you know, this
week, um, as we record this, you
know, the founder of Glenview Capital,
you know, a very prominent firm.
You know, he said it plainly
as, as anyone honestly recently.
He's been doing this for decades,
you know, very successful in his
own right and, you know, he said
where he's never seen this level of
dichotomy, uh, in the markets at all.
Um, so you're seeing a lot of
separation between, you know, kind of
winners and losers, and really those
winners are only in the AI trade.
Um, you know, if you, if you just, uh,
stopped, uh, you know, this quarter,
let's say, uh, and you looked at it,
you know, technology was the only sector
that actually, you know, beat the market
and in that, you know, was mostly AI.
So, you know, look, when a long tenured
fundamental investor, fundamental
again focused on the companies and
the, you know, the economics of
these business is talking about how
we are in such unprecedented times
as it relates to movements, I think
that's worth analyzing and reviewing.
Obviously, we sit on the side of, you
know, uh, appreciating those comments.
On the other side, I think sits, you
know, everyone else in terms of, you
know, what's happening in the markets.
You know, really for the last of the,
you know, eighteen months or so, AI has
effectively been the only game in town.
And at, at one sense, you know, that
seems rational from a trading perspective.
If you're a trader, you know, if the
consumer, the economy, inflation were
real concerns, I think the rational
thing to do was, uh, you know, bet on
businesses that were somewhat immune to
kind of the shorter term and medium term,
you know, macro noise, you know, tariffs
on, tariffs off, you know, war on, war
off, ceasefire this, ceasefire that.
You know, the build-out though in
AI Itself, you know, is the story.
Um, you know, as we're doing this, you
know, Meta's talking about, uh, you know,
releasing a, you know, AI infrastructure
product that would-- they would sell
as a third party, uh, as a service.
Think of AWS, you know, you know,
twenty years ago or fifteen years ago.
Um, but in general, you know, the
build-out itself is really the story.
A lot of capital was already isolated
for data centers and model build-out,
so it looked more like a long duration
infrastructure cycle than, you
know, a traditional cyclical trade.
Um, and where it kind of gets
interesting is, look, AI is a structural
theme, yet trading appears to be
central to the AI investment story.
In most cases, though, you know,
it's hard to gauge the duration
and the durability of the revenues,
the margins, the return on invested
capital across that entire AI stack.
So a lot of hope, uh, in terms of
duration and making sure that, you
know, these returns happen over time.
I think it's clear, like someone
such as NVIDIA will benefit.
But ironically there, ironically,
you know, NVIDIA has been one
of the laggards in that space.
So that alone I think tells you, you
know, investors are chasing, you know,
hot stocks rather than focus on, you
know, the long duration structural themes
where, uh, you know, the probability
of success there, uh, is high, you
know, five, ten, fifteen years out.
Um, and you know, I think that's the
discipline we're trying to bring to the
table and it's, you know, it, it doesn't
show up in some of the parts of the
markets, but I think that's ultimately
what's happening in the markets right now
Luis Alvarez: And what's on
the other side of that ledger?
Sean: Um, I'd say, you know, the other
side, uh, that has more lag, which
has kind of just been everything.
You know, energy had its day in the sun,
you know, because of the war with Iran and
I think, uh, the, the risk premium, uh,
that that put on, you know, oil and, you
know, the commodity space more broadly.
But now that's slowed down.
The ceasefire was signed in June.
You saw actually how, you
know, our portfolio reacted
positively to those news.
Again, something we've been highlighting
as kind of a meaningful headwind time
and time again over the last, you
know, twelve, eighteen months where
you had again tariffs and then war.
Both kind of the same idea, right?
War, tariffs both in, in theory
from a, you know, a textbook
perspective are inflationary.
But, you know, when you start to look
underneath the surface, uh, outside
of just energy being the dominant, uh,
force in kind of like headline inflation,
CPI, uh, you know, most of the other
components, a lot of the components,
um, you know, are, are rather stable.
Uh, now the, you know,
Brent is below eighty.
Uh, we know that.
We've been talking about futures
curves and those ended up being right.
Um, you know, it is a long way
from where we were months ago.
And then outside of AI, you know,
leadership has been pretty choppy
and very sensitive to rates,
uh, you know, inflation and
kind of geopolitical headlines.
So you have this two kind of track
market and the gap has been pretty wide.
We shared a chart on this recently
that I, you know, point listeners to.
You know, if you look at, you know,
year-to-date returns, especially
all of the index gain has come
from AI infrastructure players.
So you strip out the na-names out
there, uh, in the market, uh, that
are not AI infrastructure, uh,
and what you find is actually a
market that's down for the year.
You h-- you see Magnificent Seven taken
as a group, you know, they're having
their worst first halves on record.
And, you know, discretionary financials
and, um, you know, communications,
uh, are all down for the year.
So that's the dichotomy that I think, uh,
you know, Robbins from Glenview that I
was mentioning earlier was talking about
and I think it's showing up in the data.
But we do have the ceasefire now,
and I think what you've seen over the
course of the last several weeks has
been more of what we think is to come.
We had that "The Rotation Is Here" piece
that we wrote on the Investment with
Data newsletter here recently, and we
think, uh, you know, outside of any other
flare-ups as it relates to geopolitical
and/or, you know, tariffs, I think, um,
you know, the, the, the path of least
resistance is, you know, probably on the
side of the laggards versus the leaders.
That doesn't mean the leaders, uh,
you know, are ones that get, you know,
totally smashed by any stretch 'cause
there's still demand in that ecosystem.
But I, I think that's the way,
uh, you know, we're looking at it
Luis Alvarez: Does that mean
that the AI side really is
macro immune as people think?
Sean: Like is, is the AI side
really macro immune or is like,
uh, or is it the opposite?
Yeah.
Um, no, I-- so I don't think
it's truly macro immune.
Um, you know, it's
well-positioned but not immune.
You know, if there were serious macro
accidents out there, I think the
build-out would slow because, you know,
the player or the payers, uh, of the
build-out are hyperscalers, Amazon,
Google, Meta, Microsoft, Oracle.
You know, they're the ones writing
the checks ultimately here, and they
are very much subject to, you know,
enterprise and consumer demand.
