Demand Is Not Keeping Pace With Supply. The Register Shows Both Numbers.

The purchase rate has fallen in every quarter this year

Off-plan purchases are not collapsing, and nobody should read this as a crash. They are declining, steadily, and the decline is visible at quarterly resolution rather than requiring a subtle reading. The first quarter of 2026 recorded 31,174 purchases, an average of 10,391 a month. The second quarter recorded 27,774, or 9,258 a month — down 10.9%. July and August together recorded 17,034, an average of 8,517 a month, down a further 8.0%. From the first quarter to now the monthly rate is down 18%. Month to month the series is genuinely volatile — May came in at 7,458 and June bounced back to 10,095 — so no single month is worth reacting to. The quarterly direction is consistent, and that is what makes it a trend rather than noise. August's 7,591 is a complete month: the register was read on 31 August.

This year's registrations alone outnumber this year's buyers

The supply side is the part the headline gets backwards. In 2026 the Land Department registered 372 projects carrying 88,926 units, with a declared value of AED 106.9 billion. Buyers in the same eight months made 75,982 off-plan purchases. So the units registered this year exceed the purchases made this year by roughly 17%. That comparison is deliberately conservative, and it understates the gap in two ways. It counts only projects registered in 2026, ignoring everything registered in earlier years that is still selling. And it counts purchases across the entire off-plan market, including stock from those earlier years — so the demand figure is inclusive while the supply figure is not. On a like-for-like basis the imbalance is wider than 17%. What it does not tell you is whether that supply will all arrive. Registration is a legal filing, not a commitment to build on schedule, and Dubai's history includes plenty of registered projects that were cancelled or delayed for years.

Three-quarters of it has not broken ground

Of the 372 projects on the 2026 register, 270 — 73% — are logged at 0% construction. A near-identical 272, also 73%, hold an open escrow account, which is the legal gate before units may lawfully be sold. Those two figures together describe the market accurately: most of what is being sold right now exists as a filing and a bank account rather than as a building. The completion dates cluster where you would expect. Of the units registered this year, 22,153 are filed to complete in 2028 and 27,571 in 2029, with 18,595 more in 2030. So the peak of this year's registrations lands in 2029, and a buyer signing today is joining a queue that arrives alongside roughly 27,000 other units from the same registration year — plus everything registered in 2024 and 2025 that completes then too. That is the number to hold on to when a developer talks about scarcity. Scarcity is not what the completion schedule describes.

What this changes, and what it does not

It does not make buying off-plan wrong. Demand of 75,982 purchases in eight months is a large, functioning market by any standard, and a declining rate from a high base is not the same as weakness. Prices in the transaction register have not broken, and this analysis makes no forecast about them. What it should change is how you price the exit. If you are buying to resell before handover, you are planning to sell into a market where the monthly buyer count has fallen for three consecutive quarters and where roughly 27,000 units from this registration year alone complete in your likely exit window. If you are buying to let, the same completion cluster is your competition for tenants. The practical steps are unchanged and worth repeating: confirm the escrow account number with the Land Department on the day you transfer money and pay into that account only; get the completion date and its penalty clause into the contract rather than relying on the filed date; and compare any quoted price against what has actually been paid in the same district rather than against the developer's own comparables.