Boise's vacancy rate is 2.8%. It's also 8.1%.

Two weeks ago I had two Boise apartment market reports open on the same screen.

One said metro vacancy was 2.8%. The other said 8.1%.

Same market, same year, and neither one is wrong.

If you’re deciding whether to put money into a Boise apartment deal, that gap isn’t trivia. It’s the difference between a market with no vacancy and a market with a real problem. So I went and found every published Boise vacancy rate I could get my hands on this year, and I lined them up.

Eight desks, eight answers

Three questions decide which number you get.

Does it count lease-ups? CoStar’s 8.1% includes a brand new building that’s 45% empty and still leasing. Marcus & Millichap’s 2.8% is the stabilized read. One is measuring the whole market, the other is measuring the market you’d actually compete against if you owned a 1970s building on the Bench.

What size property is in the index? ALN and Cushman & Wakefield track properties of 50 units and up. Most of what I buy is smaller than that. My buildings are not in anyone’s index, and neither are most of the buildings a private investor will ever look at.

Which quarter, and is it a forecast? MMG’s 2026 forecast implies about 6.3%. Cushman’s Q1 print was 5.4% and its Q2 print was 4.8%. Quoting the Q1 number in September isn’t dishonest, it’s just old.

The rule I use now: never quote a vacancy rate without the vendor and the universe attached to it. A number without those two things is a rumor with a decimal point.

What all eight agree on

Underneath the noise, they point the same direction. Boise stopped building apartments.

Permits are the honest leading indicator, because a unit nobody permitted in 2026 can’t show up in 2028. Ada and Canyon counties permitted more than 4,700 multifamily units in 2022. In 2025 they permitted about 1,550. The first four months of 2026 ran 345 units against 510 in the same months of 2025.

The 2027 and 2028 supply picture was set two years ago. It’s already fixed.

Then there’s the pipeline number, which is where most people get fooled. The headline is 11,735 units, and it sounds enormous. Commercial Northwest did something I wish more shops did: they took all 46 tracked projects and checked every one against city permit records, the Boise development tracker, and COMPASS entitlement layers.

Two thirds of that pipeline is entitled and nothing has been built. Ten percent is actually vertical. They also pulled 439 units out of the count entirely, including a 170-unit approval that expired in 2025 on a site that’s becoming an aquarium.

It’s the most useful piece of Boise research published this year, and the lesson in it is simple. Underwrite permits, not plans.

Meanwhile the demand side keeps clearing what does get built.

Cushman counts 1,371 units absorbed year to date against 452 delivered, and the fourteenth straight quarter of positive absorption. CoStar’s forecast has demand exceeding new supply every year through 2030.

Concessions tell the same story three different ways. Marcus & Millichap has the Boise concession rate falling from 19.5% to 8.8% in a single quarter. Cushman has the share of properties offering a concession at 5.8%, down 650 basis points in a year. ALN has it at 20.1%, down from 25.3%.

Three desks, three definitions, three different numbers, one direction. Landlords stopped paying people to move in.

Why the demand keeps showing up

This is the part that isn’t really about apartments.

Micron is spending roughly $50 billion across two Idaho fabs, which is the number in the federal CHIPS project record, and it expects to create more than 17,000 jobs. About 3,500 of those are direct manufacturing and facility positions, with another 6,000 in construction, spread over twenty years. The first fab reached cleanroom blowdown this spring with DRAM output targeted for 2027. The second starts construction this year and is meant to be operational by the end of 2028.

On August 20 Micron announced something separate and new: a $10 billion research campus in Boise, with groundbreaking expected in 2027. That’s on top of the fab money, and it hasn’t made it into most of the market reports yet.

The rest of the demand picture:

• Ada and Canyon counties hit 876,760 people as of April, up about 150,000 since the 2020 census. Three of every four new Idahoans came from another state.
• Idaho unemployment was 3.6% in August. Southwestern Idaho was 3.3% in July, with 32,573 open job postings at an average advertised wage of $30.09 an hour.
• The median Ada County resale home hit $595,000 in August, up 6.9% on the year, with 2.4 months of supply. Freddie Mac’s 30-year fixed was 7.03% this week.

Run that last one. A household earning thirty dollars an hour isn’t buying a $595,000 house at 7%. They rent, and they keep renting, and that’s the whole demand thesis in one sentence.

The parts that argue the other way

I’d rather hand you the counterweights than have you find them yourself.

Boise didn’t stop building. It stopped building apartments. Single-family permits in the metro are running above last year’s pace, 5,884 units through August against 7,746 for all of 2025. The shortage is multifamily-specific. Anyone telling you the whole region shut off construction is overselling it.

The forecasters don’t agree about the back half.

Underwrite the first number and 2028 looks like a drought. Underwrite the second and it looks ordinary. Nobody can tell you which is right.

