The UAE R&D Tax Credit applies to tax periods starting on or after 1 January 2026. Security and cryptography work can be part of a qualifying project, but the label “cybersecurity” does not make a project eligible. Each project is assessed on its own: the work must be carried out in the UAE, meet five legal conditions, reach the AED 500,000 project threshold and receive pre-approval from the Emirates Research and Development Council. This page shows where genuine technical uncertainty sits in security engineering, where routine security work stops, and what evidence your team should keep.
Under Ministerial Decision No. 24 of 2026, an activity carried out in the UAE as part of an R&D project qualifies only if it meets all five conditions at once. The assessment is made having regard to the OECD Frascati Manual.
The OECD Frascati Manual names “the creation of new and original encryption or security techniques” as an example of software R&D, and excludes “the use of standard methods of encryption, security verification and data integrity testing”. That line runs through everything below.
The work aims to produce new findings. A detection method, cryptographic implementation technique or verification approach that is not already available in published research, standards, open-source libraries or vendor products can count. New to your company alone is not enough.
The work is built on original concepts or hypotheses. Configuring a known tool or following a vendor’s hardening guide is not creative in the R&D sense.
The outcome, or the means of reaching it, is not known in advance. “Will this countermeasure keep leakage below our target without breaking our latency budget?” is a technical uncertainty. “Will the audit pass?” is not.
The work follows a plan and a budget, with controlled experiments, baselines and recorded results. Failed experiments count when they are documented.
The results can be applied or replicated in other contexts, for example by another team using your test harness and environment records.
Each area below is an assessment direction, not a statement that the work qualifies. Whether a specific project meets the five conditions and the entity, cost, ownership and pre-approval conditions is a fact-specific question we assess before you apply.
Fitting ML-KEM or ML-DSA onto a specific class of device within memory, energy and latency limits, without introducing secret-dependent timing, can involve real uncertainty. Integrating an existing library to meet a policy requirement does not.
Developing a countermeasure for your own hardware or cryptographic implementation, and measuring residual leakage or fault resistance against an unprotected baseline in an authorised laboratory, can qualify. Certification testing and applying a known countermeasure do not.
Making encrypted or shared-data computation meet utility, latency and leakage targets under a written threat model, where library defaults fall short, is assessable. Configuring a standard mode is not.
Building a detection method that keeps its detection and false-positive rates when attacker behaviour, software versions or telemetry change (tested on later, unseen data against realistic baselines) can qualify. Rule tuning and day-to-day SOC work do not.
Designing a protocol, or a hybrid classical and post-quantum construction, and proving its security properties under a defined adversary through formal models and testing is assessable. Implementing a published protocol to specification is not.
A new verification method that establishes properties of firmware on your own or authorised targets, where existing tools cannot, can qualify. Running existing scanners does not.
Investigating isolation or attestation limits beyond what vendor features provide, through controlled experiments on your own infrastructure, is assessable. Enabling vendor features is not.
A new analysis method that measurably improves precision or recall on your own code and public benchmark suites can qualify. Running commercial scanners and triaging bug-bounty reports do not.
A new method for finding quantum-vulnerable cryptography in compiled or embedded code, with better accuracy than existing discovery tools, is assessable. Running a cryptographic inventory is not.
The Frascati Manual treats “new or substantially improved testing methods” as R&D, while “standard testing and benchmarking activities” are not. A new, reproducible way of measuring security or performance can be assessed; re-running an existing benchmark cannot.
The UAE Cyber Security Council’s National Encryption Policy sets requirements for key management and post-quantum cryptography, and the Council is running a national post-quantum migration programme. Moving to standardised post-quantum algorithms to meet that policy is compliance work using standard methods. It is important, but it is not R&D in itself.
A migration project can contain a separately defined R&D project, for example, when a standard algorithm cannot meet your device, latency or leakage limits and no published solution exists. That sub-project is assessed on its own, with its own plan, budget and records.
Algorithm status changes. We confirm it on the date of any assessment.
Final standards, published 13 August 2024
Round 3 under evaluation; one candidate withdrawn in July 2026
Most security work is operations, assurance or compliance. The following normally stay outside the credit:
penetration tests, red-team engagements and vulnerability scans;
SOC monitoring, threat-intelligence work and incident handling;
ISO/IEC 27001 and other compliance audits and certifications;
installing or configuring firewalls, endpoint, SIEM or encryption products;
cryptographic inventory and policy-driven migration to standard algorithms;
patching known vulnerabilities, bug-bounty triage and awareness training;
incubator, accelerator or competition participation, patents, funding rounds and partnerships, which are not evidence of R&D on their own.
