This is a very comprehensive mapping of AI financing structure and points to some really important dynamics. I’d pay attention to the GPU depreciations and especially the cash flow uncertainty @advaitarun.bsky.social points to!

For Caravanserai Volume 1 Issue 6, we went farther afield—it's geopolitics week. Here are three India-China geopolitics analysts commenting on the risks and opportunities India faces from taking Chinese investment: caravanseraiforum.substack.com/p/india-chin...

this is a great way to put it—they’re framing the *easy choices* as hard choices

made a similar point on the opinions podcast this week in response to the idea that epstein is a dead end. not only are actual voters interested in this but it divides the gop and puts pressure on the president. it would be political malpractice to treat it as a sideshow

Punchy work from Gadfly's analysis of investment trends.

Really fun piece from Humraaz about how security concerns still dominate the relationship between Indian and Chinese leaders, despite photoshoots to the contrary.

George Eliot has a more pessimistic take about what happens if Chinese investment in India dries up due to broader macro trends.

George Eliot has a more pessimistic take about what happens if Chinese investment in India dries up due to broader macro trends.

Really fun piece from Humraaz about how security concerns still dominate the relationship between Indian and Chinese leaders, despite photoshoots to the contrary.

Punchy work from Gadfly's analysis of investment trends.

For Caravanserai Volume 1 Issue 6, we went farther afield—it's geopolitics week. Here are three India-China geopolitics analysts commenting on the risks and opportunities India faces from taking Chinese investment: caravanseraiforum.substack.com/p/india-chin...

Hold up

Marxists await final crisis of capitalism as Depreciation Specialist MACRSism intersects with Private Equity Marksism.

"10a labs", a 'security research' firm employed by the big AI companies has begun collecting information and attacking organised opposition to data centres, framing them as "NIMBYS" and wildly exaggerating claims. www.datacenterdynamics.com/en/news/grou...

newsletter called MACRS Mark when

‘If there is a sudden stop in new lending to data centers, Ponzi finance units “with cash flow shortfalls will be forced to try to make position by selling out position”—in other words to force a fire sale—which is “likely to lead to a collapse of asset values.”‘ publicenterprise.org/report/bubbl...

being in the Society of Depreciation Professionals sounds so fun. you can just point at assets and say shit like "it's over," "bro's washed," "you're cooked" and make some menacing scribbles on a clipboard www.cnbc.com/2025/11/14/a...

holy moly our Hopkins lab data @nzpolicylab.bsky.social is in the IEA's energy bible!! the cool kids are actually reading the IEA World energy outlook 2025 till page 76 and not having heads turned by the silly trumpist pressure CPS scenario getting all the headlines.... www.iea.org/reports/worl...

Getting a @rajakorman.bsky.social pun is a bit better than being on the odd lots newsletter, I don't make the rules

Huh. Bearish AI research from CPE leads to Arun in the market.

Me and @jmooreotto.bsky.social wrote a bunch about green bank financial strategies! Lots here for state policymakers:

Launched today: the State Economic Power Project! This dives into cost of living + climate, on 2 pillars: (1) household budgets and (2) state budgets. Thank you Sarah Bloom Raskin, @jordanhaedtler.bsky.social @kostyack.bsky.social @advitt.bsky.social climatecabineteducation.org/state-econom...

it’s worth crediting Joe Biden — yes, that one, the senile one with dementia — how he didn’t succumb to this dynamic when it was clearly Republicans’ plan during a debt-limit crisis.

Handy cap structure analysis here from outside the industry. Cc: @roqueig.bsky.social @ruoshuiresearch.bsky.social @rajakorman.bsky.social

Really thrilled to be featured in today's Odd Lots newsletter--thanks @weisenthal.bsky.social and @tracyalloway.bsky.social! The implication: The U.S.'s sole growth sector is stuck between a prisoner's dilemma in pricing and a collateral crunch. bsky.app/profile/adva...

Today's new @publicenterprise.bsky.social report is a comprehensive analysis of the capital structure of the entire AI sector: data center real estate, GPU markets, private credit, you name it. It's also a financial risk management framework for policymakers! publicenterprise.org/report/bubbl...

