Four Caregivers per 100 Seniors Is the Robot Pitch

Bloomberg Technology reported that the US spends roughly $400 billion a year on long-term care and has only four formal caregivers for every 100 people over 65, the shortfall that companion robots such as Andromeda's Abi are being built to address.

The economics of American elder care are simple and brutal. The country spends roughly $400 billion a year on long-term care, and for every 100 people over the age of 65 there are only four formal caregivers. That ratio — not the engineering — is what gives humanoid robotics its most credible commercial story to date.

That was the framing of a discussion on Bloomberg Technology featuring Andromeda Chief Operating Officer Jennifer Haroon, Brookings researcher Gopi Goda, 1X Chief Designer Dar Sleeper, and Nvidia's Kimberly Powell. The question on the table was whether machines can genuinely relieve the caregiving crunch, or whether the technology is still years from being useful in a living room.

What a four-per-hundred ratio actually means

Stated the other way around, four formal caregivers per 100 seniors works out to one caregiver for every 25 people over 65 — an illustrative restatement of the same figure, not a separate data point. No staffing model absorbs that. It is why the shortfall shows up as unpaid family labor, as adult children cutting hours or leaving jobs, and as long waits for home health visits, rather than as a line item anyone can bill.

The $400 billion figure is the size of the formal market: nursing facilities, home health agencies, assisted living, the paid portion of the system. The informal portion — spouses, daughters, neighbors — sits outside it and is not priced at all. Any honest assessment of what robotics can sell into has to distinguish between the two. Displacing paid clinical hours is a regulated, licensed, liability-heavy business. Supplementing unpaid hours at home is a consumer product.

Abi, the humanoid built by Andromeda, is squarely in the second category. It is designed for emotional companionship for dementia patients — conversation, presence, routine — rather than lifting, bathing, or dispensing medication. That is a deliberate narrowing. Companionship does not require a robot to be strong, and it does not require regulatory clearance in the way a device that touches a patient's body does.

Why companionship is the easiest beachhead

Dementia care is unusually labor-intensive for reasons that have little to do with physical tasks. Patients repeat questions, become agitated in the late afternoon, and need someone present far more hours than any staffing budget allows. A machine that can hold a conversation indefinitely without fatigue is, in principle, addressing the part of the job that burns caregivers out fastest.

The skeptical case is equally clear. Companionship delivered by a machine raises consent questions when the person receiving it has cognitive impairment. Families may accept a device that reminds a parent to take pills; they may feel very differently about one that substitutes for a human voice. Efficacy data in this space is thin, and researchers of the kind Brookings employs tend to press hardest on whether measured outcomes — hospitalizations avoided, caregiver hours returned — justify the spending.

There is also the cost question that no humanoid developer has answered convincingly. Elder care is a low-margin, price-sensitive sector. A robot that costs more per year than a part-time aide solves nothing for the households that need it most, and the households that can afford it already buy human help.

Where Nvidia sits in the supply chain

Nvidia (NVDA) is not building elder-care robots. It sells the compute and the simulation tooling that companies like Andromeda and 1X use to train them, which is why its representatives keep turning up in conversations about physical AI. That position — picks and shovels for an industry that has not yet proven its end market — is the same one the company occupied in generative AI before the revenue arrived.

Nvidia shares last traded at 225.07, up 0.22% on the day, as of the close on Friday, September 25, 2026, against a previous close of 224.58 and a session range of 223.13 to 226.94. The broader tape was firmer: the S&P 500 tracker closed at $771.35, up 0.54%, the Nasdaq 100 tracker at $744.50, up 0.46%, and the Dow tracker at $517.49, up 0.94%. Nothing in that session was about robots — but it is a reminder that the humanoid thesis is currently priced inside semiconductor names, not inside pure-play robotics revenue.

1X, whose chief designer joined the discussion, is developing humanoids for home environments. Andromeda is private. For public-market investors, there is no clean way to own the elder-care robotics theme directly today; exposure runs through compute suppliers, component makers, and eventually the health systems and insurers that would pay for deployment.

The buyer problem nobody has solved

The hardest commercial question is not technical. It is who writes the check. Medicare does not broadly pay for long-term custodial care. Medicaid does, for those who qualify financially, but state budgets set the rates and those rates are not generous. Private long-term care insurance covers a small slice of households. That leaves families paying out of pocket, and family budgets do not underwrite capital equipment.

A useful way to size the opportunity is to ask what share of the $400 billion a robotics product could plausibly touch. If companion devices ever captured 1% of that annual spend, it would be $4 billion a year — an arithmetic illustration, not a forecast, and one that depends entirely on a payer emerging. Until a health plan, a state Medicaid program, or a large operator of senior living facilities decides that a machine reduces its costs, humanoid elder care remains a consumer novelty with a compelling slide deck.

What to watch next

Three markers will tell you whether this is a business or a demo. First, any reimbursement decision — a state waiver, a Medicare Advantage supplemental benefit — that puts a public or insurer dollar behind a companion device. Second, published outcomes from deployments in dementia units, specifically on agitation, falls, and staff turnover. Third, unit pricing: the moment a humanoid is quoted at a monthly rate comparable to a few hours of weekly home health, the addressable market changes shape.

Demographics guarantee the demand. They guarantee nothing about the supply of robots that work.

Key facts

• US long-term care spending: Roughly $400 billion a year
• Caregiver ratio: 4 formal caregivers per 100 people over 65
• NVDA last close: 225.07, +0.22%, as of Sept 25, 2026, 20:00 GMT
• Named participants: Andromeda, 1X, Nvidia, Brookings

Frequently asked questions

How large is the US long-term care market?

The United States spends roughly $400 billion a year on long-term care, according to figures cited in a Bloomberg Technology discussion on elder-care robotics. That total covers the formal, paid system — nursing facilities, home health agencies and assisted living. It excludes unpaid care provided by family members, which is not priced or counted in that figure.

What is Abi and what does it do?

Abi is a humanoid robot built by Andromeda, a private company, designed to provide emotional companionship for dementia patients. Its focus is conversation, presence and routine rather than physical tasks such as lifting or bathing. That narrower scope avoids much of the regulatory burden that applies to devices making physical contact with patients.

How severe is the caregiver shortage?

For every 100 people in the United States over the age of 65, only four formal caregivers exist. Restated, that is roughly one paid caregiver for every 25 seniors. The gap is absorbed largely by unpaid family labor — spouses and adult children who reduce working hours or leave jobs entirely to provide care.

Is Nvidia building elder-care robots?

No. Nvidia supplies the computing hardware and simulation software that robotics developers use to train humanoid systems. Its role is upstream, in the same picks-and-shovels position it held in generative AI. Nvidia's Kimberly Powell participated in the Bloomberg Technology discussion on whether robots can help address the elder-care shortfall.

Can investors buy exposure to elder-care robotics directly?

Not cleanly. Andromeda is private and 1X, whose chief designer joined the discussion, is a humanoid developer without a public listing referenced in the report. Public-market exposure currently runs indirectly through semiconductor and compute suppliers, component makers, and potentially the health systems or insurers that would eventually pay for deployments.

What would signal that companion robots are becoming a real business?

Three markers matter: a reimbursement decision putting insurer or public money behind companion devices, published clinical outcomes from dementia-unit deployments covering agitation, falls and staff turnover, and unit pricing that compares favorably with a few hours of weekly home health. Absent a payer, the category remains a consumer product with limited reach.

Sources

• Why Humanoid Robots Might Be the Future of Elder Care — Bloomberg Technology