Owners choose a procurement route believing they are choosing how much risk to transfer. What they are actually choosing is who carries which obligation, whether anyone's insurance stands behind it, and what happens to their recovery if the party holding it cannot pay.
Two, in practice. Traditional design-bid-build, where the owner's consultant designs and a contractor builds to that design. And design-build, where the contractor designs and builds, shading at its far end into EPC or turnkey. Construction management is a theoretical option in this market rather than a live one.
That last statement is worth defending, because international guidance treats construction management as a standard third choice. No UAE practice guide, market report or regulator I can find describes it as a route in use here. Four independent sources enumerating UAE procurement routes list traditional, design-build and EPC/turnkey, and none of them mentions it.
There is a structural reason. Dubai's Local Order No. 3 of 1999 builds its entire regime around an Engineer — a licensed consultant assigned to design and or supervise, carrying joint liability with the contractor and a statutory duty to the Municipality. A construction manager acting as the owner's agent, with the owner holding trade contracts directly and nobody carrying design or supervision responsibility in the statutory sense, has no obvious home in that structure. That is analysis rather than a rule anyone has published, but it is consistent with the market's silence.
The FIDIC suite dominates, and the 1999 editions remain the primary forms in use. There is no standalone FIDIC design-only contract. One caution before the table: the standard risk allocation is almost always adjusted through particular conditions, so knowing which book you are on tells you less than owners assume. Read the amendments before you read the form.
| Traditional (design-bid-build) | Design-build | EPC / turnkey | |
|---|---|---|---|
| Who designs | Owner's consultant | Contractor, to the Employer's Requirements | Contractor |
| Accuracy of the Employer's Requirements | Not applicable | Provision exists dealing with errors in them | Contractor carries it, save a short closed list |
| Unforeseeable physical conditions | Owner, on a foreseeability test | Owner, on a foreseeability test | Contractor, with no employer warranty on site information |
| Design obligation standard | Consultant: reasonable skill and care | Contractor: fitness for purpose | Contractor: fitness for purpose |
| Does professional indemnity insurance respond | Yes — negligence-based obligation, negligence-based cover | Not reliably; fitness for purpose is typically excluded | Not reliably; same exclusion |
| Who administers the contract | Engineer | Engineer | No Engineer; the Employer's Representative |
| Owner's flexibility to change | Highest | Reduced; changes priced by the design holder | Lowest |
| Decennial liability | Contractor and engineer, by function | Contractor carries both limbs | Contractor carries both limbs |
| Price premium for risk transfer | Baseline | Higher | Highest |
A strict obligation rather than a shared one. Under the FIDIC design-build forms the contractor must deliver works that are fit for the purposes defined in the Employer's Requirements. That is a fitness-for-purpose obligation, and it is liability without fault — effort and expenditure are not a defence.
The 2017 wording puts it directly: the works, when completed, shall be fit for the purposes for which they are intended, as defined and described in the Employer's Requirements or, where no purposes are so defined and described, fit for their ordinary purposes.
As one law firm puts it, no matter the level of effort or expenditure the contractor has applied when designing or constructing the works, if fitness for the purpose in the relevant sense is not achieved, they will be liable. For an owner that sounds like exactly what it wanted.
There is a sting in it, and it runs the other way. The purpose the contractor must achieve is whatever the Employer's Requirements say it is. A thin or ambiguous requirements document does not create a broad obligation — it creates a narrow one, plus a fallback to ordinary purpose. So the protection an owner gets from design-build is only ever as good as the owner's own specification. Which is the precise opposite of what "we've transferred the design" usually means when a developer says it.
Because professional indemnity insurance responds to negligence, and fitness for purpose is not a negligence standard. A Lloyd's broker states the position plainly: most professional indemnity policies cover the insured only where a claim arises out of professional negligence, and in the absence of negligence the insured is not typically protected against a contractual claim for failure to meet a fitness-for-purpose requirement. Such policies, the same source notes, will typically exclude fitness-for-purpose claims, or extend cover very narrowly.
Where an extension does exist, it may not reach far enough. One firm's analysis of design-and-construct policies observes that the cover is restricted to negligence, and that the fitness-for-purpose extensions offered provided cover only for implied obligations, not express ones. A FIDIC fitness-for-purpose obligation is an express one.
The obvious objection in this market is that the UAE has decennial liability and decennial cover, so the gap must be theoretical. The source answers it directly: a wider and bespoke contractual fitness-for-purpose obligation will be difficult to insure, even in countries with a decennial liability tradition in their legal systems.
Put the pieces together and the conclusion is uncomfortable. If the contractor's design fails and the contractor was not negligent, the contractor is liable and uninsured. The owner's recovery then depends entirely on the contractor's balance sheet.
Design-build has not transferred design risk. It has converted design risk into counterparty solvency risk — a different thing, priced differently, and managed differently. Which means the route decision is partly a covenant decision. Before choosing design-build, look at the contractor's accounts as carefully as its design credentials, because in the scenario the route is supposed to protect you from, the accounts are what you are relying on.
In three circumstances, set out by FIDIC itself in its own guidance for the EPC/turnkey form.
