_{Brought to you by:}

"Time is the scarcity and it's the commodity we can’t create any more of."

— Jim Mitchell

The final arbitrage for investors is time

Warren Buffett attributes much of his success to reading.

"I just sit in my office and read all day," he once told an interviewer, with only a moderate amount of exaggeration.

Buffett reads as much as 1,000 pages of newspapers, books, 10-K filings and annual reports every day because "that's how knowledge builds up, like compound interest." 

"My job," he told the author Michael Eisner, "is essentially just corralling more and more and more facts and information, and occasionally seeing whether that leads to some action."

It’s worked, of course.

Buffett’s latest annual letter to shareholders shows that Berkshire Hathaway has outperformed the S&P 500 by an astounding 5,463%:

But it may not work much longer.

Buffett has outperformed everyone because he can read more 10-Ks than seemingly everyone else put together (I’ve tried, they’re very boring).

Over decades, he’s "corralled" more facts than anyone else. 

But AIs (which never get bored) can read more 10-Ks than Warren Buffett and corral all the information. 

It’s a long way from when gathering information from even one 10-K was an informational advantage: Prior to the advent of Edgar in 1994, you had to travel to an SEC office in person to read an SEC filing. 

Now they’re available online for anyone to read, including AIs.

1994 also happens to be the year I first started trading equities. 

Back then, you only had to be a little quicker and a little harder working than your human competitors to make money by trading. 

Markets were still inefficient enough that you could do something as simple as buy shares of the French-Italian semiconductor manufacturer STMicroelectronics in Paris and immediately sell them a couple of cents higher in Milan. 

These arbitrage opportunities became increasingly rare in the early 2000s after quants introduced algorithmic trading to the equities market.

With fewer market inefficiencies to exploit, trading became a job of picking stocks and predicting the direction of the market — which is basically impossible for anyone other than Warren Buffett.

This is partly why I switched to crypto a few years ago: It remains a quirky market where a quick-thinking, hard-working human can still find some inefficiencies to exploit.

(Alas, quicker and harder-working than I am — I don’t have the patience to hang out in Discord rooms and Telegram chats all day, farm the newest protocols or post on X for valuable "yap" points.)

But that, too, is changing: Citadel is coming to do to crypto what the quants did to equities in the 1990s. 

At first, I imagine it’ll just be tightening bid/offer spreads and looking for arbitrages between exchanges.

But how long will it be until the likes of Citadel deploys AI to scrape the human-generated alpha from Discord, Telegram and social media and use that to predict crypto prices?

I expect that will eventually make the crypto market perfectly efficient (in the sense of being perfectly unpredictable).

Or, at least, too perfectly efficient for humans.

AIs, however, will still be able to find some informational arbitrages to exploit, as a new study suggests: "Advanced machine learning techniques, including deep learning, reinforcement learning, random forest and generative AI," the authors predict, "[will revolutionize] research, automation and prediction in the stock market."

No human investor will be able to compete.

Worse yet, AIs might even collude against human investors — perhaps by manipulating prices, exploiting predictable human patterns or spreading disinformation.

Notably, this behavior may be emergent: It will happen without the AIs being instructed to collude or even having to communicate with one another.

This kind of robot takeover won’t make for as good of a movie as Terminator, but it should probably be just as terrifying for investors.

One AI researcher, NYU’s Caleb Maresca, predicts that investors will soon sense that this Terminator moment is coming — and that there will be a mad scramble for investment assets when they do. 

Investors, he believes, will recognize that after the advent of "Transformative AI" (TAI), nearly all labor will be performed by AIs, so the only way for humans to earn money will be to own a piece of that AI labor. 

If so, however much time there is between now and TAI is how much time you have to secure your share: "Future control over AI labor could depend on wealth at the time of AI deployment."

Maresca predicts that a race to accumulate assets before TAI sets everything in stone will drive one-year interest rates to 10-16%.

10 to 16%!!

It’s difficult to imagine how investors (not inflation!) could drive interest rates to that level.

10-16% implies a level of desperation akin to Walmart shoppers on Black Friday — but instead of a race to secure a flat-screen TV, it’s a race for economic survival.

That desperation may be warranted. After TAI, both the economy and markets will be so dominated by algorithms that the only way to survive will be to own a big enough claim on "AI-mediated production" (paid for with your pre-TAI earnings and investment returns).

"The wages previously earned by human workers," Maresca warns, "will flow to whoever controls the AI system performing that job."

Investors, too, will be replaced post-TAI — there will be no hope of increasing your share of AI labor by outperforming markets, because markets will be perfectly efficient.

This is all a bit theoretical and I’m hopeful that it never quite comes to that.

But, if nothing else, I take it as a reminder that the only real arbitrage for investors is time.

Warren Buffett acknowledges that this explains much of his success: "My life has been a product of compound interest."

That, at least, is unlikely to change.

As markets become perfectly efficient, time will increasingly become the only thing that separates one investor from another.

— Byron Gilliam

Morpho is the first and only DeFi protocol integrated by Coinbase. As the go-to infrastructure for onchain loans, Morpho’s immutable code sets the gold standard for decentralization, giving builders full ownership and control.

Leverage Morpho’s flexibility to create lending and borrowing solutions tailored precisely to your users’ needs.

1. Pump.fun said to be ditching Raydium, launching own AMM

2. Issuers pile more crypto ETF plans

3. The price of bitcoin is at a ‘critical juncture’ 

4. Traders pray for Nvidia boost as stocks dip red

5. Wildcard moves NFTs from Polygon to Ethereum, shares token strategy

Futarchy Deep Dive: Can Markets Make Better Decisions?

MetaDAO’s Proph3t joins the show to discuss the line between prediction markets and futarchy, historical criticisms of this mode of governance and where it intersects with crypto. 

Listen to Bell Curve on Spotify, Apple Podcasts or YouTube.

With capital rotating, regulation evolving, and liquidity shifting, the smartest players are already adjusting.

• Paul Brody (EY) on where enterprises are deploying blockchain beyond the headlines.

• Ambre Soubiran (Kaiko) on the market trends that funds are acting on before retail catches up.

• Jake Chervinsky (Variant) on the legal shifts that could create—or kill—new opportunities.

• Rob Hadick (Dragonfly) on the allocation strategies driving institutional plays this year.

Some will watch. Others will act. Which side are you on?

📅 March 18-20 | NYC