The Last Investment

The Singularity Tracker

The tracker stands at — out of 100, — from — a year earlier. At 100, all six indicators in the Last Investment whitepaper would sit at their singularity readings.

— / 100

The average of the six indicator scores, equally weighted.

The tracker by year

2010 onward, when the real-rate series begins. Before 2024 the two AI indicators count as 0: no AI acceleration or AI price series had been measured.

  • 0–24 Not yet
  • 25–49 Moving toward it
  • 50–100 Strong

How close each indicator is now

Each bar runs from the indicator's historical normal (0) to its own singularity reading (100).

The six indicators over time

US labor productivity growth

— / 100

Now — —

0
—
100
—

The whitepaper's singularity reading: Sustained growth well above historical rates. Source: —

Labor share of US business output

— / 100

Now — —

0
—
100
—

The whitepaper's singularity reading: A steep, continuing fall. Source: —

30-year real interest rate

— / 100

Now — —

0
—
100
—

The whitepaper's singularity reading: A sharp, sustained rise. Source: —

Pace of AI research

— / 100

Now — —

0
—
100
—

The whitepaper's singularity reading: Accelerating gains driven by AI itself. Source: —

Price of AI performance

— / 100

Now — cheaper a year —

0
—
100
—

The whitepaper's singularity reading: Cognitive work priced near the cost of electricity. Source: —

Share of stocks held by top 10%

— / 100

Now — —

0
—
100
—

The whitepaper's singularity reading: Rising further, then pressure to redistribute. Source: —

How the tracker is scored

Each indicator in the Last Investment whitepaper gets a score from 0 to 100. Zero is its historical normal, before AI mattered. One hundred is the whitepaper's "what a singularity would look like," turned into a number. Readings past either end are capped at 0 or 100.

  • Straight-line scale for productivity, labor share, the real rate and stock ownership: the score is how far the reading has moved from 0 toward 100.
  • Ratio scale for the two AI indicators, which move by multiples: going from 1× to about 3.2× scores 50 on a 1×-to-10× scale.
  • Productivity uses three-year average growth, because the whitepaper asks for sustained growth and single years jump around (2020's spike came from the pandemic, not AI).
  • The real-rate target follows the Ramsey rule used by Chow, Halperin and Mazlish: with standard settings, real rates rise about one-for-one with growth, so the target is the 2010s average plus the same rise as the productivity target.
  • The tracker is the plain average of the six scores, so every indicator counts the same.

Where the data is thin. The pace of AI research is one lab's measurement (METR's 95% range is 1.5–2.8×). The price of AI performance is one Epoch AI estimate from March 2025, carried into 2026 until a newer one is published. Across tasks its rate runs from 9× to 900× a year, which would score from about 29 to 98; the chart shows that range. Both AI indicators count as 0 before 2024.

The 0 and 100 points are editorial choices, not standards. Each card shows them, and changing them changes the score.

Get the Tracker every Monday.

The Last Investment: the day's AI news for investors, in 47 seconds. Free.