Is Seattle's Cheap Office Sublease Window Closing? The Q2 2026 Answer

Four weeks ago I published this argument on Q1 2026 numbers and closed it by saying one quarter is not a trend, and that anyone quoting the data in September without re-checking it deserved what happened. Q2 2026 has now published. I re-checked. The trend held, and the story got sharper in a way I did not expect.

Sublease availability across Puget Sound fell to 4.72 million sf, or 11.7% of total available space (Kidder Mathews, Q2 2026). That is down from 4.97 million sf and 12.1% in Q1 (Kidder Mathews, Q1 2026). Roughly 250,000 sf drained out of the discount pool in three months, on top of the 278,000 sf that drained out the quarter before.

Two consecutive quarters. Same direction. Roughly half a million square feet of the cheapest built-out space in the region, gone from the menu.

Kidder Mathews describes 11.7% as the lowest sublease share since 2018 and reads it as an early sign of stabilization. Same number, different job description: for a tenant-rep broker it is a countdown, and the current reading is two quarters in.

One Q1 number did not survive. My July draft cited a $0.58 per sf increase in sublease asking rents. No primary source I trust republished that series for Q2, so I dropped it rather than carry it forward.

Because sublease space and direct vacancy are fed by different pipes.

Sublease space leaves the pool four ways. It gets leased, because it is usually the cheapest furnished product available and it moves first. It expires, because a sublease posted in 2022 against a prime lease ending in 2026 simply stops existing and rolls back to the landlord as direct space. It disappears with the building, as older inventory comes out for renovation or conversion. And the faucet refilling it has slowed, because the era of quarterly tech give-backs cooled while return-to-office policy firmed up.

Direct vacancy, meanwhile, keeps absorbing everything the sublease pool sheds. Regional office vacancy closed Q2 at 23.4%, a 30 basis point increase over Q1 (Kidder Mathews, Q2 2026). The Seattle CBD sat at 34.9%, up from 34.7% (Kidder Mathews, Q2 2026).

That is the whole point of this piece. Vacancy going up and the discount pool going down are not contradictory readings of one market. They are two different inventories moving on two different clocks.

In Q2 the two split almost to the square foot. East King County posted 248,731 sf of positive net absorption. Seattle Close-In posted negative 247,471 sf (Kidder Mathews, Q2 2026). One region, two nearly equal and opposite quarters.

The Eastside vacancy rate ticked down to 21.0% from 21.6% (Kidder Mathews, Q2 2026). Downtown, the average Seattle CBD Class A rent quote slipped to $40.29 per sf from $40.60 the prior quarter and $43.34 a year earlier (Kidder Mathews, Q2 2026). Landlords downtown have moved past defending the package and started conceding the face rate.

Large blocks on the Eastside are the scarce item. Downtown Bellevue was left with 124 full-floor options in Q2 2026, down 51 from the July 2024 peak, per Broderick Group's Eastside report. Databricks signed roughly 160,000 sf at Four106 in June, in the same new tower where Uber took 170,103 sf earlier in the year. Downtown got its own wins, including DocuSign at 116,000 sf at JPMorgan Chase Center, and technology tenants drove 42.5% of all Q2 leasing (JLL, via GeekWire, Q2 2026). But the Eastside delivers zero new office buildings in 2026 after a 4.9 million sf wave, so every large block that leases there is one that does not come back this year.

The sublease window is not closing everywhere at once. It is closing on the Eastside first.

Usually, yes. It remains the fastest path to built-out, furnished, shorter-term space, and it still clears well under direct economics once you price the landlord's required term and buildout.

The risks did not change, but they matter more as the shortlist shortens. The sublandlord's credit becomes your client's counterparty risk, which here means underwriting a tech balance sheet rather than a landlord's. Consent and recognition language decides whether your client keeps the space if the prime lease dies. Term mismatch is structural, since your client's horizon rarely lines up with the remaining prime term. As-is condition cuts both ways: free furniture, no rebuild.

None of that kills the trade. It means the remaining inventory needs more diligence at exactly the moment there is less of it.

Split the advice by submarket, not by headline.

Eastside requirement, plug-and-play discount space, 10,000 sf and up: move the search forward. Two quarters of drain plus 51 fewer full-floor options plus zero 2026 deliveries is not a market that rewards waiting. The best-fit space, meaning right size, right term, creditworthy sublandlord, exits the pool first. Six more months of patience buys a worse menu, not a cheaper one.

Downtown direct requirement: patience still pays, and the negotiating room now sits in the face rate as well as the package. Class A quotes have dropped two quarters running. The Fed held its target range at 3.50% to 3.75% on July 29, 2026, which keeps landlord capital stacks tight and keeps concessions on the table. Push on free rent, TI, and termination flexibility, and then push on the rate too.

Either way, date the advice out loud. "Sublease share hit 11.7% in Q2 2026, per Kidder Mathews, its second straight quarterly decline" is a sentence a client can act on. "The market is tightening" is a sentence a client has heard from four brokers this year.

Start with scale. 4.72 million sf is still an enormous amount of sublease space by any pre-2020 standard. The window is narrowing, not shut, and downtown it is barely narrowing at all.

Then the denominator. A share can fall because the numerator shrank or because the denominator grew. Back-solving Kidder's two published figures puts total available space at roughly 41 million sf in Q1 and roughly 40 million sf in Q2, so in this case both moved the honest direction. That is my arithmetic off their rounded figures, not their published number, and I would rather show the work than let someone else catch it.

Then the fragility. Seattle's economy specializes in single large give-backs. One AI-driven headcount reset at a major employer refills this pool in a quarter, and the same technology story driving Eastside absorption is the story that could reverse it.

Then the provider problem. Kidder Mathews put Q2 regional net absorption at negative 46,836 sf. JLL, via GeekWire, put it near positive 372,000 sf for the same quarter. Savills counted 2.4 million sf of Q2 leasing and 4.6 million sf for the first half, its strongest quarterly figure since early 2019, while Kidder counted 3.8 million sf for the half. The direction of travel is consistent across providers. The magnitude is not. Anyone quoting a single absorption number as market truth is quoting a methodology.

Sublease is far from over. The real signal is that the discount window became finite, measurable, and submarket-specific, and almost nobody in this market is measuring it.

Every quarter, the sublease figure sits in the same public brokerage PDFs everyone skims for the vacancy headline. Track one line: sublease square footage, share of total available, quarter-over-quarter change, split Eastside versus downtown. That is the entire system. It takes fifteen minutes when the Q3 reports drop in October.

Here is the practical version. For an Eastside client, your job this quarter is to compress the timeline and defend the compression with numbers, because "the pool shrank two quarters running and there are 51 fewer full-floor options than at peak" beats "I have a feeling this space goes fast." For a downtown client, your job is the opposite: slow the process down, run a real competitive set, and use falling Class A quotes as the opening rather than the concession package alone. Same broker, same week, opposite advice, both defensible because both are dated and sourced.

The deeper habit matters more than the numbers. Quarterly data has a shelf life of about ninety days, and a broker who quotes Q1 figures in August is telling the client something about their process, not about the market. I rewrote this piece from scratch rather than patch the old one for exactly that reason, and I dropped a stat I liked because I could not re-verify it. That is the standard. Timing authority is the cheapest credibility available to a tenant-rep broker right now, and it costs one download and fifteen minutes a quarter.

I build software for Seattle tenant-rep brokers, so treat that as an interested party talking. The tracking habit above works in a spreadsheet and always has.