OmniOps launches Bunyan Inference Hub for Saudi-sovereign AI
OmniOps, a Riyadh-based AI infrastructure company founded in 2024, has launched Bunyan Inference Hub, a managed platform giving governments, enterprises, startups and developers access to open-source AI models and GPU compute while keeping data and workloads within Saudi Arabia. The launch marks the commercial debut of OmniOps' Bunyan inference product, which the company describes as the Kingdom's first sovereign inference-as-a-service platform.
The platform consolidates model access and compute provisioning in a single environment, addressing a friction point for organisations that currently juggle multiple providers, GPU pools and integration layers to run production AI. Customers access predefined token bundles across supported models, with dedicated GPU infrastructure available for workloads that require additional capacity or isolation. The platform is designed to be hardware-agnostic, routing requests across different GPU and accelerator types rather than binding customers to a single vendor's silicon.
Under the hood
Central to the offering is what OmniOps calls a proprietary inference optimisation engine, developed internally by its R&D team. The company says the engine adapts how models execute on available hardware to match workload requirements, improving GPU utilisation and reducing cost per token as demand scales. OmniOps has not released independent benchmark results or published throughput figures, so those claims remain unverified at launch.
Mohammed Altassan, founding chief executive of OmniOps, said the hub brings together the company's inference technology and sovereign deployment experience "in a simplified platform that public and private sector organizations can plug into their operations and use to deploy AI solutions more efficiently."
The company cited two market data points in its release: a ServiceNow study indicating that Saudi-based organisations have increased AI spending by 124% year-on-year and plan to allocate 20% of IT budgets to AI through 2027, and a Gartner forecast placing global spending on AI inference workloads at $23.3 billion by the end of 2026. OmniOps secured $8 million in pre-seed funding prior to this launch, though no Series A or follow-on round was announced alongside the product.
Market and regulatory context
The sovereign AI infrastructure space is becoming increasingly competitive as governments across the Gulf, Europe and Southeast Asia insist on in-country data residency for sensitive workloads. Hyperscalers including AWS, Microsoft Azure and Google Cloud have all announced or expanded localised cloud regions in Saudi Arabia and the wider Gulf Cooperation Council in recent years, while regional sovereign cloud providers are positioning themselves as alternatives that more cleanly satisfy national data-protection and security frameworks.
OmniOps enters this market as Saudi Arabia accelerates its Vision 2030 digital transformation agenda, which has generated substantial public-sector AI procurement. The Kingdom's National Data Management Office sets data classification and localisation requirements that make a Saudi-hosted inference platform commercially attractive for entities handling government or regulated data. For organisations navigating those frameworks, running workloads on a hyperscaler's local region may still require contractual diligence around subprocessor flows, giving a purpose-built sovereign platform a potential compliance edge.
Longer term, the hardware-agnostic routing architecture positions OmniOps to integrate accelerators from a widening supplier base, an important hedge given ongoing GPU allocation constraints affecting the region. Investors and prospective customers will be watching for published performance benchmarks, named reference customers, and evidence of model-catalogue breadth as the clearest near-term indicators of platform maturity.