Guide

Gulf AI data residency: the architecture questions buyers should ask

AI data residency in the Gulf is three separate questions that buyers usually collapse into one. First, where the law actually requires data to stay. As of September 2026, and as a reading of the instruments rather than legal advice: in Saudi Arabia, the UAE and Qatar the personal-data statutes regulate cross-border transfer rather than imposing general localisation, and the hard in-country rules sit in sectoral instruments such as the Saudi government cloud policy, UAE health and banking rules, and the Qatar Central Bank's cloud regulation. Second, whether the cloud region runs the model at all: a live region in-country can host the AI platform while serving few or none of the models in-region. Third, where inference actually happens, because data at rest is not data in use, and the managed routes to frontier models from Gulf regions are mostly globally routed by design. A buyer who asks only whether there is a region in-country gets a compliant-looking architecture that quietly processes prompts on another continent.

AI data residency. The set of architecture decisions that determine where data governed by a given jurisdiction is stored, whether the in-country cloud region runs the model that processes it, and where inference on it actually takes place.

Written by Tayyeb Mahmud, founder and Principal Architect. Claude Certified Architect – Professional · LinkedIn profile of Tayyeb Mahmud

Checked 30 September 2026. The Saudi, Qatari, Kuwaiti, Bahraini and Omani texts and Google Cloud's region, model and Data Boundary pages behind the sections added on this date were read on the publishers' own sites; the rest of the page was checked through September 2026. Laws and region listings change, so confirm against the sources below before relying on them. This page is not legal advice.

Three questions, and only one of them is legal

Three different things get called data residency in the same sentence, and they have different answers. The first is legal: what an instrument actually obliges you to keep inside the country. The second is infrastructural: whether the cloud region you chose runs the model you want. The third is operational: where the computation on your prompt happens, which is not settled by where your database sits.

Buyers usually ask a version of the first question, accept a version of the second as the answer, and never ask the third. This page covers Saudi Arabia, the UAE and Qatar in the most depth, because that is where the pattern causes the most damage, and gives Kuwait, Bahrain and Oman a shorter section each.

The legal question: Which instrument binds you: the personal-data statute, a sector regulator's rulebook, a government cloud policy, or a free-zone regime. They give different answers, and the sectoral ones are the strict ones.
The region question: Whether a live region exists in the country, and separately whether that region serves the model you intend to call. As of September 2026 those are not the same fact in any Gulf country.
The inference question: Where the prompt is processed and the output produced, as distinct from where the data is stored. Vendor documentation answers this explicitly, and the answer is often not the region you selected.

Where does Gulf law actually require data to stay?

Start with what the personal-data laws do not say. Saudi Arabia's PDPL, the UAE's Federal Decree-Law 45 of 2021 and Qatar's Law 13 of 2016 all regulate cross-border transfer: conditions, safeguards, assessments. None of them contains a general requirement that personal data be kept inside the country. If your residency requirement came from a summary of one of those laws, it is probably stricter than the law.

The binding localisation rules are sectoral, and where they apply they are strict. What follows is a reading of primary instruments as of September 2026, not legal advice, and each of them should be re-verified against the sources listed below and with your own counsel before it becomes a design decision.

US health data runs on a different principle again: HIPAA turns on which organisations handle patient data and under what contract rather than on where it is stored, and our guide to HIPAA-compliant AI sets out what that asks of an AI system.

