All this AI and productivity is shrinking?

GM, this is Milk Road Stocks, the newsletter giving you the best of the market in 5 minutes or less.

Yesterday, we sat down with Raoul Pal to talk about what the economy needs to continue growing over the coming decades. Today we’re breaking down his thesis.

First, a quick detour to private markets.

On Sept 23, Milk Road's Martin sits down with Alumni Ventures to talk about blockchain and private markets. Save your free seat here.

AMAZON IS A COUNTRY, AND WASHINGTON WANTS THE REST OF US TO COPY IT 🤖

We had Raoul Pal (CEO of Real Vision) on The Milk Road Show yesterday, and he spent a good chunk of it explaining where economic growth actually comes from.

His formula: GDP growth = population growth + productivity growth + debt growth.

I.e. More workers + better workers + borrowed money = economic growth.

But that formula is breaking down…

Baby boomers are leaving the workforce, and there aren't enough people behind them to fill the gaps they left.

U.S. labor force participation (the share of adults either working or looking for work) peaked at 67.1% in 2000 and currently sits at 61.6%. 👇

Source: BLS

Then there's the borrowing…

Gross national debt crossed $40T in August (yikes!).

Raoul pointed out that, since 2008, most new borrowing has gone toward refinancing the old borrowing rather than funding anything new.

So that lever is technically still there, but it’s kinda jammed.

Which leaves productivity…

Kevin Warsh took over as Fed chair in May, and one of the first things he did was stand up five task forces. One is dedicated to productivity and jobs, with a remit to survey the economic impact of AI specifically.

Then at Jackson Hole in August he called AI "structurally disinflationary" and said the potential for substantially higher growth is on the rise.

I.e. If productivity climbs, he gets to keep rates lower without inflation turning up to ruin it.

So what does that look like inside an actual company?

Raoul's example is Amazon.

They spent the back half of the 2010s pouring money into warehouse robotics - and for a long time it looked like nothing was happening. Profit per worker wobbled sideways, before falling off a cliff in 2022.

But then, all of a sudden: payoff hit.

Amazon's operating profit per employee went from roughly $7.9K in 2022 to about $50.8K last year, while headcount barely budged from where it sat in 2021.

Source: Amazon

The way Raoul put it: you should treat Amazon as a country, because the whole U.S. economy needs to follow its path.

(Fewer workers → more machines → better margins out the other side.)

Problem is, it hasn't shown up in the national numbers yet.

Official productivity growth across this business cycle (so, since late 2019) has run at 2.1% a year. Which beats the 1.5% of the 2007 to 2019 stretch, and lands exactly on the long-run average going back to 1987.

(All that AI compute, and the numbers look thoroughly ordinary.)

Source: BLS

What’s nuts is - productivity has been slowing this year. Annual productivity growth was 3.0% in 2024, 2.1% in 2025, and the first two quarters of this year came in at 0.3% and 1.4%.

Long story longer:

We all need to be keeping a close eye on the productivity number.

Warsh has already called AI disinflationary and argued the Fed can hold rates lower because of it - so if productivity starts to inflect, he's got the argument he needs to cut (which would lift risk assets along the way).

If it keeps coming in average, the gains will likely stay concentrated within a dozen or so companies rather than spreading out across the U.S. economy.

Btw, this was just a small snippet of what we covered in yesterday’s discussion with Raoul.

If you haven’t already watched the full episode, you can do so here!