The Gulf's AI ambitions are getting louder and more local‭. ‬At the recent GITEX Global 2025‭, ‬MENA's biggest tech showcase‭, ‬the well-deserved spotlight was on homegrown AI‭. ‬Startups like Arabic demoed large language models trained to think‭, ‬read and respond in Arabic‭, ‬proof that the region's race to build its own intelligence is past the talking stage‭.‬

Across the UAE and Saudi Arabia‭, ‬the focus has shifted from adoption to creation‭. ‬New systems are being built to process regional data and nuance‭, ‬not just English text‭. ‬What began as a push to digitise economies has become a contest to build and even export intelligence that mirrors the region‭. ‬“We're already seeing Arabic large language models integrated across finance‭, ‬government services and customer support‭,‬”‭ ‬says Bashar Alhafni‭, ‬Assistant Professor in Natural Language Processing at the Mohamed bin Zayed University of Artificial Intelligence‭ (‬MBZUAI‭). ‬Alhafni‭, ‬who leads Aram‭, ‬the university's Arabic AI Modeling Lab‭, ‬studies how Arabic language technologies can be built around human context so that systems are linguistically precise and socially grounded‭. ‬He points to major Arabic LLMs such as Jais from the UAE‭, ‬ALLAM from Saudi Arabia‭, ‬and Fanar in Qatar‭. ‬“The combination of strategic policy and robust infrastructure has made the GCC the primary engine for Arabic-first AI development‭,‬”‭ ‬he says‭. ‬That foundation rests on three pillars‭: ‬clear government direction‭, ‬serious infrastructure and technical talent‭, ‬with‭ ‬efforts to advance technologies beyond labs and integrate them into broader economic strategies‭.‬

The Gulf Advantage ‭ ‬

Google's 2024‭ ‬MENA Economic Impact Report found that AI tools generally added Dh21.8‭ ‬billion to the UAE economy‭ ‬while a Crowell‭ & ‬Moring analysis projects regional tech spending could reach‭ $‬169‭ ‬billion by 2026‭. ‬In Dubai‭, ‬the State of AI Report found that nearly‭ ‬three-quarters of organisations are already investing in Arabic-language or localised systems‭ ‬-‭ ‬a sign that building‭ ‬“in Arabic”‭ ‬is becoming a business strategy‭.‬

Simone Vannuccini‭, ‬Professor of Economics of AI and Innovation at University Cote d'Azur studies how policy affects deployment of intelligence across industries‭. ‬“I think the Gulf countries are positioning themselves as a‭ ‬'middle-way'‭, ‬alternative platform for AI compared to Chinese and US trajectories‭,‬”‭ ‬Vannuccini says‭. ‬“This approach may seek to present itself as a‭ ‬'global hub'‭ ‬for AI‭, ‬or at least a regional one‭, ‬while the rest of the world seems to close itself within borders‭.‬”

Alhafni refers to the growing global market for Arabic AI‭, ‬within and beyond the Arab world‭. ‬Governments and universities are looking to license or integrate Arabic LLMs for education‭, ‬translation and cross-cultural communication‭. ‬While the interest isn't new‭, ‬today's scale and capability have turned it from an academic pursuit into a real opportunity‭.‬

Sonny Arcot‭, ‬founder and CEO of Arcot Group‭, ‬is one of several entrepreneurs betting on the Gulf's AI momentum‭. ‬The company operates across the UAE‭, ‬the US‭, ‬and India‭, ‬offering several AI products that automate workflows in various sectors‭. ‬

Arcot's focus is on solving language and compliance gaps‭. ‬The company's AI offerings are designed to help organisations process data‭, ‬personalise user experiences‭, ‬and improve compliance‭, ‬areas where most off-the-shelf systems built for English markets still fall short‭. ‬Arcot chose Dubai as its base because of the UAE's forward-thinking approach to technology and regulation‭. ‬“Just like the U.S‭. ‬has Silicon Valley‭, ‬I believe that shift is happening here‭,‬”‭ ‬he says‭. ‬While his goal is to make AI adaptable in the region across languages and industries‭, ‬he believes a technology's ability to process Arabic-language data securely and contextually will determine a company's success‭. ‬