Saudi government data: The MCIT Cloud First Policy states that all data in both the Government Cloud and the Commercial Governmental Cloud should be located geographically inside the borders of Saudi Arabia. The PDPL, enforced by SDAIA since September 2024, instead governs transfer: standard contractual clauses, binding common rules or accredited certification, with a transfer risk assessment in defined cases, no general prior approval requirement, and no adequacy list published as at September 2026.
Qatar, which runs the other way: The MCIT Cloud First Policy P005 classifies government data from C0 to C4, works through an endorsed provider list, and keeps the C4 tier off cloud entirely. The PDPPL pushes in the opposite direction: Article 15 forbids a controller from restricting cross-border flow except where the processing would breach the law or risk serious harm, and it sits in the penalty schedule. Qatar is the Gulf jurisdiction whose data-protection statute argues against localisation rather than for it. One sectoral exception is the Qatar Central Bank, whose cloud computing regulation, in force since April 2024, requires licensed entities to process personal and financial information within Qatar only, with QCB approval before any cloud arrangement.
UAE health data: Federal Law 2 of 2019 provides that health data related to services provided inside the state may not be stored or processed outside the UAE except under a health-authority resolution, and it applies including in the free zones. The exemption instrument is Ministerial Decision 51 of 2021; we have not been able to read its operative text, so treat the exception categories as something to confirm with counsel rather than to design around.
Abu Dhabi health, which is stricter again: ADHICS V2, issued by the Department of Health in May 2024 and effective from August 2024, requires cloud hosting physically within the UAE with none of the environments, infrastructures or systems outside the country, including backup and disaster recovery, and states no exemption route. A companion control catches encrypted, anonymised and pseudonymised copies, and another prohibits offshore analytics and offshore remote support. For an Abu Dhabi health estate this is the binding constraint, not the federal law.
UAE banking and payments: The CBUAE Outsourcing Regulation for banks requires that the Master System of Record, which includes all Confidential Data, is continuously maintained and stored within the UAE. The Consumer Protection Standards extend consumer and transaction data holding to all licensed financial institutions, and the stored value facility and retail payment regulations carry comparable requirements.
The UAE's three coexisting regimes: Federal, DIFC and ADGM. The federal PDPL has been in force since January 2022 but its executive regulations have still not been issued, so the compliance clock, the penalty schedule and the adequacy list are not operative, and in June 2026 the Cabinet approved a new AI and Data Authority to absorb the federal Data Office. DIFC entities sit under DP Law 5 of 2020 and Regulation 10; ADGM under its own 2021 regulations. Free-zone entities are not under the federal PDPL. The health data law reaches them anyway.

Saudi government data: which AI workloads can run where

The Saudi line above decides whether a government system can use a hosted model at all, so it is worth reading the policy itself rather than the summary. The MCIT Cloud First Policy, in its October 2020 version, applies to government entities except the Saudi Arabian Monetary Authority, as the policy calls it, and the bodies primarily responsible for security and defence, naming as examples the Ministry of Defense, the Presidency of State Security, the Ministry of Interior and the NCA. Government-owned companies and the private sector are only encouraged to write similar policies of their own.

It works with four data levels and two kinds of cloud. A Government Cloud Service Provider is NIC's government community cloud, or any commercial provider that meets the NCA's requirements for all four levels. A Commercial Governmental Cloud Service Provider meets them for open and restricted data only. The policy states that all data in both should be located inside the borders of Saudi Arabia, and it describes no third route: nothing in it covers sending government data to a model API outside those two clouds.

The cybersecurity side has changed since. The NCA's Cloud Cybersecurity Controls, CCC-2:2024, which the NCA's page says were updated for data localisation in July 2025, deleted the two subcontrols that required providers to deliver cloud services from inside the Kingdom, covering storage, processing, disaster recovery, monitoring and support. Its change log says data localisation controls now sit with the National Data Management Office at SDAIA, and that entities must refer to that office before taking any action on localisation. A question about where restricted data may be processed therefore goes to the NDMO, not to a reading of the NCA controls.

Saudi Cloud First Policy (October 2020): each data level, where the policy hosts it and what that leaves for an AI workload. Read 30 September 2026; a reading of the policy, not legal advice.
Dimension	Where the policy puts it	What that leaves for an AI workload
Levels 1 and 2 (“top secure” and “secure”)	The Government Cloud: NIC, or a commercial provider cleared by the NCA for all four levels. Internal hosting, preferably a private cloud, only with approval from the cloud computing adoption team at Yesser.	A model hosted inside that environment. A managed model API outside it is not a route the policy offers.
Level 3 (“restricted”)	A Commercial Governmental Cloud, with approval from the National Data Management Office, which may decide the data belongs at level 1 or 2 instead.	Get the office's decision in writing before a restricted data set reaches any model, and design the system so it still works if the answer is the Government Cloud.
Level 4 (“open”)	A Commercial Governmental Cloud directly, where its security, technical and commercial requirements are met.	The widest option, and still inside the Kingdom. A globally routed model endpoint is neither of the policy's two cloud types, so open data does not by itself make one acceptable.

For a supplier to a Saudi government entity, that turns into four questions to answer before anything is built: what level each data set is classified at, and by whom; which provider holds which level; whether the NDMO has ruled on the restricted sets; and where the model runs relative to all three. The architectures further down this page are the ones that survive those answers.

