AI Trade — the handoff from building AI to using it, in one race
Four Closelooknet indices, one for each stage of the AI trade, each started at 100 on the
first trading day of 2026. Rubin, the build-out, ran first: to
241 by 22 June. The Agentic Ecosystem,
the companies paid each time AI runs, kept climbing to 181 on
9 October. The Agentic Winners bottomed at 71
on 25 June, 3 days after Rubin’s high. And since July the build-out has given back while the
AI Users moved up to a high for the year.
The AI Handoff — from building AI to using it, 2026
$100 in each stage of the AI trade on the first trading day of 2026, equal-weight: the build-out, the companies paid each time AI runs, the applications and the users.
0%
What you are watching
Each bar is one of our functional indices, equal-weight, so every
company counts the same. Together they follow the money through the AI trade, from the
companies that build the machines to the companies that put the machines to work:
Rubin Build-Out — the capital spending: chips, memory, packaging, power and the data-centre build. Rubin 100 →
Agentic Ecosystem — the operating spending: the companies paid each time an AI agent runs, from data and identity to security and compute. Agentic Ecosystem →
Agentic Winners 40 — the software applications built around AI agents. AW40 →
AI Users — companies outside tech putting AI to work in their own business. AI Users →
The bar in front always paints brand red. Apart from one day in late October, when the Ecosystem passed Rubin, the order has held since February; the story is in the gaps.
Three phases of 2026
The build-out first. From January to 22 June, Rubin rose from 100 to
241. Chips, memory and power led the AI trade, while the software
applications fell: Agentic Winners were 29.2% down on the year by
25 June, and the AI Users touched 83 on 27 March.
Then the operating spending. The Agentic Ecosystem bottomed early, at
91 on 5 February, and kept rising after Rubin had
peaked, to 181 on 9 October. Once the machines are in place, the money
moves to the companies paid each time they run. The Winners made a first low in the spring,
72 on 10 April, and their lowest on 25 June, as Rubin turned down.
Since the summer, the users. From the first trading day of July to
13 November: Rubin −29.7%, the Agentic Ecosystem
−1.6%, Agentic Winners +21.6%,
AI Users +11.5%. AI Users closed at 102, near
its high for the year.
Why this matters
Rubin still leads the year, at 156 against 102 for the users. The race is
not about who is in front but about the order in which the stages rise and fall. Their highs
came in sequence: the build-out on 22 June, the operating spending on
9 October, the users on 4 November. So far only the first two
have fallen hard from their highs: Rubin −35.1%, the Agentic Ecosystem
−19.6%. The Winners are −9.1% from their January
high, and the users sit near theirs. That sequence is the handoff this diary
has followed since the summer, and the books lean with it: a little lighter in the
build-out, more in the companies that put AI to work. The
Weekly Signal of 15 November sets out how this diary
positions for it.
Embed this chart
Paste the snippet below into any HTML, Substack post or LinkedIn article. The embed page
tracks back to closelook.net and resizes itself in the parent iframe.
<iframe src="https://closelook.net/embed/race/ai-handoff/" width="100%" height="560" frameborder="0" loading="lazy" title="Closelooknet AI Handoff Race"></iframe>
Methodology
Source: the equal-weight series from the Rubin, Agentic Ecosystem, Agentic Winners 40 and AI Users index workers, the same numbers behind each index page.
Value: each index rebased to 100 at its close on 2026-01-02, the first trading day of the year and the base of every Closelook index’s year-to-date figure, so the last frame matches the YTD on each index page.
Time window: 2026-01-02 → 2026-11-13, 225 snapshots, one per trading day.
Refresh: nightly, Tuesday to Saturday, after the index workers have run.
Closelooknet indices are research constructs published as an investment diary, not advice
or a recommendation. Rebased index levels show relative performance only.
FAQ · from the current data
Bar Races — data Q&A
What is the Bar Races?
Animated bar races over the house data — AI megacaps, index leadership, sector rotation and Money Temperature — the same numbers as moving pictures. Part of the Closelooknet Lab (Visualisations).
What does the Bar Races cover?
Animated bar races over the house data — megacaps, index leadership, sector rotation, temperature. 4 races · Animated. Sub-surfaces: AI megacaps, Index leadership, Sector rotation, Temperature.
How often does it update?
Re-baked after the US close on trading days from the Closelooknet lake, the index workers and vendor series; the page states its own as-of date.
Is this investment advice?
No. Every Lab tool is a data surface in an investment research diary — it describes what the engine measures, never what to do.