Um, so you can't really have, you know,
a weak macro, uh, economy and hope that
the build-out continues at the same rate.
And again, I think people are just
playing the rate of change game here.
Um, you know, an example of
this I think captures the issue.
If you look at street estimates
for someone like SK Hynix, which is
obviously big in the memory space, has
done a really, you know, good job, you
know, over the last several decades.
You know, you have earnings going
from, you know, call it five hundred
dollars a share to roughly, you know,
twenty thousand dollars at peak.
Uh, so you can hear the
magnitude of that move.
But then the estimates are
falling back to around a thousand
dollars a share in the out years.
So round numbers here, but
directionally, that is the shape.
So yes, you're getting, you know, a
short, shorter term, you know, magnitude
lift in, in profitability bu-- uh, you
know, driven by kind of this supply chain
bottleneck that everyone talks about.
But normalization is
also expected thereafter.
So, you know, it's essentially a one-time
one X upside, even though in that cycle
it's gonna feel like, you know, a ten X
upside or even, you know, more than that.
Uh, and that forces a valuation question.
You know, do you value these
assets on a cycle average basis?
The way we typically here at Avory analyze
businesses with clear, you know, cycles.
I think you, you wanna take the cycle,
try to estimate it, what is that average
through line of, of earnings and,
you know, apply a multiple to that.
You know, in our view, I think
that's what you should be doing.
I think that's what maybe
people are doing in Nvidia.
But the right framework is to predict
directionally what peak revenue could
look like through the cycle, what peak
and trough margins could look like.
You know, take that median margin,
assign a full cycle multiple, and
again, you start to see hints at
the, that same tension I think you're
seeing in Nvidia that I just mentioned.
And we keep hearing good things
about AI, but the price action in
Nvidia, I think is, is, uh, you
know, kind of what it's been doing.
You know, it's been kinda
somewhat flat for the last year.
Um, you know, that gap between, I
think, narrative and what's happening
in the market is a tell to us that,
you know, we're going from, you
know, what seems structural and
durable to, you know, bottlenecks
which are much more short-lived,
uh, in the grand scheme of stuff.
So I think that's ultimately how,
you know, the immune, i-immunity of,
uh, the AI trade and, you know, h-how
you should probably be looking at,
uh, measuring the valuation there.
Luis Alvarez: And speaking of
the buildup, the numbers this
quarter have been hard to ignore.
just marked up the hyperscaler CapEx
forecast to roughly seven hundred
eighty-five billion this year, closing
in on a trillion by twenty twenty-seven.
How, how do you begin to
process a number like that?
Sean: Yeah, it's huge.
You know, a few ways though, you know,
first if you, if you have to ask whether
the level of spending is supported by
underlying demand or whether it is, you
know, faith-based build, I think the
encouraging signal there is, um, is a
remaining performance obligations data.
You know, the hyperscalers added,
you know, something like, you know,
$700 billion combined in r- you know,
remaining performance obligations,
which are these are again contracted,
you know, obligations that their
customers, uh, will essentially
be, um, spending, uh, over time.
And a- again, that's contracted, not
yet delivered revenue, so this is not
entirely a build it and they will come
story, and I think that's a good thing.
That is true backlog.
Uh, that's the way I'm
looking at that side
Luis Alvarez: What?
Sean: Yeah, I mean, but, yeah, I
mean, if you, um, if you have to
look at the cash flow picture, the
hyperscalers are now spending roughly,
you know, 90-something percent of
operating cash flow on cap- CapEx.
So, you know, I think that leaves
very little flexibility there.
Uh, and the market reaction earlier
this year when those numbers were
first announced was, uh, a near,
you know, $1 trillion tech sell-off.
Uh, investors, I don't
think, are blind to it.
Uh, that's why they're asking
questions across the board.
I think the data suggests the build-out
continues, but the bar for execution just
keeps rising and, you know, that's the
danger side of, you know, the AI trade.
The single most, I think, watched data
point over the next, call it six to eight
weeks, really is earnings here, um, as
they start in the next couple weeks is,
you know, what Microsoft, Google, Meta and
Amazon say about their fiscal, you know,
20 to-- 2027 CapEx, um, you know, guide,
uh, in their late July and early August,
you know, earnings calls and some in June.
You know, any flattening of
those guides is, uh, I think a
signal to, you know, the markets.
It's probably a good thing for them,
bad for the infrastructure players
Luis Alvarez: Is that a reason to
trim AI infrastructure exposure?
Sean: Um, yes.
I mean, so like, again, if it's just
like the direct AI infrastructure
exposure, uh, I think it's, it's
clearly a reason to be selective.
The names that have, you know,
rallied the hardest, the ones priced
with continuation of this exact, uh,
you know, CapEx pace carry the most
risk into those prints, obviously.
And, you know, the ones with, uh,
diversified revenue streams or with
their own kind of like, um, you
know, captive demand, I think they
carry less risk associated with them.
And, uh, you know, ultimately,
this is where the framework I
think matters more than, you know,
what we read in the headlines.
It's kinda like what the framework
these companies are building and, you
know, how do you model these things out?
Are you short-term cycle, long-term
cycle, traders versus investors?
And that's kind of like the
whole, you know, back and forth.
Luis Alvarez: We've been talking
about first order and second order
beneficiaries for a while now.
Can you please walk through
what that actually means?
Sean: Yeah, I mean, look, first
order means companies, um, that,
uh, are building critical parts
of the compute stack itself.
You know, the chips, the memory, the
cooling, the networking, the power.
You know, some of that is going,
uh, to be very valuable, clearly.
Uh, some of it is going to
be commoditized as the supply
catches up to, uh, the demand.
And so we're cautious in the areas of the
first order stack that have historically
been commoditized, and we are constructive
on, you know, I, I'd say the platforms
with real proprietary positioning.
Uh, you know, Amazon's
building its own chips.
M-Meta's building its own chips, and
we just heard again here recently
that they're likely to-- going to sell
that to third parties in terms of,
you know, their, their compute stack.
Uh, they're also building their own models
and is focused on, you know, what they're
calling personal, uh, super intelligence.