Qatar, Kuwait, Bahrain and Oman: what changes for AI data residency

The four smaller GCC jurisdictions answer the transfer question in four different ways, and the answer decides the architecture. Qatar's statute argues against localisation while its central bank requires it. Kuwait's regulator keeps its two most sensitive data tiers in the country. Bahrain keeps a list of the countries data may go to. Oman makes the data subject's explicit consent the default condition for sending personal data abroad. Each entry below gives the instrument, the transfer rule and what to have on file before a model call carries personal data out of the country.

Qatar: Law No. 13 of 2016 forbids a controller from limiting cross-border data flow unless the processing breaches the Law or may seriously damage the data or the individual's privacy (Article 15). The binding local rule is sectoral: the Qatar Central Bank's Cloud Computing Regulation requires a licensed entity to process personal and financial information within Qatar only (21.4), and to get QCB approval before entering a cloud arrangement (21.5). For a QCB-licensed firm that settles the model question, because neither Google's nor Azure's Doha region serves a chat model as of September 2026, so generation on that data runs on compute you host in Qatar or on a Qatari provider's capacity. Keep the QCB approval and a record that no step of the pipeline, embeddings and logs included, runs outside Qatar.
Kuwait: The regulator is CITRA, and its data rules are regulations issued under its own law, Law No. 37 of 2014. The Data Classification Policy (v2.3) sorts data into four tiers, and its third tier, private sensitive data, gives medical records and internal project reports as examples. The Cloud Computing Regulatory Framework (v2.4) applies to public-sector subscribers and to private-sector subscribers who host government data (1.4), and its classification chapter separately bars private-sector subscribers from storing individuals' third- and fourth-tier personal data with a provider outside Kuwait (3.2.1.2.2). It says third- and fourth-tier data may not be hosted or stored outside Kuwait for any purpose, “whether temporarily or permanently” (4.2.1.1); data below those tiers may use providers abroad (3.1.4.1). The Data Privacy Protection Regulation adds that a provider must notify the data owner when it intends to transfer personal data out of Kuwait, record the destination country and report a breach to CITRA within 72 hours. Read at its word, a prompt carrying third-tier data to a model hosted outside Kuwait is temporary storage abroad, and Google's regions page lists no Kuwait region. Keep the tier of every data set the system reads.
Bahrain: Law No. 30 of 2018 prohibits transfer outside the Kingdom except to a country on the Personal Data Protection Authority's record of adequate jurisdictions, or with its authorisation case by case (Article 12), with exemptions that include the data subject's consent (Article 13). The record, attached to Ministry of Justice Order No. 42 of 2022, lists 83 countries and territories, among them the UAE, Saudi Arabia, Kuwait, Oman, the UK and the US; Qatar is not on it. A transfer by contract to a recipient outside the listed countries needs prior authorisation with the contract attached. For AI the consequence is about routing: an endpoint pinned to a listed country fits the record, while a route that may process in any commercial region, which is how AWS describes Bedrock's Global routing, cannot be shown to stay on it. Keep the destination country of every model endpoint, and the authorisation where one is needed.
Oman: The Personal Data Protection Law issued by Royal Decree 6/2022 is enforced by the Ministry of Transport, Communications and Information Technology. Article 23 allows transfer outside Oman under the Executive Regulation's controls, and Article 29 attaches the Law's highest fine band, 100,000 to 500,000 Omani rials, to breaching it. The Executive Regulation (Ministerial Decision 34/2024) requires the data subject's explicit consent before a transfer unless it fulfils an international agreement involving Oman or the data is anonymised beyond linking to anyone (Article 37), an adequate level of protection at the recipient (Article 38), and a written evaluation of the transfer that the Ministry may ask to see (Articles 39 and 40). For an AI system that means consent captured before a prompt with personal data leaves the country, or anonymisation that meets Article 37, with the evaluation on file either way.

