Positron AI Ltd, a US-based developer of specialised AI inference infrastructure, has established its first international presence in the Dubai International Financial Centre (DIFC), the leading global financial centre across the Middle East, Africa and South Asia (MEASA) region.
The DIFC licence represents a significant milestone in the company’s global growth strategy. Positron AI has secured more than US$300 million in funding, including a US$230 million Series B round, deployed its inference technology with a major US hyperscaler, and attained unicorn status.
The expansion comes at a time when AI inference is emerging as the primary driver of global computing demand. AI inference involves the deployment, operation and scaling of large language models in a reliable, cost-effective and commercially viable manner.
Husni Khuffash, Managing Director, MENA at Positron AI, said: “Establishing our presence in DIFC reflects our commitment to innovation and to advancing the future of AI infrastructure. Our focus on power-efficient, deployable inference solutions supports the growing demand for scalable AI capabilities across industries. Being part of DIFC enables us to collaborate closely with partners, regulators and industry stakeholders while contributing to Dubai’s vision of becoming a global leader in digital transformation and artificial intelligence.”
Mohammad Alblooshi, Chief Executive Officer of the DIFC Innovation Hub, said: “We are delighted to welcome Positron AI to the Dubai AI Campus ecosystem. The company’s innovative approach to AI inference infrastructure aligns with the next phase of growth in Dubai’s technology and innovation landscape. Positron AI’s vision complements DIFC’s transformation into an AI-native jurisdiction, reinforcing the Centre’s role as a platform for innovation, scale and responsible AI development.”
As industry demand shifts from AI model training to large-scale model inference, Positron AI is positioning itself as a next-generation alternative to conventional AI hardware providers. Its technology delivers lower cost per token, increased memory density and improved energy efficiency while addressing key challenges such as high power consumption, memory limitations and the escalating costs of GPU-based infrastructure.
The company’s first-generation server, Atlas, was developed to support large language model inference for small and medium-sized models with up to 500 billion active parameters. It delivers performance comparable to DGX-H100 systems while operating with lower power consumption and reduced costs.
Looking ahead, Positron AI’s next-generation platform, Titan, is expected to launch in the first quarter of 2027. The system is designed to support frontier AI models with enhanced performance and greater power efficiency than NVIDIA’s Rubin and Blackwell GPU architectures.
