# Ends Up Here media layer

Every episode page now has three stable, clickable media anchors:

1. **Reel** — a 30–60 second vertical video for discovery and later Google/Facebook/Instagram/YouTube distribution.
2. **Brief audio** — a concise 1–3 minute spoken introduction that makes the question clear.
3. **Full episode** — approximately 19 minutes, recorded as a natural conversation using a run sheet rather than a forced word-for-word script.

The episode page remains the source of truth. When a media asset exists, add its player and distribution links to the matching detail block. Do not create a separate orphan page for each format.

## First production model: What Happens to All That Compute?

### Reel — 45 seconds

**Hook:** We keep hearing that better AI means more AI use, more inference, more chips, more data centres and more electricity.

**Turn:** But AI is also learning to become dramatically more efficient. If the same useful intelligence takes 10, 100 or 1,000 times less compute, today's tasks may need much less infrastructure.

**Complication:** That does not mean compute disappears. It may move into agents, robots, infrastructure, science, health, governance and space.

**Close:** Compute does not disappear. Its purpose migrates. The question is not only how many data centres we build, but who consumes the compute and what it is for.

### Brief audio — about 2 minutes

Open with the three curves: AI efficiency falls, some conventional human work-compute changes, and civilisational demand may expand. Explain that the page uses indexed scenarios rather than predictions. End with the investment paradox: companies may be valued on the creation of abundant intelligence even though abundance attacks the scarcity supporting those valuations.

### Full episode — 19-minute run sheet

1. **Opening image (2 min):** a data centre, a person at a desk, and the assumption that every better model means more electricity forever.
2. **Curve A (3 min):** algorithms, architectures, hardware, compression and allocation make useful intelligence more efficient. Explain 10×, 100×, 1,000× and 10,000× as scenarios, not forecasts.
3. **Curve B (3 min):** if employment contracts, some document, spreadsheet, email and administrative compute may cease to be necessary. Humans with more free time may also spend less time at screens.
4. **Curve C (4 min):** agents, robots, grids, transport, agriculture, manufacturing, science, health, governance, safety and world models create new compute demand.
5. **The investment paradox (3 min):** scarce intelligence creates margins; successful intelligence may become abundant and competitive. Infrastructure can still be valuable, but scarcity and ownership migrate.
6. **The human question (2 min):** what becomes scarce—land, attention, relationships, experiences, energy allocation, major infrastructure and space?
7. **Living trajectory (1 min):** keep updating the argument weekly as efficiency, costs, chips, construction, cancellations, energy, agents, robotics and employment evidence changes.
8. **Close (1 min):** compute does not disappear; its purpose migrates. Ask who pays, who owns the systems and where the value ends up.

This is a conversational recording guide, not a script. Brianna or another host can interrupt, ask for examples and bring the argument back to ordinary life.