Here is a transaction that happens in San Francisco every week.
A company needs somewhere for four engineers to live while they work on a launch. The project runs twelve weeks. There is no twelve-week product on the unfurnished market, so the company signs a twelve-month lease on a two-bedroom, furnishes it, uses it for three months, and then either carries it empty, sublets it informally, or pays to break it.
One home. Twelve months off the market. Three months of human occupancy.
That unit is 25% utilized, and the other nine months are not a transfer of housing to anyone else — they are destroyed. Nobody lives there. The city counts the unit as occupied, the rent gets paid, and three quarters of a year of shelter evaporates. Multiply by every consulting team, visiting engineering group, relocating executive, accelerator cohort and research fellow in a city built on temporary technical work.
This article is about whether furnished month-to-month housing makes that better or worse. We operate in this market, so treat us as an interested party — which is exactly why the critics' case is stated here in full rather than skipped.
What the market actually looks like in 2026
The context is not a soft market with room to absorb mistakes.
| Measure | 2026 figure |
|---|---|
| New homes ready for occupancy, first half of 2026 | 405 |
| Renter-occupied households | about 224,900 (62% of households) |
| Median two-bedroom rent | $6,020/month, up 25.9% year over year |
| Median one-bedroom rent | $4,180/month, up 22.9% year over year |
| Active rental listings | down about 30% year over year |
| Vacancy | about 3.6% overall; around 3% in the Marina, Pacific Heights and SoMa, down from roughly 13% in 2020 |
| Units in the development pipeline | about 75,000 |
Four hundred and five homes in six months, against an AI hiring boom. That single number explains more about San Francisco rents than any other fact in this article. The pipeline holds 75,000 units; the city delivered 405 in a half-year. Supply is not responding to price.
The critique, stated properly
San Francisco's response to furnished corporate rentals has been hostile, and the concern is not frivolous.
The city defined the category in legislation as Intermediate Length Occupancy — furnished, serviced units rented on terms typically longer than 30 days and shorter than a year. The political objection runs: operators master-lease ordinary apartments from building owners, re-let them furnished at a premium, and in doing so remove homes from the permanent rental market while sidestepping the restrictions that govern short-term rentals. In a city where roughly 172,000 units sit under rent control — about half the rental stock — converting long-term homes into a rotating product looks to many officials like extraction dressed as hospitality.
Where that critique lands, it lands hard. An operator who takes a rent-controlled family apartment out of permanent circulation to rent it by the month to visiting engineers has made the housing problem worse, not better, whatever the occupancy rate. There is no honest argument otherwise.
Why the arithmetic still does not support the shortage claim
The question is scale. The Corporate Housing Providers Association has estimated the number of these units in San Francisco at around 3,000.
Three thousand, against roughly 224,900 renter-occupied households — about 1.3% of renter households. In the same city that delivered 405 new homes in six months and has 75,000 units stuck in a pipeline.
You cannot get from 1.3% to a 25.9% annual increase in two-bedroom rents. The shortage is produced by a development process that turns 75,000 approved-or-pending units into 405 completions, meeting demand from an industry hiring aggressively. Corporate furnished housing is a rounding error in that equation — which is also, to be fair, a reason not to oversell it as the solution.
The actual mechanism: utilization, not unit count
The useful way to think about temporary housing is not how many units it holds but how many people-months of shelter it delivers per unit-year.
Take the same four-engineer, twelve-week project. Solved with a twelve-month unfurnished lease, one unit serves one team and sits empty for nine months. Solved with furnished month-to-month housing, the same unit serves that team for three months and then three more groups across the rest of the year — a consulting team, a visiting researcher, a family between homes.
| 12-month lease for a 3-month need | Furnished month-to-month | |
|---|---|---|
| Unit-years consumed | 1.0 | 1.0 |
| People-months of shelter delivered | 12 (3 months × 4 people) | up to 48 |
| Months the home sits empty | 9 | near zero at healthy occupancy |
| Long-term tenant displaced | none in either case | none in either case |
The unit count is identical. The output differs by a factor of four. That is the entire argument, and it is an argument about waste rather than about adding buildings — a distinction worth being precise about, because inventory is created by construction, not by booking software.
There is a second-order effect that matters more than it sounds. Every company that solves a three-month need with a three-month product is a company not bidding on an unfurnished twelve-month lease against a household that wants to live here permanently. In a market with 3% vacancy in the desirable neighbourhoods and bidding wars on ordinary rentals, taking corporate demand out of the long-term queue is a real service to the people standing in it.
Where the model clearly helps, and where it should not go
Having run furnished homes in San Francisco for years, we think the distinction a policymaker should care about is not lease length. It is what kind of building is being used.
Legitimately additive:
- Large houses — six, eight, thirteen bedrooms — that are hard to let to a single household at a price anyone will pay, and sit under-occupied or empty otherwise. A six-bedroom house holding six founders is doing more housing work than the same house holding a family of three.
- Buildings between uses, in slow sale processes, or awaiting renovation, which would otherwise be vacant.
- Odd stock: ground-floor units, in-law configurations, converted institutional buildings. San Francisco's founder-housing boom has already absorbed converted dormitories and an archbishop's mansion.
- Demand that is structurally temporary. An accelerator cohort will leave in twelve weeks whatever product it is sold. Nobody is going to live in that room permanently.
Not additive, and reasonable to restrict:
- Converting rent-controlled family housing into a rotating furnished product.
- Taking ordinary long-term apartments off the permanent market in buildings where a household would otherwise have stayed for years.
That is a line we would accept in regulation, and it is a sharper line than lease duration. Judge the use by the alternative use of that specific building, and measure occupancy rather than term length. A home occupied 340 nights a year by successive one-month residents is working harder than one leased for twelve months and lived in for ninety days.
What we do
hyperstay runs furnished homes across San Francisco for founders, cohorts and corporate teams: private rooms or an entire house with six or more bedrooms, 30-night minimum, with rent, utilities, internet and cleaning in one monthly price. Our portfolio is deliberately weighted toward large homes — the stock that is genuinely difficult to let to a single household and easy to under-use — and toward demand that was never going to become a permanent tenancy: the batch, the fellowship, the three-month sprint, the team in from abroad to raise.
If you are a company weighing a twelve-month lease for a three-month need, the nine empty months are the real cost of that decision. Write to team@hyperstay.com and we will tell you what the months you actually need cost instead.
Policy researchers and journalists: we are happy to share operating data on occupancy, stay length and resident mix. team@hyperstay.com.