UAE’s sovereign AI playbook — how G42 is positioning itself in the global open-weight race

The United Arab Emirates is moving rapidly to cement its position as a sovereign AI hub, and G42 — along with its cloud and AI infrastructure arm, Core42 — is at the center of that effort. The recent deployment of OpenAI’s open-weight GPT-OSS models on Core42’s AI Cloud is more than a product launch. It is a calculated step in a broader national strategy to combine high-performance artificial intelligence with sovereign control, giving the UAE a competitive edge in a crowded global technology landscape.

Core42’s integration of the gpt-oss-20B and gpt-oss-120B models via its Compass API offers organizations access to some of the fastest inference speeds available today, with throughput of up to 3,000 tokens per second per user. The platform allows enterprises, research institutions, and developers to run these models on a choice of silicon architectures — from top-tier GPUs to alternative accelerators — without compromising performance. This silicon-diverse approach is designed to optimize both cost and efficiency, and more importantly, to ensure that sensitive workloads can be processed entirely within national borders.

Why is the UAE betting on open-weight AI models as a foundation for sovereign infrastructure?

Open-weight models have emerged as a strategic asset in the AI arms race. Unlike closed-weight commercial APIs, they offer full transparency, fine-tuning flexibility, and the ability to be deployed in-country without relying on a vendor’s global cloud. For governments and regulated industries, this is a critical differentiator.

Core42’s offering aligns neatly with the UAE’s sovereign AI priorities. The ability to process data locally not only meets strict compliance requirements in sectors such as healthcare, finance, and defence, but also reduces the strategic risks of vendor lock-in. By giving customers the option to adapt AI models to local languages, regulations, and industry-specific contexts, Core42 is effectively positioning the UAE as a neutral yet technologically advanced player in the open-weight space.

This strategy builds on a series of high-profile infrastructure investments by G42. In the past year, the group announced the creation of a 5-gigawatt US-UAE AI campus and launched the 1-gigawatt Stargate UAE facility, one of the largest AI supercomputing platforms in the region. These developments, combined with a $1.5 billion investment from Microsoft in 2024, have strengthened the UAE’s ability to compete with both Western hyperscalers and emerging Asian AI powerhouses.

Institutional sentiment around the GPT-OSS launch has been cautiously optimistic. Analysts highlight that Core42’s combination of performance, sovereignty, and hardware choice is rare in today’s market, where most providers still lock customers into single-supplier environments. For large multinational corporations, especially those operating across jurisdictions with conflicting regulatory demands, the flexibility to choose silicon and control deployment locations could be a deciding factor in procurement.

Beyond the technology layer, there is also a geopolitical dimension. The UAE’s neutral diplomatic positioning enables it to serve clients from a diverse set of regions without the same restrictions faced by providers in the U.S. or China. If leveraged effectively, this could make the country a preferred hub for cross-border AI collaboration, model training, and compliance-centric deployments.

Looking ahead, the next phase of competition is likely to focus less on raw compute power and more on ecosystem readiness. Core42 will need to build an extensive developer and tooling environment around its models, offering APIs, orchestration frameworks, and fine-tuning pipelines that make integration as seamless as possible. Enterprises will expect not only sovereign control and speed, but also the ease of deployment that closed commercial APIs have perfected over the past few years.

If Core42’s performance benchmarks — including its claim of delivering up to 3,000 tokens per second per user on silicon-diverse infrastructure — are validated through independent, transparent testing, it could mark a turning point in the emerging open-weight AI infrastructure segment. Verified benchmarks would not only provide credibility in technical circles but also serve as a powerful procurement lever when competing for large-scale contracts in sectors like healthcare, banking, defence, and national research programs. In enterprise AI, performance numbers that stand up to third-party scrutiny often become a “trust currency,” influencing long-term purchasing decisions and vendor selection.

Such validation would significantly strengthen the UAE’s global AI brand, positioning it as a credible alternative to the AI infrastructure ecosystems dominated by U.S. hyperscalers and Chinese cloud giants. The ability to deliver world-class performance while guaranteeing sovereign control is a combination that few providers can match at scale. For policy-makers and chief information officers balancing innovation with compliance, this dual proposition — speed plus sovereignty — could be a decisive factor.

From a strategic perspective, Core42’s rise would also carry a symbolic weight. It would demonstrate that smaller but well-capitalized nations with cohesive industrial policy and targeted infrastructure investment can go head-to-head with the largest players in the AI race. This challenges the long-held assumption that only countries with massive domestic markets or decades of tech sector dominance can lead in advanced computing. If replicated, the UAE model could become a template for other mid-sized economies seeking to carve out a place in the global AI value chain, leveraging open-weight frameworks to accelerate adoption and innovation without ceding control to foreign vendors.