So that is a different posture, I think,
than buying off-the-shelf compute.
We're also seeing, uh, I'd say
something interesting play out.
Many of the companies that started
off as chip makers only are now moving
into other parts of the infrastructure
stack, so they're entrenching
themselves, uh, deeper and further.
And I think that is, you know,
where some of the opportunity lies,
where you go from commodity to,
you know, necessity, let's say.
Uh, and it's also a signal of trying to
become, you know, more or, you know, uh,
less durable, um, as it relates to, again,
these cycle dynamics that are at play.
NVIDIA's going-- trying
to go full stack here.
You know, started as a chip manufacturer,
networking, and now, you know, software,
and then, you know, there's some
of their models that are out there.
So you can imagine them, you know,
going full stack, uh, end-to-end over
time, and what does that do to their
suppliers, uh, to some degree as, you
know, they continue to verticalize?
What always worries me, though, is Is
when you do get these kind of like sudden
demand and supply shocks is that they very
quickly test, you know, the elasticity of
supply, demand, and kind of that price.
Uh, they open the door for, you
know, enormous amounts of capital
to flood the system, uh, to really
solve some of those lowest friction,
you know, bottleneck points.
You know, HBM, which is, you
know, high bandwidth memory,
is the case in point here.
Uh, you know, we know
it's a real bottleneck.
I think everyone does at this point.
You know, it'll be a bottleneck for a
while, but the alternative solutions
are, are being funded at a record pace.
You know, Cerebras, which we talked
about before, uh, in, you know,
different podcasts and our, you
know, our newsletter, is not using
HBM at all, uh, you know, for
their wafer scale architecture.
You also have to assume that, you
know, the biggest buyers of chips
will, will seek alternatives as well.
Um, you know, Etched, uh, just came out
with, uh, their own inference product.
You know, they've been at it for
four years and already have a,
uh, a billion-dollar backlog.
Four years is not a long time, and
apparently they're running, you know,
their compute at, you know, fifty percent
less energy capacity, uh, and getting
much more throughput, uh, in terms of
how they're building their architecture.
Remember, you know, before LLMs came out,
there wasn't like dedicated architect--
like, you know, architecture for this.
Um, and so we're using what was there,
and now companies are building, you
know, purpose-built architecture for
this, and I think that's important.
You know, we heard Apple is looking
to negotiate The ability to buy from,
uh, you know, a Chi- a Chinese, you
know, manufacturer, uh, that is not
a, you know, net new company, uh, but
it would open the door to additional
supply for, you know, one of the largest
buyers of this kind of technology.
So you have to question, you know,
some of the durability, uh, in some
of the positions out there, in some
of the companies that, you know, that
are, you know, up and to the right.
You have to ask yourself, you
know, "Am I willing to own this
thing for the next decade and
feel confident in its durability?"
Uh, you know, for some of these
companies, I think the answer
is definitively yes for some.
Um, I would say for most, though,
it's, it's generally unknown.
It's probably a no.
And staying prudent, I think,
in that first order, um, you
know, seems rational here.
You know, we have shared that history
lessons in the past, you know, staying
patient, letting those cycles play out,
and seeing, you know, which companies
are, are most durable has historically
allowed, uh, you know, I'd say prudent
investors to participate, uh, in, you
know, that massive structural theme
without having to, you know, nail that
exact timing or pick the winner early.
You know, the internet build-out
I think was the clearest example.
You know, I think we put that in the
last, um, you know, annual piece.
Uh, you know, the winners of that
era were not even, you know, defined
until likely, like five years later.
Um, you know, many of those early winners
eventually faded, uh, into, you know,
mediocre businesses at, as, you know,
really competition started to scale up.
Uh, and again, were purpose-built for the
new era, uh, as the internet turned on,
and yet there were still plenty of stocks
that compounded, you know, five hundred
percents, uh, you know, years later.
We always have that Cisco chart where,
you know, the, the best money you could
have made in Cisco was actually five years
after, you know, the internet boom and,
you know, riding that from then to today
And you still, again, you still have these
opportunities as patient investors to let,
you know, some of the, uh, the return,
uh, on investment capital to play out.
We, you know, we think AI is a
structural cycle, so, like, we're
not saying that, uh, you know, we're
not, um, negative on AI at all.
Uh, we think it'll be with us for the
next, you know, five, 10, 15 years
and really our lifetime in different
ways, and we still think the layer-one
winners are being decided today.
You're seeing some of what
I'm describing in real time.
You know, Anthropic has quickly
surpassed OpenAI in run rate revenue.
Uh, there are some fair questions around
how each side calculates their run rate,
but it's pretty clear Anthropic has taken
mind share and budget share, and that just
shows you how quickly the space can move.
Uh, you know, you can be a leader or
the perceived leader and then quickly
be, be the laggard to the new leader.
And then on top of that, you have open
source models that are catching up to
the frontier models, and if we ever
hit a moment where a frontier bends the
curve slightly or stops riding kind of
like that exponential of intelligence
the way it has been, I think you change
the shape of the economics drastically.
And then token use, you
know, trends towards free.
We're seeing some of that already.
And the models start to act more like
the way we all act with the Internet.
Uh, you know, we use it, we live on
it, and the value gets built on top
of it, and that's probably where
we get into, call it, like, the
next phase that we've talked about
Luis Alvarez: And how
about the second order?
Sean: Yeah, so I mean, that
was the first order, right?
Iâ¦
A little long-winded there, but the second
order is where this technology actually
gets used, and it's the application layer
and the infrastructure around it, how it
gets protected and governed, controlled.
You know, that is where we spend,
uh, I would say a disappr- you know,
proportionate amount of our time,
you know, trying to find things.
Think of, um, you know, identity,
uh, and, and, kind of like
trust as clear examples for us.
You know, as more workflows get
delegated to agents, uh, you know,
the question of who is allowed to act
on whose behalf with what permissions
becomes a critical layer of that stack.
This is why we keep coming
back to names that we own.
So full disclosure, Okta and Clear, um,
you know, they've done, you know, a good
job at, you know, building verification.
You know, this week Clear
announced a, uh, inside of AWS
for, um, their identity product.
Uh, again, that's well
outside of the airport.