FIDIC's own remedy in those circumstances is to use the design-build form instead. Its contracts guide is characterised as concluding with a list of circumstances in which FIDIC definitely does not recommend, and warns against, use of the turnkey book.
The chairman of the committee that prepared the 1999 conditions repeats the same three exclusions and adds the commercial point: the more risks a contractor is required to bear, the higher the premium a prudent contractor must add to his price to cover his extra risk-taking.
Read those exclusions against a typical UAE development. Compressed programmes with short tender periods. Deep basements and piling in variable ground. An owner's team that fully intends to review the drawings. That is a large share of the market, and an owner selecting a turnkey form for such a scheme is using it against the express written warning of the body that wrote it.
There is an honest way to buy back the first exclusion. Early contractor involvement, paid for through a pre-construction services agreement, gives bidders the time and information to scrutinise the requirements properly. It costs money at the front end. It is considerably cheaper than pretending the tender period was adequate.
It depends which form you signed, and the difference between the two design-build forms is larger than most owners realise. The design-build form contains a provision dealing with errors in the Employer's Requirements. The EPC/turnkey form has no equivalent.
Under the turnkey form the contractor is deemed to have scrutinised the Employer's Requirements before the base date and carries responsibility for their accuracy — the employer bears no responsibility for any error, inaccuracy or omission of any kind in them, subject only to a short closed list. That list covers data stated in the contract as immutable or as the employer's responsibility, the definitions of the intended purposes of the works, testing and performance criteria, and information the contractor cannot verify. Everything else is the contractor's problem, which means everything else is priced.
Order of precedence matters too. In the design-build form the Employer's Requirements rank above the contractor's proposal. Owners assume their requirements govern; few have checked where the two documents sit relative to each other in the contract they signed.
On what happens when a specification is itself wrong, a 2017 UK Supreme Court decision is instructive. A contractor was held liable for failing to achieve a twenty-year design life despite having complied with the international standard the contract specified — a standard that itself contained a fundamental error. The court's reasoning was that no inherent inconsistency arises where the performance criteria prove impossible to achieve if the agreed design or specification is adhered to. Compliance with the specification was not a defence.
The owner-side reading of that is double-edged, and worth being honest about. Compliance is not a defence for the contractor. But the claim still has to be paid by somebody, and we are back to the balance sheet.
No. Under the Civil Transactions Law, in force since June, contractor and engineer are jointly liable for ten years from handover for total or partial collapse and for defects threatening the stability or safety of the structure. Any agreement excluding or limiting that liability is void. The route determines who sits in which chair. It does not change whether the exposure exists.
The allocation follows the function. An engineer whose role is limited to design is liable for design defects. An engineer who supervises shares responsibility for both design and execution. Under design-build the contractor occupies both limbs, because it designed and it built.
Three points owners routinely miss. Decennial liability reaches developers acting as contractor, or building through wholly-owned entities, and consortium members on joint ventures — so a developer that self-performs has not stepped outside it. Contractor recourse against subcontractors is not part of the statutory regime and has to be written expressly into subcontracts, where it runs on a shorter limitation period than the ten-year decennial clock. And claims run for three years from discovery, so a defect found in year nine remains claimable into year twelve.
One further point, which nobody appears to have published and which we offer as analysis rather than authority. The statute attaches liability to the function — designing, supervising — and not to the identity of the party who signed the appointment. So an engineer engaged by the contractor rather than by the owner should still carry joint and several liability by virtue of what it did. Moving the consultant onto the contractor's side of the table does not move the liability off the project. It only changes who the owner has to sue, and how easily.
The permission is clear. The framework is not. Dubai Law No. 7 of 2025 formally recognises turnkey contracts and permits contractors to undertake design, supervision and execution works — subject to a framework to be provided by Dubai Municipality. The law took effect in January of this year. As of the most recent tracking of UAE construction legislation, that framework has not been published.
The assessment from one international firm is worth quoting because it is unusually direct for published legal commentary: it is not yet clear how Dubai Municipality will apply these principles in practice.
Two things sit alongside that. Design and construction still cannot be combined on a single commercial licence — the possibility that they could be in future is described as a noteworthy development requiring coordination with the economic department, which tells you where the position stands today. And Local Order No. 3 of 1999, with its Engineer of record carrying joint liability and a statutory duty to the Municipality, survives to the extent it does not conflict with the new law. So the old construct and the new permission coexist, with the newer law prevailing where they clash.
Dubai Municipality operates a distinct registration route for design-and-build contractors, so the pathway plainly exists in administrative practice.
What does not exist, so far as I can establish, is any published requirement that the supervision consultant be independent of the design-build contractor. Nothing in the Municipality's material, in the 2025 engineering consultancy law, or in any named legal analysis states an independence rule or a conflict-of-interest rule for this situation.
The instruction for an owner is therefore straightforward and slightly unglamorous. Settle the supervision appointment in the contract, in writing, before award — who holds it, who pays for it, who can remove it, and to whom it reports. The regulation has not settled it for you, and the first project to find out how the Municipality applies the new framework should not be yours.
Sometimes. But the regional evidence cuts against the assumption, and an owner should see it before choosing on speed.
