T·01
Artificial Intelligence
Intelligence becomes scalable
The spending is real and the adoption is measurable. Whether shareholders capture the value is the question I take most seriously.
Read the Artificial Intelligence thesisTony Mai · Six theses · T·01–T·06
Six long-term theses on where the world is heading, and what each one means for how I invest.
These pages are how I think about the technologies I have put my own money behind. They are personal investment theses, not stock recommendations and not research reports. Each one sets out what I believe, why I believe it, the evidence as it stands, the strongest argument against my view, and what would change my mind.
Every page is built around the same chain:
A technology succeeds.
Economic value is created.
A company captures some of that value.
Shareholders may or may not capture attractive returns.
How much of my capital does that uncertainty deserve?
Each link has to hold before an investment works, and none of them guarantees the next. A technology can change the world and still reward its investors poorly. Most of the thinking on these pages happens in that gap between importance and returns.
Why these six? Because they are the areas where I have concluded the structural change is real, the horizon is long, and I can hold a view worth writing down and defending in public. They are also connected: intelligence, the machines it will drive, the compute underneath both, who owns that compute, where infrastructure goes next, and what the same pattern looks like in medicine.
Two honest warnings before you read. The evidence changes. Numbers get refreshed as new quarters arrive, and a statistic that supports an argument today can weaken it tomorrow. And the companies change. A thesis is a view about where value accumulates, not a permanent attachment to a ticker, so the positions that express each view can move while the thesis holds. The portfolio behind these pages is public, and it moves.
01 — The collection
In order. Each page stands on its own; together they form a progression.
T·01
Intelligence becomes scalable
The spending is real and the adoption is measurable. Whether shareholders capture the value is the question I take most seriously.
Read the Artificial Intelligence thesisT·02
Intelligence becomes physical
A cheap clumsy robot is still a bad robot. Capability decides when this thesis becomes real, so I separate what is deployed from what is demonstrated.
Read the Robotics thesisT·03
Compute becomes infrastructure
If compute keeps getting cheaper, does the market shrink or grow? I don't have a settled answer, so the page follows the evidence on both sides.
Read the Semiconductors thesisT·04
Access expands, ownership concentrates
Three companies still hold most of a market that doubled. The variable I actually watch is how long AI hardware stays economically useful.
Read the Cloud Computing thesisT·05
Space becomes the next infrastructure frontier
The most speculative page on this site, and it says so. Whether orbital infrastructure deserves capital is a different question from whether it can work.
Read the Space thesisT·06
Medicine moves closer to the cause
The science works and the returns have not followed. My own holding is the strongest argument against my own thesis.
Read the Genomics thesis02 — The progression
Artificial intelligence asks whether machine intelligence becomes an economic force, and who captures the value if it does. Robotics follows that intelligence out of the screen and into physical work. Both run on the compute that the semiconductor thesis treats as infrastructure, and most of that compute is rented from the handful of companies the cloud thesis examines. Space asks whether the same infrastructure logic eventually extends beyond Earth.
Genomics stands slightly apart: the same pattern of a powerful technology and uncertain investor returns, applied to medicine. The order is a progression, not a dependency. None of these pages requires the others to be right.
03 — Maintenance
A thesis should change when the evidence does.
I review all six pages twice a year, in February and August. Evidence gets refreshed on that cycle: figures are checked against their sources, and numbers that have moved are updated. The arguments themselves change only when my view genuinely changes, and when that happens the change is made openly rather than silently edited.
The portfolio that sits behind these pages is public, so you can check what I actually own against what I say I believe.
First published
August 2026
Last reviewed
August 2026
Review cadence
February and August
Investing involves risk. Copy trading does not guarantee returns, and past performance is not a reliable indicator of future results. Your capital is at risk.