In the financial markets, but not of the financial markets.

United States of America
Joined March 2025
Does this get funded by market participants selling other investments or by an expansion of the money supply?
Total investment in data centers and related AI infrastructure is projected to total $10.3 trillion from 2025 to 2032, or 3.6% of GDP per year, according to new estimates. That would dwarf other huge U.S. infrastructure projects such as the railroads, the highway system and the plumbing for the internet.
20
The same people who have been saying for 15 years that sovereign bond yields were too low are now acting like the world is ending because bond yields have reverted back to where they were before they were too low.
1
11
With the exception of Japan, to call these "record high" bond yields implies that recorded history in the bond market began <20 years ago.
1
1
2
429
Matt S retweeted
The Silvergate team are some of the best humans I know. They were failed by our leaders and systems, and yet they remain graceful, humble and positive about the future. The article from Alan Lane below is must read. Also, as fate would have it, I have the privilege of interviewing former COO/Chief Risk Officer @jaredkate today at @thestablecon, and former CTO @D_CentralBanker tomorrow at @PubKey DC. • Today, noon on The Node Stage w/ Kate Fraher • Tomorrow, 6PM at PubKey w/ Chris Lane – Free event, register here: luma.com/FightForClarity
Silvergate’s Voluntary Liquidation: Setting the Record Straight by Alan Lane silvergateceo.substack.com/p…
3
3
17
1,536
I know a guy who resigned from Anthropic today after spending three years working on model training who thinks everything is going to be alright
7
3
1
95
2,987
We are going to be so hardened on the other side of this moment (if we survive).
Our third-party e-mail provider has been breached. Please be aware that the email named ‘Critical Security Alert: STM32 Entropy Vulnerability’ is not coming from us, and it’s a phishing attempt. Do not click on any link. We have taken down the domain, and we are investigating the situation, including how the hackers got access to our legit domain.
1
17
He's right, you know.
I fucking hate these Bitcoin Spaces. It’s the same doom and gloom circle jerk every day: “The dollar dies within two years.” “Unemployment is exploding.” “We’re already in a recession.” “Government spending is out of control.” “Japan will blow up the carry trade.” Cool. Then sell your house, liquidate your portfolio, close your business and go live in a fucking bunker. But you won’t. Because most of you got rich using the exact system you now tell young people is pointless to participate in. You bought assets. Built businesses. Started families. Took risks. Benefited from decades of monetary expansion. Now that you’re comfortable, you sit and LARP on Spaces telling a generation with nothing that the entire economy is about to collapse and the world is gonna end. Then when it doesn’t, you quietly move the goalposts, predict the next apocalypse and take ZERO accountability for everyone you kept sidelined. Your words have consequences. Why should young people work hard, invest, get healthy, build families or believe in the future if you’re certain there isn’t one? Bitcoin should give people hope and a reason to build. Instead…you have turned it into a doomsday religion that monetizes fear. You’re not brave for predicting the end of the world every six months. You’re just a hypocrite with a microphone. GFY.
11
This summary is a great way to think about the Power Law. Credit to Noah Frank: A very high Bitcoin (R^2), like 0.96, simply means that the power-law curve has matched Bitcoin’s historical price extremely well, but it does not automatically mean that Bitcoin is literally controlled by a fundamental power law. A good way to think about it is this: imagine you look at a section of road that appears perfectly straight from where you are standing; that does not prove the road stays straight forever, because it might actually be part of a very large curve that only looks straight over the distance you can see. Bitcoin may be similar. When you plot its price against its age on a log-log chart, the data form an impressively straight line, which is why the power-law model looks so convincing, but other types of growth can also produce something that looks almost straight over a limited period. One warning sign is that the estimated power-law exponent, usually around 5.6 or 5.7, depends heavily on where you decide that “time begins.” Bitcoin researchers normally count time from the genesis block in January 2009, which makes sense because that is when Bitcoin actually came into existence, but if you mathematically shift that starting point, you can get a very different exponent while still keeping a surprisingly good fit. That means the high (R^2) alone cannot prove that 5.7 is some deep, universal Bitcoin constant; to make that stronger claim, you would need a convincing economic or network-based explanation for why Bitcoin’s creation date must be the exact starting point and why the exponent should specifically be around 5.7. There is also another possible explanation for Bitcoin’s history: instead of following one smooth power law forever, Bitcoin may have gone through several waves of adoption. Think of the way a new technology spreads: at first hardly anyone uses it, then adoption accelerates rapidly, and eventually that group becomes saturated and growth slows down. That produces an S-shaped, or sigmoid, curve. Bitcoin could have experienced several such waves. Early enthusiasts, then retail investors, then institutional investors, and perhaps future groups after that. If several S-shaped waves overlap, the combined result can look remarkably similar to a power law for many years. In other words, what looks like one smooth, permanent rule might actually be the combined effect of several separate growth phases. This matters because the two explanations have very different long-term implications: a true power law keeps rising indefinitely according to the same basic mathematical relationship, while an adoption-wave model eventually slows unless another major wave appears. Researchers have found that multi-sigmoid models can sometimes fit Bitcoin’s past history better than the simple power law, which suggests that Bitcoin’s cycles may contain meaningful structure that the power-law line smooths over. But fitting the past better does NOT necessarily mean predicting the future better. The safest conclusion is that Bitcoin HAS behaved in a very power-law-like way over its historical lifetime, but we still cannot be sure whether that pattern is a deep law of the system or simply the result of several growth and adoption waves that happen to create a very similar looking curve.
