Dubai off-plan sell-through 2026: how much stock is left

61% of the tracked book is buyable, 35% is already gone.

Of 2,337 tracked Dubai projects, 2,054 publish a sale status. Group them honestly and the split is close to even: 1,256 buyable (61%), 724 gone (35%), and 74 pre-market (4%) that have not opened yet. The headline pipeline number and the shoppable pipeline number are two different things, and the gap between them is roughly half.

Raw counts behind the groups, largest first: on_sale 805, out_of_stock 733, sold_out 546, unknown 365, booking_started 344, available 258, might_be_sold_out 72, waiting_for_sales_start 72, presale 24, start_of_sales 2, announced 2. Shares in the last column use 2,054 projects with a stated status as the base, never 2,337. Dividing by the full tracked count would silently treat every missing status as available stock and understate the gone share by about 4 points.

The size and shape of the pipeline itself, before any status filter, is the handover pipeline note. The same corpus cut by price is the Q3 2026 market note.

703 projects are sold out while the building is still going up.

Sold out does not mean finished. 703 of the 724 gone projects, 97% of that book, are still under construction, and 425 of those are not due until 2027 or later. Read the other way, 50% of completed projects are gone against 35% of those under construction. Stock does not survive to handover: it is cleared during the build, and what a buyer sees at completion is the residue.

Construction status carries the same 283 unknowns as sale status, so the three rows sum to 2,054 rather than 2,337. The 54 not-started projects are a thin base and are shown for completeness, not as a trend.

The practical consequence for anyone shopping a specific completion year: the closer the keys, the shorter the list. Handover 2026, 2027 and 2028 rank the communities holding each book.

The 2026 handover book is 56% gone; 2030 is 12% gone.

Sell-through falls in a straight line as the handover date moves out. Among projects delivering in 2026, 226 of 401 are gone and only 175 remain buyable. By 2030 the gone share is 12% of 78 projects, and the 2025 book, already delivered, is 82% gone. 1,040 of the 1,256 buyable projects, 83%, deliver in 2027 or later. Buying near-term stock means buying what nobody else wanted.

The price column moves the same way for a reason worth separating: later stock is newer stock, launched at higher asking prices. Which years the pipeline is actually loaded into is the 2026 to 2030 handover note.

The Valley is 73% gone, Meydan 13%.

Across the 35 Dubai communities holding 20 or more projects with a stated status, 1,643 projects in all, the gone share runs from 73% in The Valley to 13% in Meydan. The pattern underneath is age, not desirability: the tight communities are the established ones where most of the tracked stock has already completed, and the open ones are the reclamation and edge districts still selling their first cycle.

The six tightest and the six most open of 35 qualifying communities, highest gone share first. The completed column is there so the ranking can be read against build stage rather than instead of it. AED/sqft is the community median asking price per sqft across its own priced records, gone and buyable projects together.

Both ends have a benchmark page: The Valley and Meydan. The full set is under communities.

Majid Al Futtaim has cleared 67% of its tracked book; Azizi 16%.

20 developers carry 15 or more Dubai projects with a stated status, 1,016 projects between them. Majid Al Futtaim sits at 67% gone across 15 projects and Azizi at 16% across 58. Emaar, the largest tracked book at 202 projects, is 57% gone with 72 projects still open. A high gone share is a portfolio that has aged, or a launch cadence slower than its sales, and the two are not distinguishable from this field alone.

The twelve most sold-out of 20 qualifying developers. Counts are Dubai projects only, so a developer active in other emirates has a larger portfolio than the second column shows. Developers with fewer than 15 such projects are excluded, which removes most of the 676 names in the corpus and all of the single-project entities.

Who is still launching rather than clearing: developers by pipeline, and the portfolio-level view in the developer note.

Gone stock is asking AED 1,602 per sqft, buyable stock AED 1,938.

The cheap stock went first, but not in the way the numbers first suggest. Median asking price per sqft is AED 1,602 for gone projects (478 priced) against AED 1,938 for buyable ones (991 priced). Median entry ticket barely moves at all: AED 1.68M across 636 gone projects that publish one against AED 1.50M across 1,165 buyable ones. Hold construction stage constant and the psf gap narrows but survives, AED 1,603 against AED 1,929 among projects still building.