Personal data leaving Qatar, Kuwait, Bahrain and Oman, from each regulator's own text read 30 September 2026. Not legal advice.
Dimension	Instrument	Transfer rule	On file before a model call
Qatar	Law No. 13 of 2016; QCB Cloud Computing Regulation for licensed entities	Free flow under the statute; personal and financial information stays in Qatar for QCB-licensed entities	QCB approval of the cloud arrangement, and a map showing every pipeline step in Qatar
Kuwait	CITRA Data Privacy Protection Regulation; Cloud Computing Regulatory Framework v2.4	Third- and fourth-tier data may not be hosted or stored abroad, even temporarily	The tier of each data set, the destination country of each transfer, and notice to data owners
Bahrain	Law No. 30 of 2018; Order No. 42 of 2022	To the 83 listed countries and territories, otherwise by authorisation or an Article 13 exemption	The destination country of every endpoint, and any authorisation
Oman	Royal Decree 6/2022; Ministerial Decision 34/2024	Explicit consent unless an international agreement applies or the data is anonymised	The consent record or the anonymisation method, and the Article 39 evaluation

A live region in the country does not mean the models run in it

This is where most Gulf residency architectures break, and they break quietly, because the region genuinely exists and the platform genuinely runs in it.

Azure Qatar Central in Doha is live and hosts Microsoft Foundry. As of September 2026 it serves no Azure-sold models in any deployment type: the region does not appear in Microsoft's model region tables at all, Foundry Models is marked preview there, and the Foundry Agent Service is not offered. A team that reads “Foundry is live in Doha” as “we can run a model in Doha” has mistaken the platform's presence for the model's.

AWS shows the adjacent version of the same trap. Bedrock runs in Bahrain me-south-1, and the number of models it serves In-Region from there is zero. Everything reachable from that region is reached over Bedrock's Global routing.

The footprint itself needs the same care, and all of it is dated. As of September 2026 AWS runs me-central-1 in the UAE and me-south-1 in Bahrain, with a Saudi region announced and not live. Azure runs UAE North in Dubai and Qatar Central in Doha; UAE Central in Abu Dhabi is a restricted-access disaster-recovery region with no listed products, Qatar Central has no paired region, and Saudi Arabia East is not live. Google runs me-central1 in Doha and me-central2 in Dammam, which is Dammam rather than Riyadh and worth checking against whatever your diagram says, and has no UAE region and none announced. Announced capacity is a plan; a plan is not a control.

Google Cloud's Doha region: what runs in me-central1 for AI

Of the Gulf regions, Doha shows the widest gap between the platform being present and the models being present. Google's own tables, updated in the last week of September 2026, are specific about what an architect gets in me-central1.

The platform is there. Vertex AI in Doha lists custom model training, batch and online inference, Model Registry, Vector Search, Pipelines, TensorBoard, Vizier and Explainable AI. It does not list AutoML, Model Monitoring or Experiments, and Google's page of supported locations for agents does not list Doha at all.

The managed models mostly are not. Google's per-region model table lists four generative entries for Doha: three embedding models (Gemini Embedding, Embeddings for Text and Embeddings for Multimodal) and Virtual Try-On. No Gemini chat model is listed, and no partner or open model, so Claude, Mistral, Grok, Llama, Qwen and gpt-oss are not served there as managed models. Dammam, me-central2, lists the same three embedding models and nothing else from those tables. None of Google's per-model ML processing commitments is for a Middle East location either, so even the embeddings in Doha carry no published in-region processing commitment.

The hardware is missing too. Vertex AI's accelerator table lists nothing for me-central1, and Compute Engine's GPU locations list no Doha zone. Self-hosting an open-weight model in Doha today means running it on CPU, which in practice means a small model and modest throughput.

What Google offers in Qatar is control over where resources live. The Assured Workloads Qatar Data Boundary, in the free tier, sets data location controls to Qatar-only regions and lists Generative AI on Vertex AI among its supported products. It also routes support cases to global support personnel. Pinning a project to Doha pins storage and the services that exist in Doha; it cannot make a model appear there.

Google Cloud me-central1 (Doha) for AI workloads, from Google's own documentation read 30 September 2026. Region listings change month to month.
Dimension	Listed in Doha	What it means for the design
Managed chat and multimodal models	None: no Gemini, partner or open model in Google's per-region tables	Generation runs in another region, or on the global endpoint, which Google says may process a request in any location
Embeddings	Gemini Embedding, Embeddings for Text and Embeddings for Multimodal	Retrieval can be built in Doha, without a published in-region processing commitment
Custom training and inference	The Vertex AI base platform, with Model Registry, Vector Search and Pipelines	You can host your own model in the region
GPUs	None, in Vertex AI's accelerator table or in Compute Engine's GPU zones	A self-hosted model runs on CPU; Dammam lists L4 GPUs
Agent runtime, AutoML and Model Monitoring	Not listed	Those parts of the stack run in other regions
Qatar Data Boundary	Available, in the free tier	Pins where resources live; support is global, and Google Cloud MCP servers sit outside its residency controls

That leaves a workable pattern for a Qatar workload. Keep the system of record, the vector index and the embeddings in Doha, and decide data set by data set whether generation may happen elsewhere. Where personal data may leave, and under Article 15 of Qatar's statute it generally may, send redacted context to a model in another region and record which region. Where it may not, as for a QCB-licensed entity, the model has to run in Qatar, on your own compute or on a Qatari provider's.