From research to revenue

For governments‭, ‬that next phase‭ ‬-‭ ‬scaling AI beyond national projects‭ ‬-‭ ‬is where firms like PwC Middle East come in‭. ‬Moussa Beidas‭, ‬the firm's Partner and Ideation Lead‭, ‬advises the UAE and Saudi Arabia on how to turn national strategies into global workable ecosystems‭. ‬Beidas believes both countries are well-positioned to lead Arabic AI because they combine capital‭, ‬mandates‭, ‬and data and tie‭ ‬in their national AI strategies into economic transformation plans‭. ‬Pointing to Saudi Arabia's goal of becoming a regional AI hub‭, ‬he says‭, ‬“We work to bridge linguistic and cultural gaps‭, ‬ensuring that LLMs and other AI systems can generate nuanced Arabic content‭.‬”‭ ‬

When it comes to adoption‭, ‬government agencies remain the main testing ground‭, ‬powering citizen services‭, ‬digital ID systems and‭ ‬smart-city projects‭, ‬with banks‭, ‬media and entertainment space catching up‭. ‬Beidas says sectors like healthcare‭, ‬education‭, ‬and‭ ‬SMEs still face hurdles though‭ ‬at MBZUAI‭, ‬they're addressing this through ARWI‭ [‬Arabic Write‭ & ‬Improve‭], ‬an AI-powered Arabic writing assistant designed to provide pedagogically aligned feedback to students‭. ‬That domestic focus mirrors what PwC's 2025‭ ‬AI Jobs Barometer found‭: ‬AI-related roles have surged but will require building and upskilling local talent to design‭, ‬adapt‭, ‬and govern systems from the ground up‭. ‬“The real test‭,‬”‭ ‬says Beidas‭, ‬“will be whether that growth can extend beyond national markets‭.‬”‭ ‬

The UAE's AI growth is accelerating‭, ‬but most of the activity is still happening inside national borders‭. ‬Turning Arabic-trained systems‭ ‬into exports‭ ‬-‭ ‬models‭, ‬APIs‭, ‬or enterprise tools that work across the region‭ ‬-‭ ‬will depend on how well neighbouring countries align on data‭, ‬licensing‭, ‬and infrastructure‭. ‬That's where experts see the next phase of competition‭: ‬exporting Arabic AI that works across borders‭.‬

EXporting intelligence

Alhafni believes that regional adoption of GCC-trained Arabic models will require policy alignment across countries‭, ‬infrastructure for model deployment beyond the Gulf‭, ‬and dialectal coverage reflecting linguistic diversity‭. ‬Vannuccini agrees‭. ‬He says there is a space for‭ ‬'regional models'‭ ‬becoming a go-to solution for certain areas of the world like the MENA region‭. ‬“Let's imagine large public hospitals or universities want to adopt AI-based systems‭, ‬they may turn towards solutions that are‭ ‬'regional'‭ ‬because they are more proximate in the geographical‭, ‬cultural‭, ‬and language space‭,‬”‭ ‬he says‭.‬

That kind of regional openness within AI‭, ‬says Beidas‭, ‬could turn the Gulf's AI edge into a broader industry‭. ‬Alhafni says‭, ‬“Jordan and Lebanon have great engineering talent‭, ‬lower costs‭, ‬and multilingual markets‭, ‬making them ideal for refining Gulf-built models‭. ‬It creates a healthy regional value chain‭ ‬-‭ ‬the Gulf leads on large-scale development‭, ‬while the Levant helps localise and commercialise those systems‭.‬”

The Gulf still holds the biggest short-term opportunity for Arabic AI‭, ‬with Egypt close behind‭. ‬Beidas feels that Levant countries and North Africa could support‭, ‬helping fine-tune Gulf-built systems for different dialects and local markets‭. ‬But closing the region's AI gap will require steady investment in talent‭, ‬infrastructure‭, ‬and cross-border collaboration‭.‬

That cooperation is already taking shape across the private sector‭. ‬Andrew D'Souza‭, ‬CEO of Boardy‭, ‬described his company as an‭ ‬“AI super-connector”‭ ‬that turns short conversations into curated‭, ‬double-opt-in introductions for founders‭, ‬operators‭, ‬and investors‭. ‬Boardy works with Gulf-based partners and events to help link ecosystems through applied AI‭. ‬“From our vantage point‭, ‬the UAE's strength lies in its world‐class infrastructure and ambition‭,‬”‭ ‬says D'Souza‭, ‬adding that the UAE has the infrastructure‭, ‬ambition‭, ‬and funding figured out but is missing‭ ‬the‭ ‬“last mile”‭ ‬between innovation and real-world products‭. ‬The real potential‭, ‬he says‭, ‬is in scaling that kind of collaboration across MENA through curated partnerships‭.‬

