Phnom Penh Condo Occupancy Rates in 2026: The Data

Quick answer: Phnom Penh condo occupancy runs at roughly 75% to 80% on average in 2026, with rental vacancy near 15%. Serviced apartments sit at 70% to 85%. No continuous public series exists, so treat every figure as indicative and verify occupancy building by building.

Occupancy is the input that decides whether a yield figure survives contact with reality, and it is the number the Cambodian market publishes least reliably. Ask three sources and you will get three answers, none with a stated method.

This page states what is available, with dates and attribution, says plainly where the data is thin, and then shows the building-level check that replaces a market average.

What occupancy rate do Phnom Penh condos achieve?

Condo occupancy in Phnom Penh is reported at roughly 75% to 80% across districts for 2025 and 2026, with rental vacancy stabilising near 15%. Foreign buyers should read both as market-wide indications assembled from agency and portal reporting, rather than as a surveyed series with a published method.

Metric	2026 figure	Basis
Condo occupancy, market average	About 75% to 80%	Agency and portal reporting
Rental vacancy, market average	About 15%	Same, stabilised 2025 to 2026
Serviced apartment occupancy	70% to 85%	Sector reporting, centre stronger
Historical vacancy estimate	30% to 50%	Attributed to CBRE Cambodia, earlier cycle
Your own realistic assumption	1 to 2 void months	Site methodology

Three cautions belong on any occupancy checklist:

• Market averages mix segments that behave differently, from entry studios to prime one-bedrooms.
• Occupancy and vacancy are quoted inconsistently, sometimes as complements and sometimes not.
• A building-level figure is worth more than a city-level figure, and takes an afternoon to obtain.

Supply context, including the vacant-unit estimate of 15,000 to 20,000 and annual absorption of 3% to 4%, sits in the 2026 investment verdict.

Data quality deserves its own warning at the top of this page rather than a footnote at the bottom. Cambodia has no tenancy register, no obligation on landlords to report a letting, and no regulator publishing a quarterly occupancy series for residential stock, so every figure in circulation is assembled by an agency or a portal from whatever sample it can see. That does not make the numbers useless, and it does mean a 5 point difference between two published figures is noise rather than a trend, whatever headline sits above it. Anyone quoting occupancy to one decimal place in this market is quoting a sample, not a census. Treat the 75% to 80% band as the shape of the market, model your own unit at 1 to 2 void months, and put the verification effort into the building you are actually buying rather than into reconciling two aggregators.

Occupancy against vacancy: the same fact stated twice

Occupancy and vacancy describe one condition from two directions, and mixing them is the most common error in Cambodian market commentary. An occupancy of 80% implies vacancy of 20%, yet reporting pairs 75% to 80% occupancy with a 15% vacancy figure, and those two numbers cannot describe the same population of units.

Statement	What it implies	Where the mismatch comes from
Occupancy 80%	Vacancy 20%	Physical units in a sample
Vacancy 15%	Occupancy 85%	Listed units, or a different sample
Occupancy 75% to 80% and vacancy 15%	Inconsistent	Two samples, two methods, one paragraph

Read published figures with three questions:

1. Is the denominator all units, or only units offered for rent?
2. Does the sample include unsold developer stock, which is empty by definition?
3. Is the period a snapshot or an average across the year?

Unsold developer inventory is the usual explanation. A tower that has sold 60% of its units and let most of them can be described as well occupied by an owner and poorly occupied by a market analyst counting every unit in the building, and neither is lying. For a foreign buyer the distinction is practical rather than academic. The units competing for your tenant are the ones actually offered for rent, so a tower holding 40 unsold apartments the developer will eventually market is a different proposition from a fully sold building with 3 units available today, even if both report the same headline occupancy. Ask which population a figure describes before using it in a model, and where nobody can answer, treat the number as commentary rather than as data you can underwrite against.

Insider tip: count the live rental listings in your target tower on two portals, then ask the building manager how many units are currently tenanted. Where the manager’s answer and the listing count disagree sharply, the listings are the more honest signal, because a manager rarely counts units an owner is quietly marketing.

What the serviced apartment segment shows

Serviced apartment occupancy in Phnom Penh is reported at 70% to 85%, stronger in central districts, and foreign buyers should read the segment as a demand signal rather than as a direct comparable. The segment wins corporate and short-stay tenants at monthly rates an unbranded condo does not reach.