Design-build accelerates by overlapping design with construction. A dispute-analytics dataset covering 480 Middle East projects and around USD 626bn of capital expenditure found that late design information affected 32.1% of regional projects, and that incomplete design disrupted 26.3% of them against 17.7% elsewhere. Speed-to-build pressure appeared in 52.9% of regional projects against a global average of 47.6%. The firm's own conclusion is that overlapping design with construction inevitably leads to more design changes, delays and disputes as projects progress.
So the failure mode design-build institutionalises is the one this region already suffers from disproportionately. That does not make it the wrong route. It does mean the speed case has to be made project by project rather than assumed.
It is also worth discarding the numbers that circulate. The claim that design-build is around six per cent cheaper and a third faster traces to a study of US buildings published in 1998, and is repeated secondhand across the industry with no normalisation for market, project type or client sophistication. Its modern replacement is also US data, promoted by a design-build trade body. None of it describes this market and we do not use it.
What we do have is current market direction. Single-stage tendering remains dominant in the UAE. No respondents to the most recent regional survey reported intense competition, and a quarter reported a shortage of contractors with minimal competition. Owners are responding by introducing rate-fluctuation provisions on key materials and fully re-measurable contracts to attract bidders as competition declines.
That is the commercially important point. In a market where owners are actively removing risk from contractors in order to get tenders at all, transferring design risk through design-build runs against the prevailing current — and gets priced accordingly. The chairman of FIDIC's contracts committee makes the same point in principle. The UAE data suggests it is happening here now.
One honest note on the evidence. The same market authority described design-build use in the UAE as declining significantly in 2024 and as increasing in late 2025. Rather than pick the more convenient reading, take it as evidence that the market has not settled — which is itself a reason to decide the route deliberately rather than by convention.
Answer six questions before you choose a form rather than after. Most are about your own organisation rather than the contractor's, and three of them are FIDIC's own tests for when its turnkey form should not be used.
| Question | Why it matters | What the answer implies |
|---|---|---|
| Is the design complete enough to tender? | Design-build does not solve incomplete design; it relocates it into the contract | If not, either complete it or buy contractor input early and pay for it |
| Is there substantial underground work, or ground bidders cannot inspect? | One of FIDIC's three stated exclusions for the turnkey form | If yes, turnkey is against FIDIC's own guidance unless special provisions are made |
| Does your team intend to review the construction drawings? | Another of the three exclusions | If yes, you want design-build or traditional, not turnkey |
| Is the tender period long enough for bidders to scrutinise your requirements? | The third exclusion, and the one most often ignored | If not, extend it or buy the scrutiny through a pre-construction services agreement |
| Can you write Employer's Requirements precise enough to define the purpose? | The fitness-for-purpose obligation is only as wide as your specification | If not, design-build gives you far less protection than you think |
| Is the contractor's covenant strong enough to stand behind an uninsured obligation? | If the design fails without negligence, recovery depends on the balance sheet | If not, the design risk has not left the project |
The honest summary is short. Traditional procurement gives the owner control over design and flexibility to change, at the cost of holding design risk and coordinating two contracts. Design-build gives a single point of responsibility whose value depends almost entirely on two things the owner controls — the quality of the Employer's Requirements, and the strength of the contractor's covenant behind an obligation its insurance may not reach. Turnkey adds a further layer of risk transfer that FIDIC itself confines to a narrow set of circumstances, and prices accordingly.
And one thing does not change with the route. The decennial exposure sits on the project for ten years from handover whichever form you sign, and it cannot be contracted away. The route decides who carries it and who is capable of paying for it. It does not decide whether it exists.
The figures suggesting it is trace to a 1998 study of US buildings, repeated secondhand and never normalised for this market. What is established is that risk transfer is priced: the more risk a contractor is required to bear, the higher the premium it must add. Design-build may be cheaper on a given project, but not because a general rule says so.
The contractor, under a fitness-for-purpose obligation that applies regardless of the care it took. The complication is insurance: professional indemnity policies respond to negligence, and typically exclude fitness-for-purpose claims. If the design fails without negligence, the contractor is liable and uninsured, and recovery depends on its balance sheet.
Yes, but they are more expensive and less transparent than under traditional procurement, because the contractor holds the design and there is no independent designer to price the change against. Change flexibility is one of the things an owner trades away for a single point of responsibility.
You lose direct control of the design and retain control of the specification. That is why the Employer's Requirements matter so much: under the FIDIC forms the purpose the contractor must achieve is the purpose your requirements define, with a fallback to ordinary purpose where they are silent.
FIDIC names three circumstances: where tenderers have insufficient time or information to scrutinise the Employer's Requirements, where construction involves substantial underground work or areas tenderers cannot inspect, and where the employer intends to supervise closely or review most of the construction drawings. Its own recommendation in those cases is the design-build form instead.
Rarely without cost. The route is embedded in the contract form, the design status at award and the appointments already made. What is usually available instead is a variation of approach within the route — bringing a contractor in early under a pre-construction services agreement, or converting a provisional sum to a fixed package once the design catches up.
Bring the project. The first conversation is a working session, not a sales call.