Bassel Haidar

"What to Expect When Governments Start Offering AI Compute or Intelligence-as-a-Service" For years, AI compute has been the domain of hyperscalers, big tech & well-funded enterprises. That’s about to change I increasingly see governments, especially in the GCC and Asia, moving toward AI Compute as a Service (AI-CaaS) or even Intelligence as a Service models. And when governments step in, the impact is rarely incremental. It’s systemic. Here’s what I believe we should expect. 1. AI Becomes a Public Utility, Not a Luxury When governments offer AI compute, access shifts from “who can afford GPUs” to “who has a legitimate use case.” Startups, SMEs, universities, hospitals, and even individual researchers suddenly gain access to serious compute power. This will accelerate: - local innovation - national AI talent development - sovereign AI capabilities AI stops being elitist & starts becoming infrastructure. 2. Sovereign AI Will Matter More Than Global AI Government-backed AI services will prioritize: - data residency - national language models - cultural, legal & ethical alignment - protection from foreign dependency This isn’t about rejecting global platforms, it’s about choice and resilience. Countries will want AI systems they can trust, govern, and audit. 3. Regulation Will Become “Built-In,” Not Bolted On Unlike commercial platforms, government AI services will come with: - predefined guardrails - embedded compliance - usage monitoring - explainability requirements For regulated industries like healthcare, etc. this is huge. AI adoption becomes faster because risk is pre-engineered out, not debated endlessly after deployment. 4. Cost Models Will Shift from CapEx to National Opex Instead of every organization buying GPUs, licenses & talent independently, governments will: - pool demand - negotiate scale - subsidize access where it matters This dramatically lowers the barrier to experimentation and reduces duplication of effort across sectors. 5. Intelligence-as-a-Service (AIaaS) Is the Real Game Compute is just the beginning. The real shift happens when governments start offering: - pre-trained national models - AI agents for citizen services - sector-specific intelligence (health, mobility, energy, security) - decision-support systems, not just raw models At that point, governments stop being users of AI & become platform providers. 7. Enterprises Will Rethink Their AI Strategy When sovereign AI is available, enterprises will ask: - What do we build ourselves? - What do we consume from the state? - What must remain proprietary? This will reshape enterprise AI roadmaps, especially in sensitive sectors. 6. Cybersecurity & Trust Will Be Non-Negotiable Government AI platforms will be judged not by speed alone, but by: - security of models & data - resistance to model theft & poisoning - protection against misuse - resilience against nation-state threats AI without trust will not scale. Period.

53

4 Comments

At iHorizons, our AI-first approach is grounded in delivering practical value and measurable impact. Dr. Hussein Mohamed, Ph.D., Manager of AI & Data Science, shares insights into leading a team dedicated to building dependable AI services and addressing real-world challenges through data and advanced analytics.   #LifeAtiHorizons #InnovationAtWork #InTheSpotlight

15

1 Comment

Nearly all data leaders in the UAE (94%) confess they lack complete visibility into how their AI systems make decisions, according to Dataiku’s Global AI Confessions Report: Data Leaders Edition. Conducted by The Harris Poll, the study surveyed over 800 senior data executives across eight countries, including 100 from the UAE, revealing major risks in AI governance, transparency, and accountability. As the UAE accelerates its ambition to become a global AI powerhouse under the National Strategy for Artificial Intelligence 2031, the findings underscore a growing tension between innovation and oversight. Only 17% of UAE data leaders said their organizations always require AI systems to “show their work” — among the lowest globally. Alarmingly, 62% are not confident their AI systems could pass a basic audit, while 72% would still trust an AI agent to make autonomous business decisions even if it couldn’t explain its reasoning. “UAE enterprises are betting big on AI — but trust and transparency must keep pace with innovation.” Florian Douetteau, Co-founder and CEO, Dataiku Learn more: https://lnkd.in/gNUjqZ_q SK Mohapatra Sanjib Mohapatra

5

The Phylogeny of Intelligence traces the evolution of Agentic AI ecosystems through a biological lens, progressing from simple code blocks to complex, adaptive multi-agent societies with nervous systems built on transport protocols like MQTT for lightweight IoT pub-sub messaging, gRPC for low-latency cluster communication, WebSockets, and REST for human interfaces. A pivotal divergence emerges between orchestrated systems centralized conductor models ensuring precision but risking single points of failure and choreographed swarms that foster decentralized resilience in dynamic environments. Structural adaptations draw from ESB legacies, including Agent-Controller hierarchies for strategic execution, Splitter-Aggregator patterns for parallel problem-solving, and content-based routing with message transformation. Cognitively, agents shift from brittle rule-driven instincts to hybrid intelligence blending LLMs, reinforcement learning, and neuroplasticity for creative adaptation within safety rails, culminating in a future where enterprises cultivate thriving digital biospheres rather than mere tools. #AgenticAI #AIEvolution #MultiAgentSystems #AIOrchestration #DigitalTransformation Full article on Medium [check below link]

8

Mohamed El-Shazly is a Cloud Solution Architect and HPC/AI/ML specialist currently serving as Lead Solutions Architect at OmniOps. In this role, he leads and mentors architecture engineering teams while designing AI-driven, cloud-native solutions tailored to enterprise and government use cases across Saudi Arabia. Operating at the intersection of hyperscale cloud, GPU-powered AI clusters, and DevOps automation, Mohamed translates complex business requirements into scalable, secure, and cost-optimized architectures across Azure, AWS, and Google Cloud. From AI/ML platform enablement and LLM inference environments to high-performance computing (HPC) infrastructure, he acts as a trusted advisor guiding organizations through AI adoption, cloud transformation, and modern DevOps practices. Previously, Mohamed played a pivotal role at SDAIA | سدايا within the National Center for Artificial Intelligence (NCAI), deploying large-scale AI/ML cloud services powered by GPU infrastructure and containerized HPC environments. He integrated NVIDIA-based HPC systems with enterprise identity platforms and secured multi-cloud workloads with private endpoints and zero-trust controls. Earlier, at Elm Company and the Ministry of Education, Saudi Arabia, he architected resilient cloud infrastructures, implemented Kubernetes clusters, and built end-to-end CI/CD pipelines. With deep expertise spanning AI platforms, cloud security, DevOps automation, and high-availability architectures, Mohamed stands at the forefront of enabling AI-powered, performance-driven digital ecosystems. #MohamedElShazly #CloudArchitecture #HPC #AIInfrastructure #MachineLearning #LLMDeployment #DevOps #MultiCloud #SDAIA #EnterpriseAI #CloudSecurity #DigitalTransformation #TechLeadership #AIMena

46

Now, MIT research from the Jameel lab has uncovered that LLMs recommended a much lower level of care for female & minority patients, suggesting a higher level of harmful bias in commercially available LLMs https://buff.ly/2L3zRkS #bias #llms #ai #genai #healthcare #mentalhealth

4

2 Comments

I often talk about the need to treat #AI as a direct extension of data and not just as another technology. This is especially true for organizations that must rely on internal, non-public, and proprietary data for genAI use cases. Here is the segment I recently recorded for Fed Gov Today with Patrick McGarry and Francis Rose on that point. #BeyondHype #DataGovernance https://lnkd.in/eCJkKDnm

19

Excited to share that Oracle database and cloud application customers in the utility, AI, and nuclear industries can now tap into the power of OpenAI GPT‑5. By merging trusted business data with cutting-edge AI, we’re helping organizations make smarter decisions and drive real innovation. #AI #OCI #FUSION

1

Hybrid Computing Solves Complex Problems Highlighted in Gartner's October 21, 2024, trends for 2025, hybrid computing combines diverse compute mechanisms like quantum and classical for advanced problem-solving. Verified applications include research labs using hybrid setups for optimization tasks in drug discovery, based on collaborations initiated in 2024. This integration enables high-speed innovation, surpassing limits of single systems as demonstrated in simulations for materials science. Security risks are managed through orchestration, with early adopters reporting enhanced automation in business processes. The technology supports augmented human capabilities, allowing real-time personalization at scale in sectors like finance. As an experimental field, it requires specialized skills but promises autonomous operations in the near term. What complex problems could hybrid computing solve in your field? Share your ideas and questions below! #HybridComputing #AdvancedTech #GartnerTrends #QuantumClassical #Innovation2025