16
3
1
73
10,421
Macro wonks, give me your best reason why term premium in Japan can't revert such that BoJ has another 100bps hiking headroom with the long-end holding low/mid 4s?
17
Guys, it happened. I forgot how to do my job already. Been using Claude / Grok / Googledotai for the past few months. Recently spent days going deep with Claude on JPY-JGB-UST-USD market mechanics and risk asset smoke signals. Was so excited to tell Claude about today's move...
1
59
And it's not working! (Grok neither but all my context is w Claude). Anyway, the punchline is that I actually don't know what website to visit or thing to search for to help me make sense of today's move. What would I have done before Claude? Honestly sorta don't know. Wild.
28
Probably not a bad idea to allocate to int'l equities in commodity rich countries with stable governments. Double whammy of weaker dollar benefit plus commodity cycle.
Can something as obscure as a 30-year cycle in commodities really matter? Decide for yourself. But, um, so far, so good. @sentimentrader
13
Replying to @BitPaine
Counterpoint: Fed Funds 3.25-3.75; inflation 3-3.5; nominal GDP 5-5.5; 30Y 5.5-6 (10Y flat curve steepens) Run it out 3-5 years US Debt / GDP 140 JGB 10Y 3.25, 30Y 4.5; JPY/USD 145; DXY ~90; Gold $6K (+33%); S&P 9,250 (+20%). "Nothing stops this train" is a long, boring ride.
1
38
"i like broccoli now" no you don't, you like cauliflower because that's all we're gonna put in ads.
1
1
1
10
5,484
Wrong but right... about 15 years ago, I convinced myself that non-poisonous food would become unaffordable or unbuyable. You'd have to pay a ton for groceries or even grow / raise your own food. A more precise description is that you have to process your own food...
1
1
89
Reality is that it's hard to avoid processed food, especially if you have young kids and interact with other humans. You can pay up for organic / non-GMO / clean labels, but many of these are in some way toxic. Environmental contaminants (lead, inorganic arsenic), mold / fungus, PFAS packaging, etc. It seems the answer is that you have to research your personal supply chain, get as close to whole food as possible, and be your own processor. Example: Dave's Killer Bread is inferior to bread you bake at home with flour you imported from Italy, which is inferior to flour that you milled at home with wheat you bought from a U.S. farm that you have diligenced in person (bonus points if you can drive there). So, long story short I didn't have to move out of suburbia and buy big plots of land in the boonies. Thesis can be mostly correct but you have to figure out the right way to express it.
1
61
That's all for now. Gotta get back to figuring out how to express the Japanese carry trade unwind thesis.
5
I've got a pretty clear picture now of where we're headed next year, when Fable-class models become ubiquitous and cheap, and every enterprise is flooded with hundreds of new Fable-class AI employees. Spoiler: They create a constitutional legal system. yegge.ai/essays/fences-not-s…
52
46
41
612
98,305
Bloomberg: Advisers remained the largest holders of spot Ethereum ETFs at $782mm, while hedge funds fell to fourth after heaving selling in 2Q. Amazing! FAs spent > 10 years telling clients to avoid bitcoin. Now they're on board and telling their clients to diversify into eth!
2
77