And then Okta, you know, continues
to build their, their stuff.
So on Okta specifically, you
know, we believe This is directly,
um, something we think about.
You know, we run our own agents,
right, internally at Avory Those
agents need to be governed.
They need to be given access rights
to, you know, files and permissions.
There has to be tracing, audit trails
of, you know, what an agent did and how.
Some of that sits adjacent to,
call it orchestration, like, you
know, what the agents are doing.
But the identity is the, I think the
deepest issue there because identity
is what governs agents doing real work
inside of enterprises and owning the
leaders inside enterprise identity, the
ones that have already built products
around agentic identity looks like,
uh, you know, a powerful tailwind.
Uh, I think that's an obvious powerful
tailwind and, uh, you still have
to pick the winners, let's say.
Uh, but someone that's already entrenched
there has a pretty good, um, trust
system with their un- underlying
clients who are gonna be the buyers.
Uh, you can also make the argument
that agentic identity gets pushed
towards a consumption-based model.
And if we end up with, call it, you
know, thousands or more AI agents per
huge-- per human, which is already
starting, I think, uh, Cloudflare has
a data point where there's more, um,
you know, bot traffic on the internet
than human traffic on the internet,
and that just happened here recently.
I think the opportunity, you know, for
that, uh, surface is that the identity
players, um, see orders of magnitude
larger volume, uh, than they see today,
and that could effectively become,
you know, a quasi-transaction payment
platform type style business model.
You know, pay per use or pay per identity.
You know, different shape than, you
know, the traditional seat-based
software model we're all used to, and
I think the CFO has to get comfortable
with that, but, um, of any company.
And so we may just end up back in
a seat-based model that has, you
know, clear spend trajectory versus
kinda this pay per use and, uh, I
think that's kind of the risk there.
But, uh, ultimately, that's how we're
looking at the second order effects
Luis Alvarez: And what
about the capital side?
Sean: Um, yeah, I mean, look, there's
a lot of capital flowing here.
Uh, you know, I'll, I'll stick one out
there that we obviously, you know, bought
here for the second time in our portfolio.
It's, it's Blackstone.
Um, you know, the financing
side of the build-out itself
is an enormous opportunity.
Clearly, you know, data center
development, power infrastructure,
the real assets that sit underneath
the compute, that is all, uh,
you know, part of the AI story.
And, you know, there's
economics that's tied to that.
Uh, and the firm has, you know,
been pr- you know, positioned
aggressively in that direction.
Uh, and so, you know, they're not all AI
compute and AI infrastructure, but, you
know, there's logistics in there, so it's
a diversified way of playing, you know,
call it, um, you know, the build-out
and the capital side of the equation
Luis Alvarez: And where has
the money actually been made
so far in second order AI?
Sean: Um, I mean, you have to say
enterprise, obviously Anthropic,
but, you know, the real part is,
you know, the application layer.
Uh, you know, there has to be ROI there.
Um, it's been visible in different
pockets, you know, coding
workflows, developer productivity,
you know, those have been the
standouts, call centers, et cetera.
You know, some of the marketing
stuff has seen, you know, ROI.
I think, uh, you know, some of the ROI
is clearly flowing to Meta's top line.
We talk about that.
It's in-- It's a holding of ours, but,
you know, they're-- again, they're the
second fastest-growing, uh, you know,
Mag Seven company, only behind NVIDIA.
Uh, they have their own chips.
They have their own models.
Obviously, they're doing work there.
Uh, they have the s- the best gross
margins of any of the Mag Seven,
and they're the cheapest there.
Um, so again, I'm throwing out that.
But, you know, in general, you know,
Anthropic's revenue run rate is, you
know, approaching reportedly fifty
billion as of, uh, you know, the
end of, uh, June, and that's up from
nine billion at the end of last year,
so it's definitely a hockey stick.
Uh, you know, that's one of
the fastest revenue ramps we've
seen in, you know, in history.
So you could argue, uh, you know,
enterprise and, uh, concentrated
in code marketing, uh, and kind of
like developer productivity, I think
is where, you know, much of the
second-order impacts, uh, are happening.
Uh, and, you know, we think the reward
there that, uh, you know, some have
realized has, you know, has been deserved.
Luis Alvarez: You've mentioned
Zoom as a second-order beneficiary
in the past, but people don't
always associate Zoom with AI.
Why does that fit, fit this framework?
Sean: Um, so Zoom has two things
going for it, I think in AI.
You know, the first is the more
underappreciated part of the story.
We've sent out a couple charts on this,
you know, so go check out our work.
Um, you know, the second is the
financial part of the story.
So it's, you know, product
side, you know, financial side.
I think, again, both
those things have to work.
Uh, you know, I'll start with
the underappreciated piece.
You know, AI's job fundamentally
is, you know, wrangling context
and spitting out intelligence.
And, uh, context is king.
You know, context is, you
know, data that can be parsed.
It can be analyzed, you know,
thrown into kind of like a, uh, you
know, a bl- a blender, but a model.
Um, historically, the context
that companies could actually
use was physical context.
Um, and, you know, then it, you know,
migrated somewhat into, like, files and
folders, and some of it was structured
like Excel sheets with rows and columns,
which is easy to extract, uh, 'cause
it's very well documented and siloed.
Some of it is, you know, is
from documents and other, you
know, less structured content.
We've, we've talked a lot
about data being unstructured.
A lot of it is what you call dark data.
We have that presentation we often do,
and we talk about dark data, and the world
of AI is opening up the doors to that.
And what is new is that AI has, has been
able to take a lot of that unstructured
data, extract those insights with a
pretty high probability of success.
And, uh, you know, I think the thing
that's missing is the, is, is the context.
That has historically, uh,
not been captured throughout
an entire organization.
And the context that gives a company, you
know, a full cycle view to iterate on its
own business and drive value is really
around the conversations we all have.
Um, you know, it is all the
conversations we have, right?
It's, uh, you know, in-person meetings,
sales channels, calls with customers,
calls with partners, calls internally.
I know some of this stuff is sensitive,
so there needs to be guardrails in
place, but historically, all of that
was lost, never captured It is only
really the last several years, really
think COVID, where everything went to
digital infrastructure to communicate
and produce, um, you know, work.