Data at rest is not data in use

Even where a model is reachable from a Gulf region, the vendors say plainly that reachability is not residency. Their own documentation is the strongest evidence a buyer has on this, and it is worth reading in the vendors' own terms. Who may use a model at all is a separate test again: Anthropic restricts use of Claude by location and by ownership, which our analysis of Anthropic's 2026 Usage Policy update covers.

AWS separates the two routes in Bedrock. Of In-Region inference it says your requests never leave the Region you specify. Of Global inference it says Bedrock routes your request to a supported commercial Region worldwide, and that data may be processed in any commercial Region. In the Gulf, as of September 2026, exactly two models are In-Region: Amazon Nova Pro and Nova Lite, in me-central-1 only. Every Claude, GPT and Grok model listed in me-central-1 and me-south-1 is Global-only. There is no Middle East Geo tier, the geographies offered being US, EU, Japan and Australia. AWS also notes that input prompts and output results may be stored in the opt-in Regions for abuse detection purposes, which is storage, not merely transit.

Microsoft's own deployment-type documentation draws the same line: data stored at rest remains in the designated Azure geography, but Global deployment types may be processed in any Azure region. There is no Middle East Data Zone. In UAE North, the deployment types that are genuinely in-region cover four models as of September 2026: three text-embedding models and whisper, and no chat model. The only in-region route to a frontier chat model there is Regional Provisioned Managed, which requires purchasing committed throughput capacity. Microsoft also describes the order in which deployment types reach a region: Global first, Data Zone next, single-region last, with no guaranteed date for the last of those.

Google states it as a warning rather than a footnote: endpoints do not guarantee data residency or in-region ML processing. Its machine-learning processing residency commitments cover thirteen locations, and not one of them is in the Middle East. Assured Workloads offers a Qatar Data Boundary, and a KSA Data Boundary with Access Justifications for customers billed outside the Kingdom. Both set data location controls, which pin where resources live, and both state that they give no residency controls for data in use or in transit through Google Cloud MCP servers. Neither changes the ML processing list.

Every figure in this section is a snapshot. Vendor model catalogues change frequently and without notice, so re-check each claim against the vendor documentation in the sources below before it turns into an architecture decision, and record the date you checked it.

What the workable architectures actually are

Six paths survive contact with the facts above: five architectures and one procurement gate. None of them is free, and the trade-off in each case belongs in front of whoever signs it off rather than in an appendix. All of the vendor and legal detail here is as of September 2026 and should be re-verified before it is relied on.

Self-host open-weight models in-region: SageMaker AI runs in both me-central-1 and me-south-1, and Azure Machine Learning runs in UAE North and Qatar Central. The constraint is accelerators rather than software: on Google Cloud, Dammam lists L4 GPUs only and Doha lists no Vertex AI accelerators at all, and Vertex AI in both is the base platform, without AutoML, model monitoring or agent runtime, which sit in Singapore. You also take on the model operations you were buying a managed service to avoid.
Buy committed capacity for an in-region managed model: On Azure UAE North, Regional Provisioned Managed is the route to a frontier chat model that stays in the region. It is a reserved-capacity purchase with a commercial commitment attached, so it belongs in the business case at the start rather than in a late architecture review.
Use the two models that are actually In-Region: On AWS in the UAE, Amazon Nova Pro and Nova Lite are the In-Region options. For workloads those models handle well, that is a real architecture. For workloads they do not, it is an honest constraint rather than something to engineer around.
A sovereign or in-country platform, with the live-versus-announced line held: Core42's sovereign public cloud on Azure in the UAE is live per its own product pages, scoped to open and confidential or sensitive tiers. Ooredoo's sovereign AI cloud in Qatar has been live since July 2025, per Ooredoo. In Saudi Arabia, AMD reports MI355X capacity live as of 31 August 2026 serving HUMAIN's customers, while the AWS and HUMAIN AI Zone is a build-out its partners plan to 2028, and Stargate UAE's first 200MW is expected rather than confirmed. Buy against what is running, on the provider's own published status rather than the press release.
Accept cross-border inference where the law permits it: For personal data outside the sectoral regimes, the transfer route is frequently lawful: Saudi transfers on standard contractual clauses or binding common rules with a transfer risk assessment where required, and Qatari law that positively discourages restricting flow. This is the option most often ruled out by assumption rather than by an instrument. Ruling it in means writing down which instrument applies to which data set, which is the work most residency requirements skip.
The procurement path, in Saudi Arabia particularly: Customers billed in Saudi Arabia buy Google Cloud through CNTXT, per Google's own Dammam region documentation, and the Dammam region holds a CST Class C licence assessed by the National Cybersecurity Authority. That is a procurement route rather than an architecture, and discovering it late costs weeks.