Two caveats that matter more than the gap itself. First, survivorship: a project's price per sqft is computed from its listed units, and a sold-out project keeps far fewer of them, 5,857 unit records across 724 gone projects against 36,170 across 1,256 buyable ones, so the gone median rests on whatever was left on the page. Second, mix: gone projects skew older and older launches were cheaper in absolute terms, which is most of the whole-book gap and some of the under-construction one. Neither number is evidence that a cheaper project sells faster.

The entry end of the market has its own sell-through arithmetic, already published: see finding six of the under AED 1M note for the gone share of the sub-1M book, and cheapest off-plan under AED 1M for what is still open there.

Only 74 projects sit in the pre-market queue.

74 tracked Dubai projects are flagged as not yet selling, 4% of the stated book. They cluster in Dubailand (13), Dubai South (10), DAMAC Islands (8), and DAMAC with 21 and Emaar with 14 account for 47% of the queue between them. Every one of them is due between 2027 and 2031, and 63 already publish a starting price despite not being on sale.

A detail that undercuts the label: all 74 of the 74 pre-market projects also carry a construction status of under construction. Either the site is moving before sales open, which happens, or the two fields were last refreshed at different times. We report the pair rather than picking the flattering one, and a queue this small, 4% of the stated book, cannot carry much weight either way.

Pre-market projects are the only part of the book where a published payment plan does not yet exist to compare. The terms on everything that is selling are in the payment plans note and lowest pre-handover payment plans. Whether a post-handover tail keeps a project on the market longer is taken up in the post-handover note.

Why we keep out_of_stock and sold_out apart.

The two statuses are not synonyms in the data even if they are in English. 442 projects read out_of_stock, with a median delivery year of 2027 and 436 of them still under construction. 225 read sold_out, median delivery 2027, 0 of them already completed. A further 57 read might_be_sold_out, a hedge the source itself publishes. Collapsing all three into one label would be tidier and would hide the fact that one of them mostly describes finished buildings and another mostly describes buildings still going up.

We group all three as gone for every share in this article because a buyer cannot transact on any of them today, and we say so each time. The distinction is preserved in the underlying data and in this table so the grouping can be re-cut by anyone who disagrees with it. What none of the three means is that the units have transferred: portal status is a marketing state, not a registry state.

What we do and do not republish from the source, field by field, is in the data policy.

283 projects, 12% of the book, publish no status at all.

The excluded block is not random noise and it is not small. All 283 of the 283 projects with no sale status also have no construction status, and 283 of them carry no listed unit at all, which is why not one has a price per sqft. What they do carry is a delivery date, 271 of them, and a starting price, 236 of them, median AED 1.96M. Nor is it a long tail of unknown developers: Emaar (37), DAMAC (32), Nakheel (20) lead it.

This is a coverage gap in one direction, project pages that were captured with headline fields but no inventory detail, rather than a class of project that behaves differently. Because the block is uniform in what it lacks, we exclude it wholesale rather than imputing a status, and every share above is stated against 2,054 rather than 2,337. If the missing 12% resembled the rest of the book, the gone share would land near 35% again; if it were entirely gone it would reach 43%. We do not know which, so we do not claim either.

Coverage counts for every field, and what each one is derived from, are in the methodology. Common questions about how the corpus is built are in the FAQ.

What sell-through here does not tell you.

Everything above is a published portal status observed once, on 2026-07-30. It is not a transfer record, so a project counted as gone may still have units held back for a relaunch, and a project counted as buyable may have nothing left that a buyer would want. Status is recorded per project, not per unit, so a tower with two remaining apartments counts the same as one with two hundred. And a single snapshot cannot measure velocity: nothing here says how long any project took to sell down.

Three narrower limits. Community and developer tables use minimum bases of 20 and 15 projects with a stated status, which excludes most small books entirely rather than ranking them on thin evidence. Gone-share differences between communities are dominated by how much of each district has already completed, so the ranking in finding four should be read with the completed column, never without it. And the price comparison in finding six sits on the listed units that survived, which is exactly the population most affected by selling out.

Where the data does not support a claim we say so rather than filling the gap. The same discipline applied to unit mix is the bedroom mix note.