Debt got more expensive while I was writing this. The 10-year Treasury closed at 4.96% on September 22, up from 4.47% on June 1, and it touched 5.01% twice this month. That’s about fifty basis points in a quarter, nearly all of it in the last three weeks. Anything underwritten on a June rate assumption is stale, including some things I wrote in June.

And the market isn’t one market.

Caldwell is at 9.0% vacancy with rents down 1.4% and inventory up 12.7% this year. Nampa has 299 units under construction against 1,491 units of existing inventory, and it has absorbed nothing at all in 2026. Meanwhile West Boise is at 2.9% vacancy with rents up 5.0%.

West Boise and Caldwell are thirty miles apart and 6.4 points apart on rent growth. When somebody sells you “Boise,” ask which Boise.

What our own two buildings say

All of the above is somebody else’s data. Here’s ours.

We own two Boise buildings. One is 45 units on the Bench. The other is 16 units in the North End, built in 1971, and we’ve owned it since October 2025. Neither is big enough to appear in any of the indexes above, which is exactly why they’re worth showing you.

One floorplan, eleven months

Ten of the 45 units at our Bench property are the same renovated two bed, one bath floorplan. That lets me watch a single plan in a single building across eleven months, with no modelling in between. We closed on it at the end of August, so the earlier leases here were signed by the prior owner. The rents are still real signed contracts either way.

The leases that started in October 2025 averaged $1,560. The ones that started in April and May 2026 averaged $1,617. That’s $57 in seven months, call it 3.7% annualized. It isn’t 2021, and it’s noisy, because ten leases on one plan always will be. But it’s signed contracts rather than an index, and it points the same way Cushman does without being downstream of Cushman.

The building is fully leased as I write this.

Ten months of mark-to-market

Four price moves in ten months, each one dated to a listing that went up or a lease that got signed. We haven’t offered a concession once in that stretch, while three of our comps were stacking them through last winter. One of those comps had its effective two-bedroom rent down at $1,215 in December.

One two-bedroom went from $1,599 to $1,675 on a resident turn. It’s a non-renovated unit with no dishwasher that got a bare-minimum turn. That $76 wasn’t renovation. That was just the market catching up to an old lease. We also re-let it before the resident had moved out.

Now the number nobody publishes.

In March we asked $1,950 for a three-bedroom. It sat for 70 days and nobody took it. We cut it to $1,895 at the end of May and it leased in 22 days. That’s the ceiling, and it’s the most useful figure in this entire piece, because every report above will tell you what rents did and none of them will tell you the rent that didn’t work.

A second three-bedroom came up for renewal on September 30. We offered $1,950 again. As I write this it’s still unsigned.

What would change my mind

Three things, and I’m watching all of them.

The 7,864 entitled units. They aren’t built, but they’re approved. This is a financing trough, not an entitlement wall. If debt gets meaningfully cheaper, some of that goes vertical faster than the forecasts assume.

A Micron delay. Fab two is the 2028 demand story. If that timeline moves, the back half of every projection above moves with it.

The 10-year going through 5.25% and staying there. At that point the conversation stops being about rent growth and starts being about what you can finance.

The short version

Supply in Boise is genuinely thin and stays thin through 2028, because the permits that would have prevented it were never pulled. Demand is real and attached to an employer spending fifty billion dollars. And the market is nowhere near uniform, so the submarket and the building matter more than the headline does.

Every model has a rent growth assumption in it, including ours. The discipline isn’t refusing to forecast, it’s refusing to need the forecast. I want a deal that works on the rents the building collects today, so the supply trough is upside we didn’t pay for rather than the reason the deal pencils at all.

If you think one of these numbers is wrong, tell me.

Sources

• Cushman & Wakefield, Boise MarketBeat Multifamily Q2 2026 (vacancy, rent, absorption, submarket table)
• Marcus & Millichap / Institutional Property Advisors, Boise MSA Multifamily Q2 2026 (vacancy, absorption, concessions)
• CoStar, Boise MSA multifamily report, August 2026 (deliveries, absorption and vacancy forecast)
• Commercial Northwest, Treasure Valley Pipeline Report Q3 2026 (the 46-project permit verification)
• Construction Monitor via TOK Commercial, June 2026 (Ada and Canyon permit series)
• ALN Apartment Data via TOK Commercial, May 2026 (stabilized occupancy, concessions)
• Berkadia, 2026 Boise offering research (delivery forecast)
• NIST CHIPS program project record and Micron investor relations, August 2026 (fab investment, jobs, the research campus)
• COMPASS, April 2026 population estimate; US Census and Idaho Department of Labor (in-migration, employment)
• Intermountain MLS, August 2026 (Ada and Canyon resale prices)
• Freddie Mac PMMS and FRED series DGS10, September 2026 (mortgage and Treasury rates)
• Our own rent rolls and asset-management records for the two Boise properties shown, through 24 September 2026

Nothing here is an offer to sell or a solicitation to buy a security, and none of it is investment advice. Market figures come from the third-party sources named above and are current as of the dates shown.