CLK 1 · INTELLIVE
2026
On time
Software, agents, digital work. Already compounding. This is the clock people are watching.
Hyperabundance · Elon · McKinsey · the household budget
Not
twelve
months.
If Elon and the McKinsey Global Institute are right that we are entering abundance, the first time a typical person would feel cheaper goods in a meaningful way is five to ten years. “Anyone can have anything” is a decade-plus claim even on Elon’s own calendar.
CLK 2 · WATTSLAG
2029
Demand ahead of supply
Data centers bid for scarce electricity now. New nuclear is late-2020s to 2030s. Near term this clock pushes prices up.
CLK 3 · HANDSLAG
2031
Thousands, not millions
Humanoids are factory tools in the low thousands. Consumer sales, if they happen, are a 2027-plus story. Not a CPI event yet.
Abundance is not “AI is smart, therefore prices fall.” Three things have to get cheap at the same time: intelligence, energy, and physical labor. Elon’s deflation math only fires when clocks two and three catch clock one. Right now they have not.
Into 2027
You will not see broad prices go down.
Rent, groceries, healthcare, and the electric bill will not get cheaper because of abundance. What you will see, and already do: much cheaper intelligence — writing, code, images, tutoring, legal first drafts, customer support, personal software. Quality of life rises for people who actually use that. It does not show up as a lower CPI. For a lot of households it shows up as a higher power bill.
Into 2029
First visible cheapness in sectors. Not in living costs.
Software, media, some professional services, maybe more manufactured goods if factory robots are doing real work. Businesses that cut labor costs and compete will lower prices. Businesses that don’t will keep the margin. Housing will not have fallen. Healthcare will not have. Electricity is still more likely tight than abundant. This is “my laptop and some of my errands got better,” not “the cost of living broke.”
Into 2031
Earliest honest window for cheaper stuff.
A median household might notice cheaper appliances, clothes, maybe cars if autonomy plus robots compress manufacturing and driving cost. It is still not Elon’s “quasi-infinite economy.” Doubling real GDP in five years is about 15% annual real growth. Large rich economies do 2–3%. China at peak did about 10%. Treat 2×-in-5-years as a stretch goal, not a base case. Elon himself put “money stops mattering” around ten years, not five.
Mid-2030s
This is when it becomes a household-budget story.
If the thesis is right: robots in the millions, nuclear / SMR and gas-plus-solar actually on the grid, intelligence near-free. That is also when the political problem gets sharp. If labor income falls faster than prices, you get abundance on the factory floor and tightness in the checking account until some form of “just issue money” exists. Elon has been explicit that last part is required.
Two different forecasts
Elon and McKinsey are not saying the same thing. Mix them and the timeline gets fake-precise. Elon is describing a phase change. McKinsey is describing a slow S-curve. History usually looks more like McKinsey until it suddenly doesn’t.
Elon · 2025–2026
Phase change
- AI plus humanoids produce so much that output outruns the money supply → deflation, then universal high income.
- Economy could double in five or six years.
- 100 million to a billion humanoids in about five years.
- Money starts to matter less around 2036.
- June 2026: “Deflation is inevitable.” Tesla mission language is now “Amazing Abundance.”
McKinsey Global Institute
Slow S-curve
- Existing agents and robots could technically do work that fills about 57% of U.S. work hours.
- Midpoint for 2030: only ~27% of hours actually automated, unlocking about $2.9 trillion in U.S. value — if firms redesign workflows.
- Plateau adoption of a new technology: typically one to three decades.
- 2025 AI survey: 39% of firms report any EBIT impact; most of those say it is under 5%.
What prices are doing this month
U.S. CPI, 12 months to July 2026. The split Elon is betting on is already starting in goods, and not starting in the things that dominate a household: housing, energy, healthcare, local services. That is the whole argument in one table.
BLS CPI-U · Jul 2025–Jul 2026
Headline +3.4%
| Line item | Share of the argument | 12-month |
|---|---|---|
| All items | Headline | +3.4% |
| Energy | Gasoline +24.6% | +14.7% |
| Shelter | ~1/3 of the index | +3.2% |
| Services | The sticky bulk of living | +3.1% |
| Electricity | The AI power bill | +4.2% |
| Core goods | Commodities less food & energy | +0.8% |
| Durables | Where abundance should show first | −0.2% |
| Used cars | Already deflating | −1.9% |
| Medical commodities | Pills, not physicians | −2.7% |
Why the buildout works against cheapness first
Data centers make this worse before they make it better. IEA-linked figures have global data-center power going from about 415 TWh (2024) toward about 945 TWh by 2030. Hyperscalers have committed ~9.8 GW of nuclear and only ~1.9 GW is operating. Three Mile Island’s restart for Microsoft is H2 2027. Western SMRs are 2029–2035. Until those watts exist, AI is a bidder for scarce electricity, not a source of cheap electricity.
Dallas Fed work already puts wholesale power 2–6% higher because of data centers. PJM capacity auctions are setting records, with data centers a large slice. Nuclear is real. It is not a 12-month, or even a 3-year, consumer-price story.
Humanoids are the same pattern. Figure and Agility have documented factory and logistics hours. Tesla Optimus is still mostly internal / R&D; Musk has pointed to late-2026 Fremont production and about end-2027 for public sales, at a hoped-for $20–30k. Industry forecasts of ~90k units in 2026 scaling toward ~1 million by 2030, even if they hit, are factory tools — a rounding error against a global labor force of billions.
Nuclear: committed, not delivering
TMI restart H2 2027
Western SMRs 2029–35
Humanoids: thousands
Consumer Optimus ~end-2027
What will not get cheap on any of those clocks
Land, housing in nice places, elite schools, doctors’ time, restaurants, live events, status. Robots do not print zoned lots. McKinsey’s own split is revealing: healthcare is among the least automatable sectors by 2030 (~20% of hours vs ~31% in manufacturing). Baumol’s cost disease is the boring name for this: as goods get cheap, the remaining human-only services eat more of the paycheck.
So “quality of life way up” and “prices way down” will diverge by category. Digital life can feel sci-fi while the rent check looks like 2024.
The honest score
Use the horizon tabs above to highlight a column. If one number has to be picked: five years for some prices and a real QoL jump for the top half of AI users; ten years for the thing Elon is actually describing, and only if energy and robots both land.
Household-facing outcomes
Highlighting 12 months
| Thing you actually care about | 12 mo | 3 yr | 5 yr | 10 yr |
|---|---|---|---|---|
| Cheaper Netflix / software / AI help | Yes | Yes | Yes | Yes |
| Cheaper physical goods | No | Starting | Plausible | If robots work |
| Cheaper electricity | Risk of up | Unlikely | Maybe starting | If nuclear/solar scale |
| Cheaper housing / healthcare | No | No | No | Policy, not tech |
| Median CPI going down | Almost no | Unlikely | Goods, not headline | That’s the bet |
| Significant QoL for people who use AI | Already | Clear | Large | Transformative |
| Significant QoL for the median household via cheaper living | No | Small | Moderate, lumpy | Elon’s decade |
The tell to watch
Not another keynote. Until these three move together, “hyperabundance” is a correct direction with a wrong speed. Don’t wait on abundance to make a housing, health, or career decision that has to work in 2026–2029.
- Core goods CPI going negative and staying there.
- Electricity prices rolling over after this buildout.
- Humanoids with verified factory hours in the tens of thousands, then hundreds of thousands.