Can UAE’s Core42 set a new benchmark for sovereign, open-weight AI infrastructure?

Core42, the artificial intelligence and sovereign cloud subsidiary of G42, has made OpenAI’s latest open-weight models — including the gpt-oss-20B and gpt-oss-120B — globally available on its Core42 AI Cloud, unlocking new performance and compliance options for enterprises, developers, and researchers.

The deployment is fully integrated with the Core42 Compass API, enabling instant access to a variety of high-performance compute platforms and delivering inference speeds of up to 3,000 tokens per second per user. This makes it possible to run real-time AI workloads at scale while maintaining sovereign control, a capability that is increasingly critical in regulated industries such as healthcare, finance, and national security.

How does Core42’s GPT-OSS launch change the economics and accessibility of high-performance sovereign AI infrastructure?

By placing OpenAI’s open-weight models on infrastructure optimized for both speed and compliance, Core42 is positioning itself as a leader in sovereign AI enablement. Organizations can select from multiple silicon architectures — including GPUs and other accelerators — to optimize for either raw performance or cost efficiency. This “silicon freedom” approach allows customers to align infrastructure with workload demands, whether for low-latency inference, large-scale training, or specialized applications.

Institutional observers note that such flexibility is a significant differentiator at a time when AI deployment models are diverging into public-cloud, hybrid, and fully sovereign environments. Open-weight access means enterprises can inspect, fine-tune, and adapt the models without dependency on a single vendor’s hosted API, which is often a limiting factor in regulated markets.

Why is sovereign control over AI models becoming a critical factor for regulated sectors and public-sector deployments?

The Core42 rollout aligns with a broader global push for sovereign AI infrastructure, in which organizations maintain control over where and how AI workloads are executed. For regulated sectors, this ensures compliance with jurisdiction-specific data residency laws and operational security requirements.

Core42’s Compass API enables in-country deployments that meet sovereign control standards, allowing sensitive workloads to remain within national borders. This capability is seen as particularly valuable for governments and industries that cannot risk their data transiting through global public-cloud regions.

Market participants point out that the United Arab Emirates has positioned itself as a proactive hub for such sovereign AI initiatives, using partnerships and infrastructure investments to court both domestic and international workloads.

What are the performance and scalability advantages of Core42’s silicon-diverse infrastructure for enterprise AI adoption?

Core42 claims industry-leading inference throughput — up to 3,000 tokens per second per user — allowing for real-time AI interactions even in high-volume, multi-user scenarios. The system’s design supports agentic AI workloads, advanced automation, and decision-support systems without performance bottlenecks.

For large-scale enterprises, the combination of high throughput and low latency means AI services can be integrated directly into operational workflows — from call center automation to industrial process monitoring — without the lag that can undermine adoption. Analysts say this can directly translate into lower total cost of ownership by enabling shared infrastructure across multiple use cases.

How could open-weight deployment shape enterprise AI autonomy and innovation cycles?

Open-weight availability gives organizations the freedom to retrain or fine-tune the gpt-oss models on proprietary data, enabling domain-specific adaptations. This removes the black-box nature of closed-weight APIs and fosters innovation by allowing internal teams to iterate quickly without vendor bottlenecks.

Institutional sentiment suggests that this move by Core42 could accelerate the creation of region-specific and sector-specific AI applications, particularly in languages and contexts underserved by mainstream large language models. By removing licensing constraints, Core42 is effectively empowering its customers to own and evolve their AI assets.

What are the potential cost and compliance implications for enterprises adopting Core42’s GPT-OSS offering?

Core42 positions its GPT-OSS deployment as cost-efficient, particularly for agentic AI workloads that require persistent, high-volume inference. Customers can deploy in committed infrastructure environments, locking in predictable performance and cost, while meeting sovereign control requirements.

Compliance is strengthened through transparent model operations, where organizations can verify model behavior, audit data handling, and ensure outputs meet jurisdictional standards. For cost-sensitive sectors like education, municipal governance, and mid-market enterprises, the ability to balance compliance with affordability could be a tipping point for adoption.

How does this launch fit into G42’s broader AI and infrastructure growth strategy?

This release follows a string of high-profile G42 announcements, including a 5-gigawatt US-UAE AI campus and the launch of the 1-gigawatt Stargate UAE facility, part of a multi-phase AI supercomputing initiative. In 2024, G42 also secured a $1.5 billion investment from Microsoft, bolstering its position as a global AI infrastructure leader.

Industry watchers view the GPT-OSS integration as a natural progression of G42’s ambition to become a central node in the international AI supply chain. By combining massive compute capacity with strategic partnerships and sovereign-ready platforms, G42 is not just targeting regional dominance — it is vying for influence in the global AI standards conversation.

What are the forward-looking expectations for Core42’s role in the global AI infrastructure market?

Looking ahead, institutional sentiment suggests that Core42’s combination of open-weight model access, sovereign deployment capabilities, and high-throughput AI infrastructure is well-positioned to resonate with a wide spectrum of high-value customers. This includes national governments seeking AI sovereignty to protect sensitive data, hyperscale cloud providers aiming to expand their infrastructure portfolios with silicon-diverse options, and multinational corporations that want to reduce dependency on single-vendor ecosystems. In an era where regulatory environments are tightening and AI supply chains are under geopolitical scrutiny, the ability to deploy powerful language models entirely within jurisdictional borders is emerging as a core procurement criterion.

Analysts say the next competitive frontier will not be limited to raw compute performance but will hinge on the depth and maturity of the surrounding developer ecosystem. Core42’s success will depend on building robust APIs, developer toolkits, orchestration frameworks, and fine-tuning pipelines that allow enterprises to integrate open-weight models as seamlessly as they would consume closed commercial APIs from providers like OpenAI or Anthropic. This means reducing friction in model customization, simplifying compliance integration, and offering turnkey deployment templates for sector-specific use cases such as healthcare diagnostics, financial compliance automation, and multilingual government services.

If Core42’s performance benchmarks — including its claim of delivering up to 3,000 tokens per second per user — are validated through independent, transparent testing, it could establish an early leadership position in the open-weight AI infrastructure market. Such proof points would not only bolster credibility with procurement teams in regulated industries but also reinforce Core42’s pitch as a cost-effective, sovereign-ready alternative to traditional hyperscale AI providers. Over time, this combination of speed, compliance assurance, and developer-friendly integration could make Core42 a strategic choice for organizations seeking both technical autonomy and geopolitical resilience in their AI strategies.