The questions to put to any vendor or consultancy

Every question below has a factual answer somebody can look up. A supplier who cannot answer them has not done the work, and a supplier who answers them confidently without a source has done something worse. Take the answers in writing, with the date attached, because all of them expire.

Which instrument are we complying with, by name?: Not “Gulf data residency”. The PDPL, ADHICS V2 control CS 1.2, CBUAE Article 6.1, Cloud First Policy P005: the answer should be a named clause, and it determines everything downstream.
Which model, in which region, on which deployment type?: All three together, on the vendor's own terms: In-Region or Global on Bedrock; Global, Data Zone, Regional or Regional Provisioned Managed on Azure. A model name and a region name with no deployment type is not an answer.
Is the region running the model, or only the platform?: Ask for the line in the vendor's model region table, not a confirmation that the region is live. The two questions have had different answers in Doha and in Bahrain as of September 2026.
Where is inference performed, and where are prompts and outputs retained?: Processing location and retention are separate answers. Abuse-detection retention can put prompt text in a region your architecture diagram does not show.
What happens to backups, disaster recovery, logs and support access?: The Abu Dhabi health standard names backup and disaster recovery explicitly and prohibits offshore remote support, so an in-country primary with an offshore secondary fails it. Most residency designs are broken by their second copy rather than their first. Monitoring is a second copy too: our analysis of Claude Platform OpenTelemetry traces shows how switching on full content sends prompts and responses to an outside tool under its own retention.
Is this capacity running today, or announced?: With a date, a source and the name of whoever confirmed it. Sovereign AI announcements in the Gulf run several years ahead of the capacity they describe.
How would we evidence this to a regulator?: A region recorded per store, a deployment type recorded per model call, and logs that show what actually ran. If proving it depends on a supplier's recollection, it is not evidence. A governance platform can hold those records, and our guide to AI governance software sets out what such platforms record and what they cannot enforce. A residency rule is one clause in a wider AI policy, and our AI governance framework template maps each clause to the control that enforces it.

Where we have had to make these decisions

We hold end-to-end technical ownership of the live production estate of a government-accredited EdTech in the Middle East. Its database was replatformed onto Postgres in Google Cloud's Doha region, me-central1, to meet Gulf data-residency requirements. That decision is the basis on which we describe the trade-offs above: the set of choices we had to make, document and defend.

The separations on this page are the ones that engagement forced. Where the data sits became a store-by-store decision with a named region against each. Where the models run stayed a separate decision, taken against the vendor documentation set out above rather than against the region's marketing. What we bring to a residency question is architecture and evidence, and the discipline of checking the vendor's own words on the day the decision is made.

For engineers: endpoints, region pins and GPUs in Doha and Dammam

The region facts on this page in configuration terms, from Google's and AWS's own documentation. Each item is something you can check in a project, a policy or a log.