And here I was, thinking I could finally plow through The Prize without distraction. Alas.

The precious moment when you find an economics paper and it feels like an economist has spent time doing exactly what you would hope for them to be doing.

Really thrilled to be featured in today's Odd Lots newsletter--thanks @weisenthal.bsky.social and @tracyalloway.bsky.social! The implication: The U.S.'s sole growth sector is stuck between a prisoner's dilemma in pricing and a collateral crunch. bsky.app/profile/adva...

One reason I decided to emphasize T-accounts is bc there are a decent few analysts and historians who have quoted Minsky's Financial Instability Hypothesis these past few months who really should try and understand what Minsky's whole project is before they enlist him as their prophet of doom.

T-acounts, they make things look so orderly. Very helpful AI bubble mapping here by @advaitarun.bsky.social.

There are FIVE BIG RISKS. Cash, collateral, tenant churn, circular financing, and debt. You've probably heard of some of them. But here's the analysis of how they connect.

Here's our analysis of what a federal loan guarantee or a tax credit for the AI sector could involve, in practice. Counterintuitively, a GPU loan guarantee is a boon to NVIDIA (and a tax credit might not be so easy to monetize...)

This is the key!!!

"The booming demand for high-end chips and the meteoric rise in NVIDIA sales look less like a reflection of healthy growth for the tech sector and more like a scramble for high-value collateral and a strategy to maintain market position among a set of firms with limited product differentiation."

Shoutout to @quantian.bsky.social for one of the best threads I found on this whole AI bubble shebang. Featured heavily in today's CPE data center report as a keystone point of our analysis:

Economist Hyman Minsky explained how speculative debt issuance can endanger an entire economy. A new @publicenterprise.bsky.social report applies this to debt finance for data centers: how the AI bubble could burst & how to think about the potential bailout. publicenterprise.org/report/bubbl...

really clear and thorough picture of the mechanics of AI bubble financing and what it might look like when it blows up https://publicenterprise.org/report/bubble-or-nothing/

Great thoughts about what's happening inside Mag7 companies as they pivot into AI.

something that I don't think the paper mentions explicitly: for the mag7 companies, because they use so much stock-based employee compensation, a drop in their share price directly lowers their cash flow because they have to pay employees more cash to make up for the lower stock price (or dilute […]

Want to understand the guts of the AI investment boom? CPE's @advaitarun.bsky.social wrote a fantastic report for all your Qs: describes who the actors are, how they finance capex, and just how they are taking on risk. Lots of lessons for policymakers too! publicenterprise.org/report/bubbl...

"Data centers are an asset with the characteristics of both real estate and infrastructure: Data centers have tenants, large tech companies, that are undertaking expensive long-term capital investment plays with fast-depreciating assets and minimal cash flow..." publicenterprise.org/report/bubbl...

Read it and weep. Corporate grift technological indifference to humanity.

token weight of training vs inference is going to make the dynamics between subscalers and hyperscalers host pricing fun

Just five little risks, Advait. What could they cost, 5 trillion dollars?

Here's our analysis of what a federal loan guarantee or a tax credit for the AI sector could involve, in practice. Counterintuitively, a GPU loan guarantee is a boon to NVIDIA (and a tax credit might not be so easy to monetize...)

There are FIVE BIG RISKS. Cash, collateral, tenant churn, circular financing, and debt. You've probably heard of some of them. But here's the analysis of how they connect.

Today's new @publicenterprise.bsky.social report is a comprehensive analysis of the capital structure of the entire AI sector: data center real estate, GPU markets, private credit, you name it. It's also a financial risk management framework for policymakers! publicenterprise.org/report/bubbl...

it's really funny how Texas and Florida are competing in a race to adopt the most idiotic property tax system known to man

My favorite part of writing this report—but also the most worrying part—was explaining why we do NOT believe the LPO can currently finance new nuclear reactors. We looked into Vogtle's project finance and argue it was a "financial anomaly" that will be impossible to replicate: