Engineered intelligence

Agreement strengthens the Mubadala Investment Company (MIC) Group’s ability to deploy sovereign, enterprise grade cloud, data and AI infrastructure

Reducing data risk and unlocking AI readiness across customer environments

Appointment marks next phase of international growth as demand for sovereign AI accelerates

Former Microsoft leader to scale Core42’s global commercial operations

AI costs can escalate quickly when usage grows faster than the budgets, visibility, and controls needed to manage it. As generative AI moves from pilots into production, every interaction consumes tokens, and costs rise with usage, context, latency, and workflow complexity. In the inference era, the goal is not simply to access capable models. It is to deliver useful intelligence repeatedly, economically, and under control. That means looking beyond token price to useful output per dollar, delivered at the speed the business requires. Achieving this demands the right model, accelerator, and deployment path for every workload. Core42 Compass makes this possible by routing, governing, and optimizing AI consumption across diverse silicon, turning inference economics from a source of risk into a managed capability. WHAT YOU'LL LEARN: Why lower token prices do not always lead to lower total AI spend. Why useful output per dollar matters more than token price alone. How model, accelerator, and deployment choices affect inference economics. Why no single accelerator is right for every AI workload. How Compass provides cost visibility, governance, and control. How to assess your organization’s readiness for production-scale AI.

GenAI moves from interesting to essential the moment teams can point to specific, repeatable use cases that deliver business value. The challenge is rarely model capability; it is identifying the right starting points, mapping them to industry context, and running them in production securely and at scale. The Compass Use Case Guide is built to bridge that gap. The guide opens with the six foundational GenAI patterns Compass is built to support: enterprise knowledge assistants, customer support and virtual agents, GenAI copilots for productivity, agentic AI workflows, AI content generation at scale, and retrieval-augmented generation applications. Each pattern represents a category of value, from grounding responses in trusted enterprise data to deploying autonomous agents that plan, reason, and execute multi-step tasks across systems. From there, the guide maps GenAI into six industry verticals with concrete, deployable use cases. In telco and media, that includes real-time AI inference for customer interactions, speech analytics, churn prediction, billing optimization, and AI-driven traffic management. In healthcare, it covers medical image analysis, AI-powered diagnostics, clinical documentation, real-time patient monitoring, and triage assistants. Public sector applications span predictive public safety, government contact centers, regulatory assistance, judicial case summarization, and traffic flow management. Banking and finance use cases include AI-powered fraud detection, virtual customer assistants, intelligent document processing, KYC/KYB onboarding, and credit risk insights, all areas where explainability and compliance matter as much as accuracy. Manufacturing and energy round out the guide with use cases like energy consumption optimization, synthetic data generation, material science discovery, predictive maintenance, anomaly detection, real-time asset monitoring, and grid resiliency planning. Across every industry, the same Compass capabilities apply: access to leading models through one unified API, white-glove fine-tuning support, sovereign deployment with secure local integration, and a future-ready architecture that scales with evolving needs.

Core42 Sovereign AI Cloud is a full-stack, AI-native cloud platform built for the full intelligence lifecycle, from training and fine-tuning to production-grade inference. The platform is anchored by two core services: GPU as a Service, providing direct access to a diverse range of accelerators via bare metal, Kubernetes, or Slurm orchestration, and Compass Inference as a Service, enabling teams to deploy and scale models with low latency, enterprise-grade performance, and built-in scalability. The platform is engineered for accelerator choice without lock-in, supporting NVIDIA, AMD, Cerebras, Qualcomm, and Microsoft silicon so teams can match the right hardware to each workload. AI-optimized storage delivers fast, reliable access for AI and HPC data, designed for large-scale training and high-concurrency workloads without becoming a bottleneck and engineered with enterprise-grade resiliency for operational continuity. Performance is validated by independent benchmarks. The Core42 Maximus-01 (AMD MI300X) system in the US ranks #20 worldwide on the Top500 HPC list and #3 worldwide on the IO500 storage benchmark. In the UAE, the Core42 NVIDIA DGX system ranks #37 globally (#1 in the UAE) and the Core42 AMD MI210 system ranks #38 globally (#2 in the UAE). These rankings reflect sustained system balance across compute, networking, and storage at production scale. The brochure walks through the full architecture stack: GenAI services (agents, RAG, guardrails, fine-tuning, evaluation), model hosting and inference (model catalog, model-as-a-service), AI Ops (training, model customization, model governance), and infrastructure-as-a-service (compute, ultra-fast storage, high-speed networking, managed Kubernetes and Slurm, vector data management, access management, billing, and metering), all underpinned by a unified security and compliance layer. The platform supports two primary consumption models. On-demand GPU instances give teams immediate access to diverse GPUs via the Core42 Sovereign AI Cloud console with pay-as-you-go pricing and no long-term commitments, ideal for ML experimentation and inference. Large-scale GPU clusters provide reserved capacity for sustained training workloads, with managed Kubernetes and Slurm orchestration and high-speed networking over InfiniBand and Ethernet. Built for global scale, Core42 Sovereign AI Cloud operates 86K+ GPUs across sovereign data centers in the US (Buffalo, Minneapolis, Stockton, Sunnyvale, Dallas), UAE, Southern Europe, with Kenya, India, and SE Asia in active deployment. The brochure closes with the MBZUAI case study: how the Mohamed bin Zayed University of Artificial Intelligence trained frontier models including Jais, K2, and Jais Climate on a sovereign, heterogeneous compute environment combining NVIDIA DGX SuperPod with AMD MI210 GPUs hosted within the UAE, accelerating research timelines while keeping all nationally significant models under UAE jurisdiction.

Despite an estimated $30-40 billion in enterprise investment into generative AI, 95% of organizations have yet to see measurable return on their initiatives. The gap between AI ambition and AI ROI is not a model problem. It is an architecture problem. Pilots succeed in isolation, then stall when they meet the realities of enterprise data, governance, and integration. This roadmap is built for digital transformation leaders who need to close that gap. The paper opens by examining why enterprise sovereign AI projects stall: a lack of unified strategy that creates fragmented stacks, data readiness gaps that undermine value and trust, compliance and sovereignty treated as afterthoughts, and a widening gap between AI talent and operating models. Each barrier is grounded in research from McKinsey, Gartner, Accenture, and others, with 65% of organizations yet to scale AI enterprise-wide and 77% of engineering leaders citing integration as a major challenge. The central argument is a shift in mental model: AI must stop being a collection of disconnected projects and become the operating system of the business. That means AI embedded into workflows rather than sitting beside them, shared governed data rather than ad hoc extracts, a coherent full-stack platform rather than a toolbox, and reusable patterns rather than rebuilt-from-scratch use cases. Sovereignty and governance become foundational architecture, not procurement details. From there, the paper provides a five-step blueprint for embedding sovereign AI into the business: define an AI strategy anchored in real workflows and KPIs, invest in AI-ready data and treat data as a product with documented lineage, plan AI with business, compliance, and risk teams in the room from day one, source a full-stack sovereign AI cloud platform that supports multi-accelerator workloads and unified orchestration, and integrate AI into systems of record like ERP, CRM, and HR so it can trigger and update real transactions. The paper closes with how Core42 Sovereign AI Cloud operationalizes this approach: a full-stack sovereign AI platform recognized by Top500 HPC and IO500, combining GPU-as-a-Service and inference-as-a-Service with unified orchestration, deep observability, and strict sovereignty controls. The result is a foundation for enterprises to move beyond experimentation and run AI as part of their production infrastructure, securely and at scale.

Precision Healthcare Starts with Trusted Data

Bringing private-market data into focus

Building the Foundation for an AI-Native Government What does it take to become the world's first AI-native government? For Abu Dhabi, which is on course to achieve this ambitious vision by 2027, it means more than adopting new technologies. It means fundamentally reimagining how government operates, how citizens, residents and businesses interact with public services, and how intelligence becomes embedded into every aspect of government decision-making and service delivery. As the entity driving Abu Dhabi Government's ambitious vision to become the world's first AI-native government by 2027, the Department of Government Enablement (DGE) recognized that achieving this transformation required a cloud foundation capable of supporting AI at scale while maintaining the highest levels of trust, security, and sovereignty. The Sovereignty Imperative DGE's vision required access to the latest AI innovations and hyperscale cloud capabilities. However, it also introduced a critical challenge. As AI adoption accelerated across government entities, DGE needed assurance that sensitive government information, citizen data, and government intellectual property remained protected and governed within UAE jurisdiction. The organization viewed every AI interaction, every dataset, and every token generated through AI systems as a government asset that required visibility, control, and accountability. The challenge was balancing two priorities that many governments around the world continue to struggle with: gaining access to frontier AI technologies while maintaining sovereignty, data privacy, and compliance requirements. For DGE, compromising on either was not an option. Your browser does not support the video tag.

Building the Digital Foundations for Growth Aldar is one of the Middle East's leading real estate developers, with a diverse portfolio spanning residential communities, commercial properties, hospitality, education, and retail. As the organization expanded its digital ambitions, technology became a critical enabler of growth, customer engagement, and operational excellence. From digital customer experiences and smart buildings to AI-powered services, Aldar's vision required a cloud platform capable of supporting innovation at scale while maintaining the highest standards of security and governance. To achieve that vision, the company needed more than cloud infrastructure. It needed confidence in how data was managed, protected, and governed. Your browser does not support the video tag.