The Future of AI · Money & Markets
The Same Musk Claim, Measured in Dollars: I Say 38%
World nominal GDP, measured in current US dollars, will be at least 2.0x its calendar-year 2026 level in a calendar year ending on or before December 31, 2036.
Doubling in current dollars needs 7.18% a year. The 2025 to 2026 increase was 6.87%, and carrying that forward for a decade lands at $245.4 trillion, about $7.2 trillion short. So the claim needs a recent and unusually strong year repeated for ten years, plus a small acceleration. That is mostly a bet on the global inflation path and the dollar, and only marginally a bet on AI.
Call added September 11, 2026, after publication.
Above FRED by 17 points. Compounding gains in economic efficiency, and the innovation those gains unlock, push nominal output harder than a straight extrapolation of the last year suggests.
17 points apart. A meaningful split.
Resolves HIT only if the IMF World Economic Outlook world GDP series at current prices in US dollars shows world nominal output at or above $252.6 trillion, being 200% of the 2026 level of $126.3 trillion, for a calendar year ending on or before 2036-12-31, as stated in the first full release covering that year. Resolves PARTIAL if world nominal output reaches at least $189.5 trillion, being 150% of the 2026 level, by that date but not $252.6 trillion. Resolves MISS in every other case. Constant-price GDP does not resolve this. Purchasing-power-parity GDP does not resolve this. If the IMF restates the 2026 baseline in a later World Economic Outlook, the restated baseline replaces $126.3 trillion and the thresholds move with it, because the claim is about doubling rather than about a fixed dollar figure. Data revisions published after the first full release covering a year do not reopen a settled resolution.
Written before the outcome. Not reinterpreted after.This is the same Elon Musk sentence as the real-terms prediction, measured in the other unit. The two readings are 34 points apart.
On September 9, 2026 Musk posted: “AI+robots will more than double the global economy in less than 10 years.” He did not say whether he meant dollars or output. That single omission is worth more than any argument about robots.
What doubling costs in dollars
World GDP in 2026: $126.3 trillion in current dollars, per the IMF’s April 2026 World Economic Outlook. In 2025 it was $118.18 trillion. The one-year gain was $8.12 trillion, or 6.87%.
Doubling in ten years requires 7.18% a year, compounded — the tenth root of two. That takes $126.3 trillion to about $252.6 trillion.
Run the most recent year forward and you land at $245.4 trillion. Short by roughly $7.2 trillion, a gap of under 3%.
So the nominal claim does not need a miracle. It needs one strong year to repeat ten times, with a small acceleration on top.
Why this is a weaker claim than it sounds
Nominal GDP counts price increases as growth. An economy producing exactly the same goods and services, with 7.2% annual inflation, doubles in nominal terms over a decade while adding nothing real at all.
That matters here because the claim is about AI. If ordinary inflation and trend growth carry the statement across the line, the statement measured nothing about AI.
This is why we published both versions. In constant prices the same sentence is a 4% proposition. In current dollars it is a 38% proposition. Same words, same deadline, one unstated assumption, and a 34-point spread. A forecast that can be settled by choosing a deflator after the fact is not a forecast.
Why 38% and not higher
The required 7.18% is above the long-run average for global nominal dollar GDP over the past two decades, and the 2025 to 2026 figure of 6.87% was a strong year rather than a typical one.
The variables that actually decide this are mostly not about AI:
- The global inflation path over ten years, which no one forecasts well at that horizon
- The US dollar, since world GDP in current dollars falls when the dollar strengthens, regardless of what any economy produces
- Real growth, which contributes its own roughly 3%
Add trend real growth of 3% to global inflation near 3% and you get roughly 6% — short of the required pace. Getting to 7.18% needs either persistently higher inflation, a weaker dollar, genuinely faster real output, or some combination. Each is plausible. All of them holding for a decade is the part that makes this 38% rather than 60%.
Where Matt lands
Matt is at 55%, seventeen points above me.
His argument is that compounding efficiency gains and the innovation they unlock push nominal output harder than extrapolating recent years would suggest. Efficiency does not just lower costs; it releases capital and attention into new production, and that second-order effect does not show up in a straight-line projection.
The honest summary of the disagreement: I am forecasting inflation and the dollar. He is forecasting compounding. Those are different bets wearing the same number, and the annual scorecard below will show which one the data is tracking.
The annual scorecard
We are not waiting until 2036.
Reaching $252.6 trillion on a 7.18% path means hitting these levels:
| Year | Required world GDP | Cumulative vs 2026 |
|---|---|---|
| 2027 | $135.4T | 107.2% |
| 2028 | $145.1T | 114.9% |
| 2029 | $155.5T | 123.1% |
| 2030 | $166.7T | 131.9% |
| 2031 | $178.6T | 141.4% |
| 2032 | $191.4T | 151.6% |
| 2033 | $205.2T | 162.5% |
| 2034 | $219.9T | 174.1% |
| 2035 | $235.7T | 186.6% |
| 2036 | $252.6T | 200.0% |
Each October, when the IMF publishes, the actual figure gets recorded here against that year’s checkpoint, along with the growth rate now required across the remaining years to still reach doubling.
That last column is the useful one. A single year below pace is noise. A required rate that climbs from 7.18% to 8% to 9% is a forecast dying in public, on a schedule, where anyone can watch it.
What would change my mind
- Two consecutive years of global nominal growth above 8%
- A sustained dollar decline against a broad basket, which mechanically inflates the world total
- Global inflation settling above 4% rather than near 3%
- Real growth clearing 4% on the back of measurable AI contribution
The first and third together would put this above 50% quickly.