Uh, and you know, really now we're
starting to capture portions of
meetings, uh, recording capabilities
in person, in meetings a-as well,
depending on obviously where you
sit on the regulatory side there.
But all that information and context
is powerful, and having full cycle
obs-- you know, observation of company
context, personal context, puts Zoom,
uh, in a very unique strategic place.
You're hearing this
theme play out elsewhere.
You know, l-look at Salesforce, how
they're positioning Slack as the, the main
hub where humans and agents work together.
You know, if you, if you go on,
like, YouTube and you, you, you look
at, you know, some of the better
creators that are, you know, producing
content around AI, a lot of them are
using Slack, uh, to communicate with
their, you know, with their agents.
So agents sit inside Slack,
communicate as, you know,
the-they're a real person, let's say.
Uh, that orchestration and communication
layer I think is very important.
Zoom has Zoom Team Chat.
We use it internally, but Slack sinceâ¦
You know, Slack also has obviously, uhâ¦
Is it inside of Salesforce?
Both companies understand, I think,
that the surface where conversation
happens is very strategic.
Uh, Zoom also has gone from
kind of like a single product
to a multi-product platform.
We've talked a lot about this in the past.
If you look at their recent product
events, you know, they're extending into
documents, into notes, into, you know,
slides, like PowerPoint almost, and a
broader kind of like workflow suite.
So it's powering all of that with
AI in a really thoughtful way.
You know, the post-meeting workflow
is, is kind of that proof point.
You know, after a sales meeting,
historically, what would you do?
You would write notes.
Uh, you would write them by hand.
Uh, then you would sum-- you know,
try to summarize them or stuff
them into a folder or something.
Now the summaries can be actionable,
so an agent can take the summary and
say, "Hey, you know, look, I'm going
to send these three follow-ups for you.
Please approve them.
And by the way, I'm adding, you
know, these two tasks to your to-do
list, and I'll remind you in a month
to, uh, you know, follow up with
customers, you know, X, Y, and Z."
Um, and with that, you have, you know,
governance, reliability, innovation.
So Zoom checks, you know, kind of that
AI box from a use case standpoint.
They have brought in serious AI talent,
including, you know, many of the
people that worked on AI at Microsoft.
So you're seeing some, um-
Some talent migration to Zoom,
which I think is also a signal.
Uh, you know, they have made a
small acquisitions in like, uh,
different verticals to deepen some
of their, um, strategic place there.
And, you know, that is
just one half of the story.
I think the other half is
the financial equation.
I'll be quicker there.
But, you know, Zoom owns
a stake in Anthropic.
They've invested roughly, you know,
fifty something million dollars
about three, four years ago now.
You know, we've constantly highlighted
this, and at Anthropic's most re-recent
round, it was, you know, prior to
this, was somewhere around three to
four hundred billion dollars, and
then, you know, now it's somewhere
closer to nine hundred billion dollars.
And there's chatter around potentially
a multi-trillion dollar IPO for them.
And if those, you know, play out,
Zoom's stake could be somewhere five
to ten billion dollars, depending
on dilution, depending on the round.
Um, you know, that alone would
equal thirty, forty, fifty
percent of Zoom's market cap.
Then you add roughly, you know, seven
to eight billion dollars of cash
and no debt on their balance sheet.
You know, you net out cash, you net
out that Anthropic stake against
the free cash flow they generate,
which is, you know, one point eight,
you know, two billion dollars.
You're left with a business trading
at four to five times free cash
flow, netting out again, all,
you know, Anthropic and cash.
And that is the same business we just
described as a core context platform.
That's probably a good way to phrase
it, a core context platform for AI.
Um, you know, that's the financial
setup the market, I think,
uh, hasn't really priced in.
Uh, it's getting some love.
Obviously, it's the best software name,
uh, for the year for the most part.
Um, and, you know, I think a lot
of that has to do with, you know,
valuation and their execution and growth
re-accelerating and the Anthropic stake.
But I think there's more to do there.
But again, not a recommendation
Luis Alvarez: Let's switch sides
of the bifurcation for a second.
Real estate has clearly struggled.
Uh, what's going on there?
Sean: Yeah.
Uh, um, yeah, I mean,
look, we own Zillow, right?
So, um, that's something to,
you know, stay top of mind.
That's, that's been a, a weak
performer here recently as of,
as of really the war, right?
When the war started, you know,
uh, what you saw was the long bonds
stayed higher and, and moved higher.
Uh, people expected, uh, you know,
inflation concerns, um, to rise,
and they did rise, uh, the concerns.
Uh, along with the headline prints.
You know, you saw, you know, three
point four, you saw four point, uh, oh
on some of the, uh, headline readings.
You know, a lot of that had to do
with the energy cost and, you know,
people, uh, perceiving that, again,
the war co-could last longer or,
you know, fall off the wayside.
And we're still kind of in
this, um, memor-- you know,
memorandum of understanding.
Uh, and so we haven't have a definitiveâ¦
We have a ceasefire, but not a
definitive, um, you know, lock-in plan,
uh, in terms of, uh, uh, maintaining
the lid as it relates to, you know,
future, uh, oil prices, let's say.
And, you know, that's ultimately
kept a little bit of, uh, the 30-year
mortgage rate h- uh, elevated, uh,
even after, again, the ceasefire
here and loyal-- lower oil prices.
Typically, that there's a l-
a little bit of a lag there.
But we do expect mortgage rates
to migrate lower over time.
Uh, again, but that path has been
slow here and a little bit deviated,
uh, you know, due to the war.
You know, if you had toâ¦
You know, if you looked at Zillow or
just real estate, you know, assets
in general, it's not just Zillow, um,
you know, many of those that are tied
to, call it like residential real
estate, uh, you know, have struggled.
And, um, you know, so it's more
of, um, uh, industry, sub-industry,
uh, issue as it relates to that.
Uh, and again, over time, we think,
um, mortgage rates migrate, uh, lower
Luis Alvarez: You have a structural
view on rates that is different from a
lot of the macro commentary out there.
Walk people through it
Sean: Uh, yeah.
