Every June, the CMS Office of the Actuary publishes its projection of national health expenditures, and every June the coverage writes itself. Big number, bigger number next decade, health care eating the economy. This year the projections put 2025 spending at about $5.7 trillion, up 7.3% from 2024, a third consecutive year above 7% growth, on the way to roughly $9 trillion and 20.6% of GDP by 2034.
Those numbers are real and worth stating. But the actuaries' report is not really a set of numbers. It is a set of assumptions with numbers attached, and this year one assumption is doing an enormous amount of quiet work. Two former assistant secretaries of HHS flagged it in Health Affairs, and almost nobody else did.
The assumption that makes everything look fine
In 2025, Congress passed a law cutting more than a trillion dollars from health coverage, mostly Medicaid, and let the enhanced ACA subsidies expire. The obvious question is what those two moves do to the number of uninsured Americans.
The CMS actuaries' answer is: not much. They project the insured share of the population drifts down only modestly, from about 91.8% in 2024 to 90.5% by 2034. A little more than a percentage point over a decade, despite the largest coverage rollback in years.
Health Affairs called this out directly. The CMS coverage assumptions are, in the authors' words, the most consequential and questionable in the entire projection. By assuming only a modest rise in the uninsured, partly on the premise that states will absorb a larger share of the Medicaid bill, the actuaries make the math behind HR1 and the subsidy expiration look far more favorable than the Congressional Budget Office's analyses did.
Sit with why that matters. If you assume the coverage cuts barely raise the uninsured, then the federal government saved a great deal of money without much human cost, and the policy looks like a clean win. If the uninsured number is larger, which CBO's framework implied, the same savings come with millions losing coverage, and the "savings" are partly just a cost shift onto emergency rooms, uncompensated care, and people who stop getting treated. Same spending projection. Opposite moral.
The projection does not resolve that argument. It assumes an answer to it, and the assumption is the thing that makes $6 trillion read as manageable rather than alarming.
Where the money is actually going
Set the coverage fight aside and look at what is driving the raw growth, because the composition has shifted in a way worth noticing.
Two categories are pulling the number up. The first is prescription drugs, now the fastest-growing major category, driven above all by GLP-1s and expensive oncology drugs. An economist on the report named those two specifically. The obesity and diabetes drugs in particular are a genuinely new load on the system, because for the first time a mass-market population is on chronic, costly medication that Medicare has started to cover.
The second driver is the one that should embarrass the industry, and it is the one I would build the story around.
Administrative costs are growing faster than care
CMS projects that commercial insurer spending grows faster than overall health spending, and a major reason is not medicine. It is administration.
The actuaries expect non-medical health insurance spending, the overhead, the billing apparatus, the profit margin, to grow at about 7.5% a year. Their stated logic is almost cynical enough to admire. Insurers lost margin when medical loss ratios rose and Marketplace enrollment fell, and the actuaries assume insurers will act to recapture that lost profitability in future years. In plain terms, the model assumes the administrative layer will grow because the companies running it will make it grow.
That is the projection saying, in actuarial monotone, that a meaningful slice of the coming trillions is not going to doctors, drugs, or hospitals. It is going to the machinery of moving money between them. It connects directly to the billing arms race already inflating spending through AI-driven coding intensity: the actuaries have now baked rising administrative cost into the official forecast as a structural feature, not an inefficiency anyone expects to fix.
The projections have been wrong before, in a revealing direction
Here is the context that almost never makes the coverage, and it cuts against the alarmism.
The scary long-run numbers have a poor track record, and they miss the same way every time. They come in too high. Total health spending is now projected to be about $2 trillion lower than CMS expected in its 2010 projections, and hundreds of billions below its 2015 ones. Back in 1992, before Medicare Part D and the ACA even existed, the actuaries projected health care would reach an astonishing 32% of GDP. It is tracking closer to 20.
This is not because the actuaries are bad at their jobs. It is because projections extrapolate recent trends and strong assumptions, and health care has repeatedly bent below the trend line in ways the models did not anticipate. The 2010 projection could not see the late-2010s slowdown coming. The current one cannot see whatever bends the curve next.
Which is the honest frame for the whole exercise. The $6 trillion is close to certain because it is nearly here. The $9 trillion is a scenario, resting on assumptions about coverage, drug prices, and administrative behavior that the actuaries themselves hedge, and that history suggests will land high.
The one thing to hold onto
Health Affairs made the point that deserves the last word. It would be a mistake to take these projections literally, because technology and policy are already moving in directions the model cannot yet price. GLP-1s were barely a rounding error a few years ago and are now a top driver. AI in billing did not exist as a line item and is now reshaping administrative cost. The next such shift is not in the projection, by definition, because nobody has seen it yet.
So read the number for what it is. $6 trillion is a fact about how much the United States is about to spend on health care. It is not a fact about whether that spending buys health, whether it reaches the people the coverage cuts pushed out, or whether the fastest-growing dollars inside it are treating anyone at all. The report answers the easy question loudly and leaves the hard ones sitting inside its assumptions, where the actual story has been all along.