Dementia will cost the United States $818bn in 2026, according to a modelling study by Johanna Thunell and colleagues at the University of Southern California, published online in Alzheimer's & Dementia on 24 June 2026. The estimate carries a 95% confidence interval of $759bn to $866bn. It is not a tally of bills paid: it is a societal cost figure produced by the US Cost of Dementia Model, a dynamic microsimulation built on the Health and Retirement Study and administrative data from the Centers for Medicare and Medicaid Services, and documented in a companion methods paper by Bryan Tysinger and colleagues published the same day.

The composition of the number is the story. The single largest line is valuation of quality-of-life loss for the 5.7 million adults aged 51 and over living with dementia this year, put at $320bn (95% CI $269bn to $363bn), with a further $15bn for care partners' quality-of-life loss. Unpaid care comes next: 5.2 million family members and friends providing 6.8 billion hours, valued at $237bn. Medical and long-term care accounts for $222bn, of which Medicare and Medicaid cover roughly 70% ($154bn) and patients and families about 20% ($46bn) out of pocket. Earnings forgone by people with dementia and their care partners adds $23bn.

Set against that, the Alzheimer's Association's 2026 Facts and Figures report, published on 21 April 2026, projects $409bn in payments for health care, long-term care and hospice for people aged 65 and over with dementia, with Medicare and Medicaid covering $263bn (64%) and out-of-pocket spending at $103bn. The two figures are not rivals and should not be added or subtracted. They count different things over different populations: one is money changing hands for care, the other a modelled societal total that assigns dollar values to unpaid hours and to lost wellbeing.

That valuation choice is what any funding argument citing $818bn is really leaning on. Roughly two-fifths of the headline is quality-of-life loss, a construct derived from model simulations of what health and economic outcomes would have looked like without dementia, not an observed transaction. Read narrowly, the transactional part of the picture is smaller. Read as the authors intend, it is a statement that most of what dementia costs the country never appears on an invoice.