So we think, um, look, AI is
structurally deflationary.
I think e-even, uh, Elon Musk just said
this, you know, tweeted this the other
day or X'd, whatever you wanna post it,
um, said it the other day, and I agree.
I mean, look, technology has
always been deflationary, and
yes, you get a, a lift in, inâ¦
When you have a supply and, uh,
demand, um, uh, you know, mismatch
there, you know, you can have, you
know, memory costs rise, uh, and
that, you know, is inflationary
for maybe some of the components.
But net-net, I think, um, look, if you're
actually using this stuff, you see,
uh, it's increases your productivity
if you're using it in the proper way.
Um, so I think it's 100% a productivity
enhancer, which is naturally deflationary.
So there's-- I, I don't think
there's a question to that.
Um, the, the, these kind of like, um,
productivity improvements I do think,
you know, compound over time, though.
You know, they show up in, again, lower
unit labor costs per output, right?
Um, uh, and then, you know,
faster software cycles as well.
Uh, and then services, you know, broadly
just become more and more efficient.
And so again, as those gains work
through the data, the real rates,
uh, should continue to trend lower
over time, uh, even if kind of
like nominal rates don't move.
Um, and we think those implications of
that are, you know, healthy for, again,
rate sensitive areas, real estate,
certain cyclical areas, long-term duration
cashflow, you know, base businesses.
Um, I don't think that
path will be linear.
It hasn't been linear in the last year,
but I think the direction matters and,
you know, you probably wanna own leaders
in those, in those categories, and that's
kinda what, h-how we're positioned there
Luis Alvarez: Is that a
contrarian view at this point?
Sean: Um, I mean, obviously when we were
preparing for this, uh, I'd say yeah.
And then when Musk says it's deflationary,
I think, um, you know, it's hard
not to say that, uh, at least, uh,
you know, the, the wealthiest person
in the world thinks it, um, uh, if
that's, you know, too contrarian.
But I, it is pretty contrarian.
I think mostâ¦
There's, there's a whole group
of, uh, market participants.
There's a reason why, you know,
gold was, you know, uh, a high
flyer and then now it's crashed.
L- like, there's signals out there.
You look at, you know, um, gold,
you look at oil, you know, some of
these, uh, deflationary-ish signals,
um, uh, continue to come down.
If you look at, uh, breakevens on
inflation, you know, those have come down.
So, uh, you know, some of, um, the
market is starting to price a little
bit, and we think that'll ev- eventually
manifest into lower mortgage r- or lower
rates and then lower mortgage rates.
And I'm talking about real rates
which impact mortgage rates.
And, you know, we think again,
this is, um, you know, most macro
frameworks still treat AI as a, um, a
productivity tailwind that will show
up, uh, in a decade as opposed to,
you know, starting to show up now.
Uh, and I think if, look, if the US is
truly winning the AI race and they do,
um, I think what happens is, you know,
we used to use labor internationally
for the last decade or two.
Uh, if AI can achieve fractions of
that work, then you start to, uh,
pull some of that here and it just
becomes a productivity enhancer
internally and, uh, an export of quasi
service, even though it's digital
service, uh, the other way around.
And I think all of this feeds into
kind of this, uh, deflationary story
Luis Alvarez: You've been pretty
accurate on the consumer for a while now.
where are, where are we today?
Where does the consumer stand?
Sean: Um, yeah, we've-- Yeah, so
we've definitely been-- We've,
we've been very accurate there.
Uh, uh, uh, obviously frustratingly, um,
you know, you've, again, you've had these
bits and pieces of, again, tariffs and war
that have, uh, created, uh, uh, renewed
concerns of like what could happen.
And so again, people pull back to, you
know, what's working trading-wise as
opposed to focused on, you know, what's
cheap and high quality and durable.
Um, but the, you know, the consumer
continues to spend and remains,
you know, pretty healthy overall.
You know, this next quarter-- I mean,
this next month here is, uh, jobs report's
gonna-- is tracking really strong.
Uh, so you're gonna see, you know,
some pretty healthy job action there.
You know, the labor market
overall, you know, is, you
know, fairly healthy to stable.
Go back to our annual, uh, report that
we wrote for this year, and our view
was that jobs would start to pick up
here in the first half of this year.
Um, and that is showing up.
You know, wage growth has come, uh,
down or at least is stable to a, a,
a, a place where you're not really
concerned about inflation, and you're
also not concerned about, you know,
what that means for the consumer.
And then we've been, you know,
pounding the table that real wages
have been positive this whole time
and higher and, and more positive
for longer than the market has,
uh, I'd say, given credit, uh, too.
Because if you're focused on real-time
inflation versus, you know, um, kind
of this broken inflation, you know,
survey process, uh, I think those
two numbers tell a different story.
Again, we've been posting that chart
for a while, and I think that's why
we have said the consumer is healthy
enough and they're spending, and
that's showing up in the data while
everyone continues to be concerned.
But I think people are
coming around to that idea.
We heard the Fed talk about bringing in
real-time data to then help inform them.
Um, and so all of that is positive
news, and I think the bigger story
for us is that, again, the gap between
real-time data and lagged official
prints, uh, has been wide, but it's,
uh, now going to be something that, um,
hopefully is used more and more and that,
that, that, you know, data gap closes
Luis Alvarez: I guess it's
fair to say that the Fed has
finally caught up to that idea.
Why, why does that matter,
uh, for how we invest?
Sean: Yeah, they have.
Um, so we're happy about that.
Uh, I know you know that
I'm happy about that.
It's a positive shift.
I mean, I've been saying this and,
and it feels like the market doesn't
care because the market's more worried
about, you know, it, there's-- it's
been like this three-legged stool.
The, if, you know, the Fed reacts to
a database that is lagged, then the
market can focus on what the Fed's
looking at as the reaction function of
what they're gonna do with the data.
But if the real-time data is saying
something else and the Fed's not
partic- you know, paying attention
to that, the question is, what do
you pay attention to as an investor?
And again, this is more like trading
activity than long-term durability.
But if we're focused on real-time
data, which is what we care about
more, like what's actually happening
in the economy, that matters more.