Segment	Occupancy	What drives it
Serviced apartments, centre	Toward 85%	Corporate contracts, hotel-style service
Serviced apartments, wider city	From 70%	Price competition, newer supply
Standard condo, central	Higher end of 75% to 80%	Expatriate and regional tenants
Standard condo, outer	Lower end	Local demand, single-employer risk

Two takeaways for a condo buyer scenario:

• Corporate demand exists in the centre, and the serviced segment captures much of it.
• A condo competing with serviced apartments needs management a corporate tenant accepts.

Management quality is the practical lever here, and it is covered in the property management guide.

Serviced occupancy is also the clearest evidence that Phnom Penh’s rental weakness is a supply problem rather than a demand problem. Corporate and relocation demand exists, and the segment engineered to capture it reports 70% to 85% occupancy while standard condo stock sits at 75% to 80% across a much larger and more varied population. What the serviced operators provide is not a better location, it is reception, cleaning, maintenance response and a billing relationship a company’s finance department accepts. An individual owner cannot replicate all of that, and can replicate enough of it to compete: a professional manager at 8% to 12% of collected rent, a furnished unit at the standard a corporate tenant expects, and a lease a company’s finance team can sign without amendment. Those three decisions move your own occupancy more than any district choice.

Why published occupancy figures disagree

Published occupancy figures disagree because there is no shared method, no central register of tenancies and no obligation to report. Estimates in the 30% to 50% vacancy range circulated during the oversupply build-up, attributed to CBRE Cambodia, while current reporting suggests something closer to 15%, and both can be defensible on their own definitions.

Reason for divergence	Effect on the figure
Different denominators	10 points or more of apparent difference
Unsold stock counted or excluded	Large, in newly completed towers
Sample skewed to marketed units	Flatters occupancy
Snapshot versus annual average	Seasonality moves the number
Segment mix	Entry studios and prime units behave differently

Practical response for a foreign buyer:

1. Use ranges, never point estimates, in any model you rely on.
2. Prefer building-level evidence over city-level commentary.
3. Where a source states no method, treat the number as directional only.

Our own methodology is deliberately conservative: 1 to 2 void months a year on central stock and 2 to 3 on outer-district entry units, applied before any yield comparison. That is the equivalent of 83% to 92% occupancy centrally and 75% to 83% further out, which is close to published market averages while being derived from something we can check. The reason for building the assumption bottom-up rather than importing a market average is simple. A market average includes buildings you would never buy in, developer stock nobody has marketed yet, and segments with entirely different tenants, so it answers a question about the city rather than a question about your unit. Ask the narrower question and the answer is both easier to obtain and more useful: how many units in this tower are let today, and how long did the last one take.

How to verify occupancy in one building before buying

Verification requires three data points and about an afternoon, and it replaces a 75% to 80% market average with a figure specific to your asset. Foreign buyers who run the check routinely find their target building sits 5 to 10 points away from the citywide number in one direction or the other.

Check	How to do it	What good looks like
Live rental listings in the tower	Filter two portals by building name	Low single digits for a mid-size tower
Tenanted share	Ask building management directly	A specific number, given quickly
Two signed leases	Ask the agent or seller	Rents at or near the quoted level
Void history on the unit	Ask the current owner	Weeks rather than months between tenants
Unsold developer stock	Ask the developer or manager	Small, or already marketed

Then convert what you find into a model input, which is the methodology we apply on every file:

• Low listing count and a fast manager answer: model 1 void month.
• Many live listings and vague answers from management: model 2 to 3 void months.
• Unsold stock still to be marketed: add a month, because that competing supply arrives later, after you own.

Tenant screening, which shortens voids once you own, sits in the tenant screening guide, and the yield effect of every void month in the net versus gross guide.

Listing counts are the most reliable of the three checks, and the reason is that nobody controls them. A manager may round the tenanted share upward and a seller may describe the tower as full, while a portal simply shows what owners have advertised this week, in the same building, at the prices they are asking. A mid-size tower carrying 18 live listings is telling you something a market average never will, and the same count taken 6 weeks apart tells you whether those units are moving. Two portal searches and a calendar reminder are the whole method, and they cost nothing beyond the 10 minutes it takes to run them twice, 6 weeks apart.

What does one void month cost you?

A void month is the most expensive line in a Cambodian rental model, and it is worth pricing before you compare buildings. On a $130,000 unit letting at $750 a month, each empty month removes $750 of rent and about 60 to 90 basis points of net yield, while the service charge and the sinking fund keep running.

