The Office Recovery Is Real, and It's a Slow Grind: What CoStar's 2026 Forecast Means for Seattle Tenant Reps

Most coverage of the office market in 2026 is still arguing about whether the recovery is real. CoStar settled that question in February, and the more useful work now is reading what kind of recovery it is, because the answer changes how a tenant rep should advise a client on timing.

In February, CoStar revised its U.S. office forecast upward and called the cycle. Phil Mobley, the firm's national director of office analytics, put it plainly on America's Commercial Real Estate Show: "We observed peak vacancy for this current cycle. It finally occurred in the middle of last year and by our reckoning that was at 14.2% and we've now come down 10 to 15 basis points from that."

By the first quarter of 2026 that number had eased to about 14%, and CoStar expects it to hold near 14.1% through the end of the year before declining toward 13% by 2030. Mobley framed the revision against the firm's prior, gloomier call: "The revised forecast outlook takes a somewhat more optimistic long-term view than the previous forecast, which called for a continued rise in vacancy until late 2026."

The leasing data backs the turn. U.S. office tenants signed roughly 120 million sq ft of new leases in Q1 2026, the strongest quarter since mid-2018 and a 25% jump year over year. So the recovery is not a forecaster's hope. It is in the signed-lease numbers.

Here is the part that gets lost in the headline. Vacancy is falling in large part because the denominator is shrinking, not because demand is roaring back.

New office construction has effectively stopped. Mobley: "We've got a generationally low level of new supply getting started. The past six or eight quarters, it's been around 5 million square feet per quarter. And if you go back to the late 2010s, you know, it was 15 million per quarter. It wasn't all that uncommon to see a quarter more like 18 or 20 million." At the same time, obsolete space is leaving inventory faster: demolitions and conversions have run above 8 million sq ft a quarter for the last year and a half, and Mobley expects that to accelerate.

Net out the math and supply growth turns negative. A market can tighten on that alone, even with only modest demand. That is most of what is happening right now.

The composition of the leasing surge tells the same story. The Q1 record came from transaction count, not deal size. Mobley noted "the number of lease transactions executed during the quarter was the highest observed in a decade," while average lease sizes remain roughly 15% below pre-pandemic norms. More deals, smaller footprints.

Because the demand ceiling is structural, and it is low. Mobley's read on the next decade is sober: "We're actually looking, if you believe economic forecasters, we're looking at slower job growth, significantly slower job growth for the next 5 to 10 plus years as a result of changing demographics," with job growth around "half a percent a year for the next decade or so, which is a third to a quarter of the rate that we got used to seeing during the last cycle."

Office demand tracks office jobs. A third to a quarter of the old hiring pace is a hard cap on how fast space gets reabsorbed. Mobley adds the wildcard that cuts the same direction: "the recent productivity-driven divergence between economic growth and job growth could persist, especially if advances in AI enable firms to expand output with fewer employees. That could dampen demand for additional space."

So the recovery is real and slow at the same time, and those are not in tension. Supply scarcity pulls vacancy down while weak hiring keeps a lid on how high demand can climb. Mobley's own summary of the risk picture: the market "is priced appropriately to the level of risk that it presents," and the risks "are better understood now, but they are still there." For a tenant rep, the practical translation is to stop waiting for a V-shaped rebound that the demand math does not support.

Seattle turned the same corner, on the same supply-led mechanics, with a sharper local split. Puget Sound posted its first positive net office absorption since Q1 2022 in the first quarter of 2026, roughly +253,352 sq ft per Newmark, with vacancy and availability each falling about 90 basis points in the quarter. For the first time in the region's modern history, there are no office projects under construction, which means the supply-drought dynamic Mobley describes nationally is even more extreme here.

But the recovery is two-speed. The Eastside is tightening fast on AI and tech demand, with OpenAI expanding to about 272,000 sq ft at City Center Plaza in Bellevue and Uber taking a large block, while downtown Seattle vacancy still sits near 36.5%. The rent gap shows it: Bellevue CBD Class A is around $67.58/sq ft and rising, against roughly $51.01 downtown and slipping. Newmark's Yemas Ly draws the timing implication directly: "As the sublease discount pool shrinks and vacancy begins to compress, tenants that have been waiting on the sidelines may find the window for favorable deal terms narrowing, accelerating leasing decisions among mid-sized users through the balance of 2026."

This is where the national forecast becomes advice, and where a generic "the market is recovering" headline actively misleads. The right call is not one call. It is two, split by submarket and by block size.

For Eastside clients and anyone chasing large contiguous Class A, the data says move sooner. There is no new construction coming, the discount pool is shrinking, and large premium blocks are the first thing to disappear in a supply-starved market. Mobley flagged the national version of this: "if you're looking for premium Class A space that's well-leased and you're looking for a lot of it, say 100,000 plus square feet, there's not a lot of it left." Waiting for a better deal that the supply math says is not coming costs the client the space, not just the price.

For downtown Seattle clients in smaller footprints, leverage holds longer. With vacancy near 36.5% and rents soft, a tenant can still negotiate hard on concessions. The risk there is not running out of space. It is signing a long term right as office-to-residential conversions and demolitions quietly pull the weakest inventory out and lift the floor under the best of what remains.

The point is that "is now a good time?" has no single answer in a two-speed market, and the broker who can show a client the supply-and-demand picture for their specific submarket and block size is giving advice the client cannot get from a national headline or a chatbot.

We build BrokerHQ around the idea that a tenant rep's edge is defensible advice, and a turning, two-speed market is exactly when that edge pays. A national vacancy number does not tell a Bellevue client whether to sign this quarter or next. The answer comes from the comps, the live requirements, and the submarket supply picture the broker is already tracking, organized so the recommendation is traceable to its source.

One honest boundary, the same one we hold in every post: software does not make the timing call. The broker does. A good system makes the call faster to support and harder to argue with, by keeping the data behind it in one place. Some of that is live today and some is on the roadmap, and we will not pretend otherwise.

A fair operator names the other side. Three real objections.

First, CoStar is a single forecaster, and forecasters miss. The 14.2% peak call could prove early if a recession or a fresh wave of layoffs pushes vacancy back up. Seattle already shows the warning sign: metro unemployment hit 5.4% in February 2026, above the national rate, and Q1 WARN notices jumped 56% quarter over quarter, dominated by Amazon. A demand shock would reopen the tenant-favorable window the supply story is closing. The hedge is to treat "move sooner" as conditional on the client's own risk tolerance, not a blanket rule.

Second, the supply-drought thesis can be oversold. No new construction today does not mean no space tomorrow; sublease blocks, conversions that stall, and tenants giving back space can all add availability faster than a forecast assumes. The discipline is to watch the actual sublease and give-back data in each submarket, not to assume scarcity everywhere because starts are low nationally.

Third, none of this is uniform, and treating a national number as local advice is the exact mistake this piece warns against. The 14% national figure is close to meaningless for a downtown Seattle client sitting in a 36.5%-vacancy submarket. The cycle frame is useful only when it is grounded to the client's block, submarket, and timeline.

CoStar called the cycle, and the call holds up: peak vacancy is behind us, leasing is at an eight-year high, and the recovery is real. It is also slow, supply-led, and uneven, which is a very different thing from the rebound the word "recovery" implies. For tenant reps, the takeaway is not to wait for a bottom that has already passed on the supply side, and not to rush a client whose downtown submarket still hands them leverage. It is to give timing advice that is specific to the submarket and the block, backed by data the client can see. That is the job the cycle just made more valuable.