But again, if the Fed's gonna react to
the other, then in theory, the Fed has
the chance of making a policy mistake.
Um, and so yes, it's a positive shift.
They've caught up to the idea.
Clearly, you know,
they've talked about it.
They have these little committees
that they're gonna run to, to see,
uh, how they change their process,
so there's nothing definitive yet.
Um, but they have acknowledged publicly,
uh, that real-time data is going to
be part of their process, and that
should lead to, again, policy that
is more tied to reality and less tied
to perception of the Fed's reality,
and that, that makes me feel a lot
better about, uh, a lot of things
Luis Alvarez: Switching gears a little
bit, Microsoft is going open source
on some of their Copilot products.
Uber publicly limiting AI
token usage internally.
What do you make of those signals?
Sean: Um, yeah, I mean, this is kind of
what we've been discussing for a while.
Uh, you know, these things are not
cheap, and I'm talking about AI.
You know, for AI to work, AI has to
have, uh, you know, return on investment.
You have to spend dollars, and there has
to be something in there, whether it's,
you know, optimizations and savings or
it's, you know, uh, new revenue streams
that are, you know, uh, uh, relative to
the size of your spend enabling something.
Um, and, you know, companies, uh,
enabling AI are valued at multiple
trillions of dollars at this stage.
That's a lot of ROA-- R-ROI,
um, return on investment that
has to surface to justify that.
Uh, there are plenty of case studies
that show real value being created,
and there's also plenty of, uh,
case studies that show the opposite.
Um, I think the ones that are not
seeing, you know, return on their
investment are typically, you know,
probably just didn't map out clear goals
around, you know, what they're trying
to achieve, uh, did not put guardrails,
uh, in their organization around usage.
Uh, I think it's very easy to spend a
lot of money very, very fast because
look, engineers, researchers, creators,
they're creatures of discovery.
So you give, uh, I would say us,
give us a tool, uh, creatures of
discovery that answers questions back
to back and reroutes us to new ideas.
I think, you know, you're just gonna keep
going down that discovery process and try
to build and create and build and create,
and that spend climbs pretty quickly Uh,
the good news is that the spend itself,
uh, is driving, you know, optimization.
You know, some of it is absolute cost, uh,
of frontier models, uh, even when those
models are efficient on a per-token basis.
So the absolute dollar cost there is high,
uh, even though again, they're, they're,
they're getting more, uh, efficient.
But the playbook, um,
becomes pretty simple.
You know, use the frontier to learn
and figure out, you know, what is
worth optimizing in your current
workflows or what you can build kind
of net new that creates real value.
Once that's established, you know, open
source follows quickly, which, you know,
you're hearing that left and right.
And if you have a clear, uh, skills
repository and a harness, you know,
around that, you know, what you're doing,
I think you can plug and play models or
even use a model, you know, orchestrator,
you know, to help you assist.
And that's becoming pretty popular here.
And a lot of that is going to be,
you know, open sourced over time.
And then the optimization pressure
pushes down to the compute layer, and
the compute layer is being optimized too.
And this takes time, but it is
the-- I think the pass, you know,
uh, path of least resistance here.
And so the Microsoft, again,
having Copilot go open sourced.
You have Uber again saying that they're,
um, gonna limit some of the spend there.
I, I think it's the coming to
light of, you know, none of
this stuff is gonna be free.
People are starting to, you
know, think through this stuff.
Uh, open source is becoming
more and more popular.
So if you are just running models
and making revenue just on models,
that is probably not sustainable
five, ten years from now.
And if it's not sustainable five, ten
years from now, you have to either
create new products for new revenue
or optimize the stack and put pressure
on, you know, the componentry that's
making up all, you know, your cost.
And so more revenue or cost optimization,
it's probably both, but ultimately I think
that's the path of least resistance there
Luis Alvarez: We hear it in
the outlets every week now.
Buffett, Ackman, others we won't
name, well-established fundamental
investors looking at this environment
and saying, "You have very dramatic
head flyers from one corner of the
market high-quality companies on
the other side getting punished AI
impacted, even though the revenues
and margins look healthy and durable."
Uh, what do you think about that?
Sean: Yeah, I know.
I heard the, um, the recent Ackman piece.
Um, you know, I tend to agree there,
you know, as a portfolio that, you
know, shares some of the same names,
uh, you know, I agree with that view.
Uh, you know, I've articulated some
of that here, uh, on, uh, already,
but, you know, the market has been
unforgiving to what I would call
high-quality durable compounders that
actually have the characteristics
you want in an AI winner, right?
So, like many of the things you're
looking for, and if you think about,
you know, those characteristics,
you know, what are they?
I think, you know, do
they have distribution?
Do they have proprietary data and
context that lives inside their
platform or data that they consume
in a, you know, privileged way?
Uh, have they shown the ability
to withstand pressure on both
the revenue side and the margin
side over the last several years?
Um, you know, when you channel Check,
like check, you know, different
industries against, you know,
their actual demand generation.
I think the top of funnel
has, um, you know, have th--
like has that been impacted?
And if it has, you know, has
that, you know, impacted revenue?
Um, and if not, you know,
have they found, you know, new
sources of opportunity there?
So look, we're big believers right
now in founder-led companies versus,
you know, just pure operators.
Uh, we think founders have, you
know, more skin in the game.
Any value loss in their equity is much
more impactful to them than to an operator
who is just mostly there for a paycheck.
You know, founders are typically more
willing to make the hardest decisions,
especially if they have control.
You know, you saw that at Block,
which, you know, kind of, um, again, an
investor-- an investment of ours, you
know, where they, you know, cut forty
percent of their staff to go kind of
like very, very AI native in the company.
And what you're seeing there is a
product cadence that I have never
really seen from them, and it's
showing up in the results and I'm,
you know, I'm, I'm happy to see that.
And again, it's a very
much a founder-led org.
Um, and it feels, you know, a very
entrepreneurial inside the stack as, you
know, many of the people also communicate,
uh, on X and, and some others.
So, you know, for a while,
control was viewed as a negative
from a governance signal.
Uh, but I think in this
environment, you know, in the
right hands, control is positive.
It, it really allows, you know, the
right calls to be made at the right time.
