I love the boogie men people will dig up when trying to explain why the market’s going down when in reality it’s pretty much as simple as “rates are going up”
Dougie
Verified account This account is verified. Learn more@DougieDeLuca
live free or die
1.3K Following 9.8K Followers 6.8K Posts
Latest posts 5
after several months at Paxos Labs working alongside Bhau, it’s obvious to me why a business would want to hold or integrate another stablecoin. he lays it out well here.
our focus has been what happens after that decision. integrating a new stablecoin is usually a heavier lift than a business expects going in.
whether you’re an enterprise, a payments platform, an exchange, a chain, or a market maker, you should be able to pick the stablecoins you want, offer them to your users, and hold them in your own treasury frictionlessly. that’s what we’ve been racing toward enabling with Transit.
I’ll be in London for DAS Nov 10-12. if you manage balances across multiple stablecoins or move between them often, I’d love to hear how you’re handling it today and trade notes on making it more efficient.
DMs open.
our focus has been what happens after that decision. integrating a new stablecoin is usually a heavier lift than a business expects going in.
whether you’re an enterprise, a payments platform, an exchange, a chain, or a market maker, you should be able to pick the stablecoins you want, offer them to your users, and hold them in your own treasury frictionlessly. that’s what we’ve been racing toward enabling with Transit.
I’ll be in London for DAS Nov 10-12. if you manage balances across multiple stablecoins or move between them often, I’d love to hear how you’re handling it today and trade notes on making it more efficient.
DMs open.
View quoted post on X
close your eyes and think about everything that's transpired in the market over the last week.
now open your eyes and see that a bunch of coins are up last 7d...
now open your eyes and see that a bunch of coins are up last 7d...
the blue wave is an early test of Saturday's thesis: when political or financial constraints limit new compute supply, owning existing capacity and controlling its allocation becomes a huge advantage.
hyperscalers have that leverage. today's divergence vs. semis/neoclouds is telling
hyperscalers have that leverage. today's divergence vs. semis/neoclouds is telling
after spending a good chunk of my Saturday morning thinking this over, here's where I'm at on AI:View quoted post on X
my enduring thesis, which I've shared a few times over the last several months, remains that AI can be enormously valuable without the companies developing frontier models necessarily being great businesses or investments at their current valuations.
what I'm increasingly focused on is the potential convergence of several pressures on independent frontier labs.
capital markets are becoming less tolerant of AI infrastructure spending, and we're starting to see financing stress among the independent compute providers these labs depend on. meanwhile, physical and political constraints are making it harder to expand compute supply.
now imagine hyperscalers face continued pressure to moderate spending and increasingly prioritize their own models and products. independent frontier labs could find themselves squeezed from multiple directions: constrained access to incremental compute, worsening financing conditions, and increasing competition from vertically integrated incumbents.
unlike those incumbents, they don't have massive, independently profitable businesses to subsidize their research. they need to continuously finance the next generation of intelligence just to maintain their competitive position.
and I think the competitive picture has changed meaningfully this year. incumbents have adapted faster than many expected, while increasingly capable open and near-frontier models are putting pressure on how much customers need to pay for intelligence.
the biggest counterargument (stating the obvious) is that the frontier itself becomes extraordinarily valuable. think biotech, scientific research, defense, autonomous work, and other industries where having the most capable intelligence could translate into an enormous competitive advantage. in those markets, customers might rationally spend extraordinary amounts to stay ahead of their own competitors.
I think that's a very real possibility, and we're seeing early signs of it. but we haven't yet demonstrated that this demand will materialize quickly enough, at sufficient scale, or with enough of the value accruing to frontier labs to justify the cost of continuously pushing the frontier forward.
so where does that leave me as an investor?
I'm increasingly inclined to favor vertically integrated incumbents and companies that benefit from widespread AI adoption. they have more control over their own destiny, more ways to monetize intelligence, and greater financial flexibility if the economics of frontier development disappoint.
but that doesn't necessarily mean I'm eager to buy them today.
the market is only beginning to seriously question the returns on this extraordinary infrastructure spending. these companies may be better positioned to survive the cycle while still being vulnerable to substantial valuation compression as investors reassess their future capital requirements and returns.
and importantly, they're not immune to the demand question either. even the hyperscalers need AI to generate sufficient economic value to justify their investment. they just have more ways to capture that value and more financial capacity to absorb being wrong about its timing.
so over the coming months, I'm watching whether financing and compute supply pressures converge, whether incumbents continue closing the frontier capability gap, and most importantly, whether genuinely high-value AI applications emerge quickly enough to support the industry's investment expectations.