Regional endpoints, not the global one: Google's regional hostnames take the form LOCATION-aiplatform.googleapis.com. The global endpoint, aiplatform.googleapis.com, names no region, and Google's own advice is not to use it if you have ML processing requirements, because you cannot control or know which region processes the request. Google documents blocking global endpoint traffic with organisation policy; do that rather than trusting every SDK default.
Check the model row, not the region: Before a model name goes into configuration, find its row in Google's per-region table. For me-central1 and me-central2 the generative rows that carry a tick are the three embedding models, plus Virtual Try-On in Doha. No Gemini chat model, partner model or open model is listed in either.
CPU in Doha, L4 in Dammam: Vertex AI's accelerator table lists nothing for me-central1 and L4 for me-central2. Compute Engine's GPU locations list no Doha zone at all, against G2 and G4 machine types in me-central2-a and G2 in me-central2-c. A self-hosted model in Doha runs on CPU.
What the Qatar Data Boundary pins: The Assured Workloads package sets location controls through organisation policy constraints and sits in the free tier. Support cases are routed to global support personnel, and Google Cloud MCP servers get no residency controls for data in use or in transit, so block them with IAM as Google's page describes.
Bedrock routes as transfers: Record every Bedrock call made through Global routing as a cross-border transfer, because AWS's own definition is that such a request may be processed in any commercial Region. In me-south-1 there is no In-Region model to call instead.
Log where each call ran: Write the endpoint hostname, model version and region into the application log for every request, next to the store each retrieved document came from. That is the record a regulator or a DPO will ask for, and it cannot be rebuilt afterwards from a supplier's memory.

Sources

In this topic

What do AI projects in the Gulf have to meet, and how do we deliver them there?

Saudi PDPL and AIGuide
Which duties in Saudi Arabia’s Personal Data Protection Law apply when an AI system processes personal data?
CBUAE AI guidanceGuide
What does the CBUAE’s AI guidance note expect of a UAE bank or insurer, and what evidence shows it?
AI in Abu Dhabi healthcareGuide
What does an AI system have to meet to be used in Abu Dhabi healthcare?
Arabic document processingGuide
How well can OCR and document AI read Arabic and bilingual forms, and how do we test it on our own documents?
Arabic-English enterprise searchGuide
How do we build AI search that answers questions in Arabic and English from documents written in both?
DIFC Regulation 10 and AIGuide
What does DIFC Regulation 10 ask of a firm that uses AI on personal data?
UAE bank AI vendor contractsGuide
What does the CBUAE Outsourcing Regulation require in a UAE bank’s contract with an AI vendor?
AI system register for Gulf regulatorsGuide
What must an AI system register hold to satisfy the QCB, the CBUAE and Abu Dhabi’s Department of Health, and how is it kept current?
Human oversight in Gulf financeGuide
Which human oversight model can a Gulf bank use for which AI system, and what do the UAE and Qatar central banks require?
Bahrain PDPL and AIGuide
What does Bahrain’s Personal Data Protection Law ask of an AI system before it goes live, and where can the data be sent?
Bahrain government AI policyGuide
What does Bahrain’s government AI policy ask of the suppliers who sell AI to government entities?
Qatar PDPPL and AIGuide
What does Qatar’s data privacy law, Law No. 13 of 2016, require before an AI system processes personal data?
QCB AI GuidelineGuide
What does the Qatar Central Bank’s AI Guideline require before a bank or other licensed entity launches an AI system?

Further

AI governance implementation · The engagement this is bought as, where residency is recorded per store alongside the other controls.
Production AI systems · The build this becomes: access, failure modes, evaluation and audit on a system that runs every day.
Engagement file D-02 · The Middle East estate behind this page, replatformed into Google Cloud's Doha region for residency.
AI risk assessment · Score one AI use case on six factors before launch, and place it on a tier.

We build these systems for a living. See the engagement files for what that looks like in practice, or write to us if yours is the next one.

Book a 30-minute call enquiries@1aym.com See the engagement files

Last reviewed 30 September 2026 · 1AYM

Gulf AI data residency: the architecture questions buyers should ask

Three questions, and only one of them is legal

Where does Gulf law actually require data to stay?

Saudi government data: which AI workloads can run where

Qatar, Kuwait, Bahrain and Oman: what changes for AI data residency

A live region in the country does not mean the models run in it

Google Cloud's Doha region: what runs in me-central1 for AI

Data at rest is not data in use

What the workable architectures actually are

The questions to put to any vendor or consultancy

Where we have had to make these decisions

Sources

Related questions

Does the Saudi PDPL require data to stay in Saudi Arabia?

Can we use Claude or GPT models and still meet Gulf data-residency requirements?

Is a local cloud region enough on its own?

What about the UAE's free zones?

How long does an answer to this question stay true?

Can a Saudi government entity use an AI model hosted outside the Kingdom?

Does Google Cloud's Doha region run Gemini?

Which Gulf country requires consent to send personal data abroad?

In this topic