You know, on, on Berkshire specifically,
you know, they've had a, a tough run.
You know, Warren stepping down,
Charlie obviously passing.
You know, neither, neither of those
is good for sentiment around the
future of the company because,
you know, two decision makers.
Uh, you know, but Berkshire has
also been deliberately patient
on their own AI bets, right?
And I think that patience may
eventually look very good.
You know, we're, we're being patient
across, you know, that very deep, uh,
technical infrastructure layer, just given
the fact that there's still a lot of known
unknowns, uh, and the market is, you know,
just pricing, you know, bottlenecks today.
And so, you know, again, you know,
circling all the way back, I think,
you know, Ackman and some of the others
talking about how high quality compounders
have been, you know, maybe not, um, left
for dead in entirely, but, you know,
have been, um, not in focus, you know,
just due to the, you know, the AI cycle
overwhelming the headlines overall.
Luis Alvarez: Thanks, Sean.
close with, uh, rapid fire.
questions, short answers.
Is AI a bubble?
Yes or no
Sean: Um, uh, uh, you know, we asked
this question when we were at the--
when we rang the bell, which was cool.
Uh, but, you know, I don'tâ¦
I, I-- no.
I don't think AI, like,
AI is not a bubble.
You know, the value AI creates
relative to the global economy will
likely be larger than it is today.
And yes, there's pockets of probably
severe speculation at the company level.
Valuations and stories obviously can get
ahead of themselves across the board.
And, umâ¦
But if you look at, you know,
the eventual size and scale of
this, of the opportunity, I think,
uh, it's, it's, it's massive.
Um, I think the job is, again,
trying to separate sentiment,
emotion from structural durability.
You know, feeling like you're,
you're missing something, um,
for the sake of just missing it.
Um, you know, it's funny, if you look
at, like, something like Bitcoin five
years ago, three years ago, two years
ago, you felt like you were, you know,
missing it, and, you know, it, it's
down from where it was five years ago.
Um, right now, again, these are
different, different things, but, you
know, similar in terms of, um, you
know, maybe the feeling emotionally.
But right now, look, most of that
work is not being done, and I think
largely because it's generally hard
to figure out who wins and who loses.
So the market defaults to chasing
the next high flyer and the next
bottleneck, uh, along the chain.
Uh, and that's kind of like an easy
path to look at charts as opposed
to trying to, you know, roughly
forecast what's gonna happen next.
Uh, I know that was not a short,
rapid-fire answer, but yeah.
Luis Alvarez: What's the most
underappreciated name in our book?
Sean: Um, I mean, I'll say Zoom just
'cause, you know, we just walked
through it, uh, so there's a little
bit more context to it, and context
is king, and context is king for AI.
And I think the market is still pricing
it as a video conferencing company.
We see a context platform with a
balance sheet kicker that has, you know,
Anthropic and cash and, uh, business
that is, you know, doing pretty good.
So, uh, an evaluation that makes sense.
So, you know, that's our view.
You know, it doesn't have to be
your view, but, uh, you know,
listener, but, um, that's our view
and it's, uh, a big holding for us
Luis Alvarez: Biggest macro risk
people are not talking about
Sean: Um, I think the flattening of the
hyperscaler CapEx guides in, in, you
know, at some point later this year.
Uh, even if like, like if one of the
four, you know, big hyperscalers, um,
send any sort of signal about the rate
of change for CapEx, I think the AI
complex reprices really, really fast.
Um, and I don't think a lot of
people are positioned for that.
Um, I don't know what
you do with that per se.
I think it's good for hyperscalers, right?
Uh, 'cause the Mag Seven's gotten hit,
but I also think what does that do to
the AI complex as more new entrants in
kind of inference continue to develop
and build, uh, in training as well.
More Neo clouds are being built,
all while there could be signals.
And again, a lot of it'll be
de-- uh, you know, predicated on
what's happening in the Mag Seven.
So if we're going into these prints
and their stocks, you know, I think
Microsoft just had its worst quarter
almost like ever or something.
Um, you know, if they walk into the
next quarter or the quarter after
that and are, you know, signal some
sort of, uh, um, peak spend, you
know, good for them, bad for the whole
infrastructure play, which would, you
know, we wouldn't be too, too mad at that
Luis Alvarez: What's the chart
on your screen right now?
Sean: Um, I'd say there's
probably, like, two.
Um, we're looking at obviously
real-time inflation 'cause, you
know, given that, you know, we, weâ¦
You know, there's still crosscurrents.
That's always on our screen 'cause we know
how, uh, sensitive that is to the market.
So that's hovering around
2%, so that's good.
Um, and then, you know, just tracking,
I'd say app usage across the ecosystem
around AI, just trying to see who's the
winners and losers, specifically around,
like, uh, app usage and token usage.
So, like, are we seeing models
continue to push towards open source?
Uh, and if they are, like, you know,
I think, uh, on one side you have
the inflation story, on the other
side you have the optimization
inside of AI and, like, you know.
That's not one chart, that's a
bunch, but I think that's how youâ¦
You have to have a mosaic to, to build,
you know, your narrative and story
Luis Alvarez: Last question.
Uh, what are you reading or listening
to right now that has nothing to
do with markets or investments?
Sean: Honestly, um, more,
uh, history than usual.
You know, every time I read
about prior cycles, railroads,
electrification, internet, all
those patterns tend to rhyme.
You know, the shape of the bubble
debate, the shape of the winners
taking time to emerge, the shape
of the durable compounders getting,
you know, looked at mid-cycle.
Um, you know, ultimately, those are
many of the things I'm reading about.
Some of them are directly or
indirectly tied to the market, so
it's not like market specific, but
it's, you know, industry specific.
Uh, so it's market adjacent.
Uh, and that when you, when you
read those, it, it, it allows you
to stay, you know, fairly patient
Luis Alvarez: Yeah, Sean, thanks
so much for walking us through it.
To everyone listening, thanks
for joining us Around the Desk.
If you enjoyed this discussion,
subscribe wherever you listen to
podcasts and follow Avory for more
data-driven investment research.
We will see you next time
Sean: Awesome.
Thanks, Liz
Luis Alvarez: Apa?
Sean: You're all