A huge hurdle is usable Arabic data‭. ‬Most existing datasets still lean towards a handful of dialects or regions‭. ‬To grow responsibly‭, ‬experts say‭, ‬AI needs to become more inclusive and trained on data reflecting the Arab world's linguistic range‭. ‬

Even with those gaps‭, ‬Beidas believes interest in Gulf-built models is rising fast‭, ‬especially among public-sector and financial‭ ‬institutions in Arabic-speaking countries that want to localise services‭. ‬What's still missing‭, ‬are standardised licensing systems‭, ‬stronger data-sharing frameworks‭, ‬and clearer commercial returns to prove the business case‭.‬

Vannuccini believes the main bottleneck is demand‭. ‬“If AI does not show real potential beyond the hype in a relatively short time‭, ‬final demand might collapse or jump to the‭ ‬'next big thing'‭,‬”‭ ‬he says‭.‬

‭ Forecasts ahead‭ ‬

Looking ahead‭, ‬the region's trajectory is clear‭. ‬“GCC-trained Arabic models are maturing rapidly‭, ‬and as they continue to improve in dialectal coverage‭, ‬efficiency‭, ‬and user alignment‭, ‬they'll start powering applications across education‭, ‬government‭, ‬and commerce‭,‬”‭ ‬says Beidas‭. ‬The next five years‭, ‬will be about turning prototypes into real-world impact‭. ‬

Others‭, ‬like Marcello Mari‭, ‬believe that exportability is already happening‭, ‬albeit not in the traditional sense‭. ‬“AI is already exported as cloud services‭, ‬on-prem deployments‭, ‬and open-licensed model weights‭,‬”‭ ‬says Mari‭, ‬former CEO of SingularityDAO‭. ‬“Ownership depends on licensing‭, ‬not geography‭. ‬Models under MIT or Apache licenses grant broad rights‭, ‬while others restrict redistribution or commercial use‭.‬”

For Mari‭, ‬the conversation is about who controls AI‭. ‬His work on decentralised systems which are designed to distribute data access and decision-making‭, ‬parallels the Gulf's push for digital sovereignty‭. ‬Initiatives like the UAE Strategy for AI and Saudi Arabia's National Strategy for Data and AI reflect that goal‭: ‬keeping data‭, ‬talent‭, ‬and economic value inside the region‭.‬

He cites the UAE's Jais model and Saudi Arabia's ALLAM‭, ‬as examples of that ambition in motion‭. ‬Both projects were designed to create Arabic-trained systems that reflect local‭ ‬context reducing dependence on foreign AI‭.‬

“The main constraint for developers in the Arabic-AI space remains the lack of high-quality Arabic data‭,‬”‭ ‬Mari says‭. ‬The UAE and Saudi Arabia are now investing heavily to close that gap through new datasets‭, ‬evaluation tools‭, ‬and training programs‭. ‬The next stage of growth will depend less on new technology and more on what surrounds it‭: ‬open licensing‭, ‬affordable data pipelines‭, ‬and transparent evaluation methods‭.‬

Beidas sees a similar trajectory‭. ‬“We're already watching a regional ecosystem take shape‭,‬”‭ ‬he says‭. ‬“The Gulf builds large-scale models‭, ‬while countries like Jordan and Lebanon help fine-tune and commercialise them‭. ‬It's a healthy value chain‭.‬”‭ ‬But real exportability will require shared datasets‭, ‬performance benchmarks‭, ‬and coordinated regulation‭ ‬-‭ ‬“the connective tissue”‭ ‬that turns policy into markets‭.‬

Realistically‭, ‬it may take a three-to-five-year timeline for the Gulf to start exporting Arabic AI through licensed models‭, ‬APIs‭, ‬and enterprise tools‭, ‬as regional data ecosystems and partnerships mature‭. ‬Vannuccini believes in the long term‭, ‬the real value capture being for those that have stacks in datacenters and physical infrastructure‭. ‬He says‭: ‬“While everyone looks at OpenAI‭, ‬it is Nvidia that has passed from being a gaming hardware company to a top worldwide giant‭. ‬The‭ ‬Gulf prospected investments in France or UK go in this direction‭.‬”‭ ‬Still‭, ‬if the region succeeds in exporting Arabic-trained models that reflect its own culture‭, ‬language‭, ‬and governance‭, ‬it won't just be creating tools‭, ‬it will be exporting identity in a way‭. ‬A homegrown technology signals the Arab world's step back from being‭ ‬consumed by imported technologies built elsewhere‭. ‬And that's the quiet revolution inside this AI race‭: ‬the move from participation to authorship‭. ‬

MENAFN31102025000049011007ID1110277380