The Dependence Ledger — At 250, What American Power Actually Rests On

The United States turned two hundred and fifty this month, and the anniversary invites exactly the wrong question. “Is America still number one?” is a scoreboard question, and scoreboards measure output. Output is not where national economic leverage lives.

Leverage lives in dependence — in the specific places where the rest of the world has no ready substitute for something American or American-allied. A currency most trade is still invoiced in. A settlement system you can be removed from. A machine that only one company on earth knows how to build. These are not the biggest things in the economy. They are the things with no second supplier, and that is a different property entirely.

Read that way, the picture at 250 is neither triumph nor decline. It is a ledger with entries moving in one direction, at very different speeds. Some dependencies are eroding gently by dilution. Some are being deliberately engineered away by a challenger with a twenty-year clock. And one of them has a property that makes it strange to hold at all: it depletes when you use it. This briefing is a structural map of that ledger — not a verdict on it.

Start with the entry that gets discussed most and understood least.

The dollar’s share of allocated global foreign exchange reserves has fallen from around seventy-one per cent at the turn of the century to roughly fifty-seven per cent by the middle of this year. Fourteen points in twenty-five years is a real decline and it is not a rounding error.

But the mechanism behind it is the opposite of the one usually described. Central banks have not, in aggregate, been selling dollars. Their dollar-denominated holdings have been broadly flat for more than a decade. What changed is the denominator: total reserves ballooned, and the new money went disproportionately into other currencies and into gold. The dollar’s share fell by dilution, not by exit.

That distinction matters enormously, because the two things have different futures. A stock being sold tends to keep being sold. A stock being diluted stops falling the moment the growth in everything else slows. It also explains why the quarterly numbers are so noisy: a meaningful part of any given quarter’s move is simply the dollar’s own exchange rate revaluing the pile, not a single reserve manager deciding anything.

The forward-looking signal sits elsewhere, and it is worth taking seriously precisely because it is about intent rather than arithmetic. Surveys of reserve managers this year have found a majority planning to reduce dollar allocations over the coming decade, and the reason they give is not yield or liquidity. It is political risk — the possibility of being on the wrong side of a sanction. That is a reputational cost attached to the settlement system, not to the currency, and it points at the next section.

Not every large American industry is a source of leverage. Leverage requires the absence of an alternative.

Two entries on that ledger are genuinely hard. The first is extreme ultraviolet lithography: the machines required to make the most advanced chips are built by exactly one company on earth, and there is no second source at any price. The second is the settlement layer — not the dollar as a unit of account, but the correspondent banking and clearing infrastructure that moves dollars, and the demonstrated ability to remove a country from it.

Two entries are softer than they look. The Treasury market is deep and liquid, and precisely because it is a financial asset rather than a physical capability, it is the most substitutable thing on the list — a reserve manager can hold euros or gold tomorrow, and increasingly does. Advanced manufacturing nodes are concentrated but contested, and capability there is diffusing faster than in the tooling underneath.

The useful habit is to stop asking “what is America big in?” and start asking “what has no second supplier?” Those are very different lists, and only the second one is leverage.

Here is the structural point that most commentary misses entirely, and it was put best by people arguing for tighter controls, not against them.

In February, a bipartisan group of lawmakers pressing for a blanket ban on semiconductor manufacturing equipment sales to China wrote that each chokepoint tool that enters China represents a permanent loss of American leverage. Read that sentence as an economic claim rather than a political one and it describes something unusual: leverage here behaves like a depleting resource, not a renewable one.

Most economic advantages compound with use. A chokepoint does the opposite. Every time it is exercised, it does three things at once: it extracts a concession, it proves to everyone watching that the dependency is dangerous, and it funds the search for a substitute. The first effect is immediate and visible. The second and third are slow and invisible, and they run in the opposite direction.

This is not an argument that export controls were a mistake — that is a judgement about priorities, and reasonable people land in different places. It is an observation about the shape of the instrument. A chokepoint used sparingly stays a chokepoint for a long time. A chokepoint used often teaches the world to route around it, and the routing is permanent in a way the concession is not.

Which brings us to the other side of the ledger, and to the thing that is easy to underestimate because it is so unspectacular: a challenger doing the same patient move over and over.

In energy, the move is stockpiling and diversification. China now holds the largest national strategic crude reserve in the world — on US official estimates somewhere around 1.4 billion barrels across state and state-company inventories — and spent 2025 adding to it at roughly 1.1 million barrels a day, which made it one of the single largest demand factors in the global crude balance. Alongside the stockpile sits deliberate supplier spread: the Middle East above half, Russia around a fifth, Brazil a few per cent. And alongside that, a steady push to settle some of it outside the dollar. None of these alone changes anything. Together they convert an acute vulnerability — an import-dependent economy that could be cut off — into a manageable one.

In semiconductors, the move is identical in shape. Denied the best tools, the response was not to stop but to build the second-best domestically and accept the cost. Chinese foundries reached seven-nanometre production using older deep-ultraviolet machines and heavy multi-patterning — more expensive, lower-yielding, and good enough to ship. The stated programme now runs to production lines using fully domestic equipment, a working domestic EUV capability somewhere in the 2028–2030 window, and eighty per cent self-sufficiency by 2030 against roughly a third in 2024.

The honest read of the export-control era is that it has worked and backfired simultaneously. It has genuinely held China back at the leading edge — the gap in the most advanced chips and in AI training efficiency is real and has not closed. It has also, by the assessment of the analysts who follow it most closely, become the single strongest accelerant of Chinese domestic tooling, by converting a commercial preference for Western equipment into a national security imperative to replace it.

Several things cut against the neat version of this story.

Targets are not achievements. Eighty per cent self-sufficiency by 2030 is a stated goal, and the history of Chinese semiconductor targets is a history of missed ones — an earlier seventy-per-cent target for 2025 was not met. A domestic EUV machine exists today as a programme and a prototype, not as a production tool, and the distance between those two states is measured in years and has defeated better-funded efforts before. Everything in the third figure above that is shown faded should be read as intention.

The dollar’s position is also stickier than the reserve numbers suggest. Reserve share is the most-quoted measure and among the least important: invoicing, trade finance, and the currency denomination of global debt are all far more concentrated in dollars than reserves are, and all of them are harder to move because they depend on everyone else moving at the same time. A reserve manager can diversify alone. An exporter cannot unilaterally decide what currency the contract is written in.

And the whole framing has a limit. “Leverage depletes when used” is a real property, but it says nothing about the right rate of use. A chokepoint held and never exercised produces no benefit at all. The question is not whether to spend the stock but how fast, and that is a political judgement this briefing is not equipped to make.

This is not a decline thesis. Nothing above argues that American economic power is ending, and the two hardest chokepoints on the ledger remain intact and are likely to for years. It is not a prediction that any Chinese technical target will be met on schedule — several explicitly may not be. It is not a position on whether export controls are good policy. And it contains no investment recommendation of any kind; no company, instrument or allocation is suggested anywhere above.

It is one structural observation, offered so the anniversary coverage is readable: power at this level is a ledger of other people’s dependencies, the entries move at very different speeds, and the most valuable ones have the awkward property of shrinking each time they are used.

So the next time you read that the dollar is being dethroned, or that export controls have crippled a rival, resist both headlines and ask the structural questions underneath them:

Is this dependency being sold out of, or merely diluted by everything else growing around it — because those have completely different futures. And when a chokepoint gets used, what did it buy, and how much of the stock did it spend?

We are not here to tell you who wins the next twenty-five years. We are here to make sure that when the ledger moves — and at 250 it is moving, quietly, in several places at once — you are reading the entries rather than the scoreboard.