I increasingly believe the technology will be transformative. I'm just less convinced that the capital markets have correctly priced who captures that value, what it costs to create it, and how long it takes to materialize.
and if the economics of maintaining the frontier become unsustainable while American technological leadership remains dependent on it, I still think the government eventually becomes the backstop.
that was part of my thesis months ago, and the developments since have made me more convinced of the underlying risk, even if the ultimate outcome remains uncertain.
palantir's earnings reaction encouraged me to think further about the AI thesis I shared over the weekend.
I'm increasingly interested in the divergence between companies monetizing the buildout of AI infrastructure and those positioned to monetize the economic value AI ultimately creates.
the market is beginning to seriously question whether the extraordinary capital investment in AI will generate sufficient returns to justify it. I think that's a legitimate concern, particularly given how much future demand and economic value is already priced into the infrastructure beneficiaries.
but I think there's a risk of conflating the economics of building AI infrastructure with the economics of actually adopting and commercializing AI.
a slowdown in infrastructure spending growth doesn't necessarily imply a slowdown in AI adoption. in fact, the transition from building capacity to extracting economic value from that capacity is a natural, and often healthy, part of technology adoption cycles.
and that's where I think the more interesting investment opportunity may emerge.
I'd increasingly favor businesses whose growth is driven by customers adopting products and services that deliver real economic value, rather than those whose growth depends on continued extraordinary capital investment.
what makes this particularly interesting is the asymmetry in what's already priced in.
much of the infrastructure trade depends on AI applications generating enough economic value to justify the capacity being built. meanwhile, there are businesses across enterprise software, biotech, defense, and other industries where AI could create substantial incremental value that hasn't necessarily been fully realized or capitalized into valuations.
if those applications succeed, i'd rather own the companies capturing that value than those already priced as though it's inevitable.
and if they don't succeed, it's hard to see how the extraordinary investment expectations supporting much of the infrastructure trade hold up either. that's what makes shorting select infrastructure beneficiaries an interesting potential hedge against a diversified basket of AI application beneficiaries.
of course, none of this means every application company is cheap or every infrastructure company is overvalued. palantir's earnings simply prompted me to revisit this framework. it's one of several names i'm watching, not necessarily the best expression of the thesis or an obvious bargain at today's price.
and I don't think the broader AI complex is finished repricing.
but over a medium-to-long-term horizon, I'm increasingly interested in gradually accumulating a basket of businesses positioned to capture the next cycle of AI spending as valuations become more attractive, rather than trying to find the bottom in those that benefited most from the last one
I'm increasingly interested in the divergence between companies monetizing the buildout of AI infrastructure and those positioned to monetize the economic value AI ultimately creates.
the market is beginning to seriously question whether the extraordinary capital investment in AI will generate sufficient returns to justify it. I think that's a legitimate concern, particularly given how much future demand and economic value is already priced into the infrastructure beneficiaries.
but I think there's a risk of conflating the economics of building AI infrastructure with the economics of actually adopting and commercializing AI.
a slowdown in infrastructure spending growth doesn't necessarily imply a slowdown in AI adoption. in fact, the transition from building capacity to extracting economic value from that capacity is a natural, and often healthy, part of technology adoption cycles.
and that's where I think the more interesting investment opportunity may emerge.
I'd increasingly favor businesses whose growth is driven by customers adopting products and services that deliver real economic value, rather than those whose growth depends on continued extraordinary capital investment.
what makes this particularly interesting is the asymmetry in what's already priced in.
much of the infrastructure trade depends on AI applications generating enough economic value to justify the capacity being built. meanwhile, there are businesses across enterprise software, biotech, defense, and other industries where AI could create substantial incremental value that hasn't necessarily been fully realized or capitalized into valuations.
if those applications succeed, i'd rather own the companies capturing that value than those already priced as though it's inevitable.
and if they don't succeed, it's hard to see how the extraordinary investment expectations supporting much of the infrastructure trade hold up either. that's what makes shorting select infrastructure beneficiaries an interesting potential hedge against a diversified basket of AI application beneficiaries.
of course, none of this means every application company is cheap or every infrastructure company is overvalued. palantir's earnings simply prompted me to revisit this framework. it's one of several names i'm watching, not necessarily the best expression of the thesis or an obvious bargain at today's price.
and I don't think the broader AI complex is finished repricing.
but over a medium-to-long-term horizon, I'm increasingly interested in gradually accumulating a basket of businesses positioned to capture the next cycle of AI spending as valuations become more attractive, rather than trying to find the bottom in those that benefited most from the last one