Void length	Rent lost on a $750 unit	Rough effect on net yield
2 weeks	$375	Minus 30 to 45 basis points
1 month	$750	Minus 60 to 90 basis points
2 months	$1,500	Minus 120 to 180 basis points
3 months	$2,250	Minus 180 to 270 basis points

Three things shorten voids, in order of effect, and all three belong on a post-purchase checklist:

1. A building corporate tenants accept, which produces scheduled renewals rather than random departures.
2. A furnished unit at market standard, because unfurnished stock lets slower in this market.
3. A manager who starts remarketing 6 weeks before a lease ends rather than after it expires.

Vacancy is also the only occupancy variable a foreign owner partly controls. The market average is a fact about the city, the void history of one tower is a fact about a building, and the length of your own voids is a consequence of decisions you make at purchase and at each re-let. Buyers who treat 15% market vacancy as their own forecast are importing someone else’s average; buyers who verify the tower and then manage the re-let actively usually land closer to one void month than to two. The yield arithmetic behind each of those months sits in the net versus gross guide.

Advantages and disadvantages of the current occupancy picture

Stabilisation at roughly 75% to 80% is better than the earlier cycle and still leaves a wide gap between a well-let building and a poorly-let one.

Advantages	Disadvantages
Reporting suggests stabilisation from the earlier cycle	No official series to rely on
Central districts perform above the average	Outer districts sit below it
Serviced segment shows real corporate demand	That demand is largely captured by serviced stock
Building-level verification is cheap and quick	Requires cooperation from management
Conservative modelling is easy to apply	Honest models make fewer units look attractive

Buyer scenarios by occupancy assumption

Three profiles use occupancy differently.

1. Income buyer, central stock. Model 1 void month, verify with listings, and expect the deduction stack to matter more than occupancy.
2. Yield-maximising buyer, outer district. Model 2 to 3 void months. The high gross band exists partly because voids are longer.
3. Buyer planning to use the unit part of the year. Occupancy becomes a scheduling question, and the short-term option is compared in Airbnb versus long-term letting.

All three profiles need the same first step, which is a count rather than an average. Get the number of tenanted units in the tower, get the live listing count, and only then decide which void assumption belongs in your model.

Serviced stock runs at the top of the ranges on this page, at 70% to 85% and stronger in the centre. The reason that occupancy premium exists, and whether it survives the cost of providing the service, is worked through in the serviced apartments guide.

Red flags in an occupancy claim

An unverified occupancy claim is easy to spot, and four phrases give it away. Foreign buyers can resolve all four inside 1 working day with a seller who keeps orderly records, because each red flag hides the same missing item: a count of tenanted units against the 75% to 80% average.

Ask all four in one message, then set your void assumption from the answers:

Red flag	What it signals	What to ask for
”The building is fully occupied”	No count has been made	Number of tenanted units, in writing
Occupancy quoted with no source	Market commentary reused	Source and date
Unsold developer units not mentioned	Future competing supply hidden	Unsold count and marketing plan
Void history unavailable	Owner does not track it	Two signed leases and their dates

Four questions close the gap:

1. How many units in this tower are tenanted today, as a number?
2. How many are listed for rent right now on the portals?
3. How many remain unsold and still to be marketed by the developer?
4. How long did the last two re-lets take, in weeks?

Our underwriting will not accept a citywide occupancy figure as an input for a specific building, because the variance between buildings is wider than the variance between years. District bands sit in the district yield comparison, the resale consequence of a soft rental market in the resale liquidity guide, and the years-to-recover arithmetic in the break-even guide.

Frequently Asked Questions

What occupancy rate do Phnom Penh condos achieve? +

Roughly 75% to 80% on average across districts for 2025 and 2026, with rental vacancy near 15%. Both figures are indicative rather than survey grade.

Is there an official occupancy series? +

No continuous public series exists. Figures come from agency and portal reporting with different methods, which is why they disagree by 10 points or more.

What occupancy should I model? +

Model 1 to 2 void months a year on central stock, 2 to 3 further out, then stress-test at 3. Your building matters more than the city average.

What occupancy do serviced apartments get? +

Typically 70% to 85%, stronger in the centre, at higher rates and with a hotel-style cost structure.

Was vacancy worse in earlier years? +

Yes. Estimates of 30% to 50% were attributed to CBRE Cambodia during the oversupply build-up, which is the context for today's figures.

How do I check occupancy in a specific building? +

Count live rental listings in the tower on two portals, ask management how many units are tenanted, and ask for two signed leases.

Can you check occupancy for a building I am considering? +

Send the tower name. We count live listings, ask management for the tenanted share, and return the void assumption we would use.