Debts, printing, inflation, risky assets, AI, crypto & storage

Published in

Vinter

3 min readOct 18, 2023

Here is a brief memo on debts, printing, inflation, risky assets, AI, crypto & storage — written so that someone without a degree or significant interest in macroeconomics gets it.

This blog post contains an investment thesis on crypto based on the following reasoning: 1. Debt => 2. Money printing => 3. Inflation & risky assets up => 4. Tech booms, including AI => 5. Invest in data & computation. It’s a summary of a post by Arthur Hayes.

1: DEBT

Government debts are high. A lot of debt must be refinanced soon.

Who will buy the debt? Mostly central banks (Hayes estimates half all gov debt will be purchased by central banks). Why? Because politicians can force them. Private capital is only forced to invest in war time. “The private sector doesn’t want them because, due to inflation, they have to hoard capital and purchase increasingly expensive inputs if they are a business or food/fuel if they are individuals.” Also, the private sector bought some low-yielding debt during COVID and then yield went up, so these portfolios have negative PnL now — private investors will say please no more of these bonds my PnL is red.

In short, large amounts of debt will be refinanced with bought by central banks.

2: PRINTING

Central banks buy this debt by printing money and handing governments this money.

Governments keep on spending because that’s what the politicians promised to get elected. And unless this spending creates productivity in the future, we just made the problem worse because this spending will need to be financed by even more future borrowing.

3: INFLATION & RISKY ASSETS

The money printing causes inflation.

Money printing also drives up the price of energy and risky assets (more dollars chasing a particular asset drives up its prices due to supply-demand).

4: TECH BOOMS

The riskiest assets are those with huge profits far into the future –technology.

Hayes writes, “Every fiat liquidity bubble has a new form of technology that captivates investors and draws a fuck-ton of capital. As the major central banks printed money to ‘solve’ the 2008 GFC, the free money flowed to Web 2.0 advertising, social media, and sharing economy startups. In the 1920’s, it was the commercialisation of technologies developed during WWI, like radios. This time around, I believe it will all be technology related to AI.”

5: AI

Hayes reasoning around AI goes like this

<1% of all AI companies will have a useful and profitable product, as it’s a winner-takes-all game. Investing in AI stocks is like picking a needle in a haystack. It’s better to invest in the infrastructure layer of AI.*
Having decided to play an infrastructure bet, what must these AI firms buy? Computational power and data storage.
When an AI selects storage, it prefers decentralized storage over centralized storage. The reason is that humans can turn off AWS but not a public blockchain, and an AI would prefer not to depend on humans (just like humans prefer not to depend on pigs for our existence).
One crypto token offering decentralized storage is Filecoin. The Filecoin network consist of many diverse storage providers, to create a secure and reliable services.

*Footnote: “This concept is very similar to how, during the Web 2.0 buildout, startups raised money and handed it to Google and Facebook in the form of advertising dollars. I challenge a reader to compile the average performance of the 2010 to 2015 vintage of VC funds and compare their compound IRR net of fees vs. buying an equally weighted basket of Google and Facebook stocks. I bet only the top 5% of VC funds (or less) beat that simple stock portfolio.”

Before talking up Filecoin, Hayes noted there is no crypto asset with convincing decentralized computational power. Maybe there will be in the future. Looking at the stock market, he thinks NVIDIA is too expensive at 101x P/E when its peers trade at 14x. Professor Damodaran agrees. Even under optimistic assumptions for NVIDIA, its price is less than its value, says Damodaran in his excellent analysis valuing the stock and since price<value he sold.