What can change the answer is a separately scoped UAE project that resolves a technical question existing standards, libraries and tools cannot answer, run with a plan and budget and recorded as it happens.
Meeting the five conditions is necessary but not sufficient. The following must also be true:
Entity. The claimant is a UAE company (including a Free Zone Person) subject to Corporate Tax and/or Top-up Tax, or a foreign company with a UAE permanent establishment. Businesses that elect Small Business Relief are excluded.
Free zones. A Qualifying Free Zone Person must be subject to 9% Corporate Tax on income from the R&D activity, or to Top-up Tax.
Only work done in the UAE counts. R&D staff must be in the UAE and under your company’s direction and control. Subcontractors must be UAE-based, work in the UAE and not subcontract onwards.
Qualifying expenditure must reach at least AED 500,000 per project per tax period, before the staff-cost uplift.
Your company must bear the cost of the work and be entitled to the returns from its results.
Staff costs (with a 30% uplift), consumables (including non-capital licence fees), qualifying UAE subcontracting, cost contribution arrangements, and capitalised internally generated intangibles. Government-grant-funded portions, employee share options, intra-group recharges and capital equipment are not included.
Every project needs pre-approval from the Emirates Research and Development Council, in the form and timeline the Council sets. As of 21 October 2026, the Council’s pre-approval process is live. Applications are submitted through TAWWER at tawwer.gov.ae. We confirm the current position directly before any application.
Rates. The credit is calculated band by band: 15% on the first AED 1 million (at least 2 R&D staff), 35% on the next AED 1 million (at least 6), and 50% on the portion from AED 2 million to AED 5 million (at least 14). It is non-refundable and is used against Corporate Tax and/or Top-up Tax.
First AED 1 million
At least 2 R&D staff
Next AED 1 million
At least 6 R&D staff
AED 2 million to AED 5 million
At least 14 R&D staff
See what your credit could be worth
For security and cryptography teams, the strongest evidence typically includes:
the threat model and the security or performance target;
a dated review of published research, standards, libraries and vendor features (the baseline);
the hypothesis and why the outcome was uncertain;
the project plan and budget;
versioned code, configurations and an environment record (hardware, compiler, operating system, library versions);
run logs, measurements and results, including failed approaches and design changes;
laboratory or firmware-testing authorisations where relevant;
monthly R&D staff time by project and UAE location;
a cost breakdown by project, with grants and subcontracting identified.
Records must be kept for seven years and include written, visual and electronic records of objectives, processes, methodologies, experiments and findings.
No. Each project is assessed against the five conditions and the UAE entity, location, cost, threshold and pre-approval conditions.
Not in itself. Moving to standardised algorithms is compliance work using standard methods. A separately defined project that solves a problem the standards and libraries leave open may be assessed.
No. Routine testing is excluded. Developing and validating a new defensive testing method may be assessed as its own project.
Routine monitoring, analysis and incident handling do not. Building a new detection method whose results are uncertain and tested systematically against baselines may.
ML-DSA is a final standard (FIPS 204). Falcon (FN-DSA) has been selected and its standard is not yet final. We check NIST’s status on the date of assessment.
They can. The Frascati Manual includes R&D with negative results, provided the work met all the conditions and was recorded.
Not as qualifying activity. Only work carried out in the UAE counts, by staff located in the UAE or UAE-based subcontractors working in the UAE.
Only if your company is subject to 9% Corporate Tax on the income from the R&D activity, or to Top-up Tax. This needs company-specific advice.
An incubator place is not evidence of R&D. Expenditure funded by a Federal or Local Government grant is excluded from the claim.
No. It is non-refundable and is used against Corporate Tax and/or Top-up Tax. Unused credit can be carried forward subject to conditions.
Now. TAWWER is open. An authorised individual signs in with UAE PASS and requests organisation access, which your FTA-registered authorised signatory approves. Because there is no backdating, projects running in your current tax period should be submitted without delay.
Seven years after the end of the relevant tax period.
If your team is working on a security or cryptography problem with genuine technical uncertainty, we will review the project against the five conditions and the UAE requirements, show you where the routine boundary sits for your work, and specify the evidence you need to prepare for Council pre-approval.
R&D tax credits are all we do. Built in Manchester in 2017, AED 396M+ (£80M+) processed in UK claims. Now claiming for the UAE.
Expert insight on R&D tax credits, UAE compliance updates and innovation incentives, straight to your inbox.