Gulf infrastructure strategy bets on both camps by combining American chips and Chinese models

Prevailing estimates in the global AI race point to a split of the world into two closed camps: an American camp led by giant cloud-computing firms, leading model labs and Nvidia, and a Chinese camp that owns its models, engineering capabilities and an integrated, escalating ecosystem. However, the on-the-ground reality in the Gulf region, especially in Saudi Arabia, reveals a different trajectory, as the countries of the region are not seeking to replicate either ecosystem; instead they are shaping a strategy to be among the few locations worldwide capable of running both camps together and drawing on them simultaneously.

This duality is clearly reflected in the moves of the Riyadh-based, government-backed company Huumain. The company relies on American technologies to build its computing infrastructure and commits not to train advanced Chinese models on computing capacity obtained through U.S. export licences, while at the same time keeping its partnerships with Chinese institutions open. During the LIP conference in Riyadh, the company unveiled an Arab-origin technology built on a model from the Chinese firm Mini Max. That is matched by a massive investment commitment from Amazon Web Services of $5.3 billion for data centres inside the Kingdom, countered by a parallel Chinese expansion that includes building ByteDance’s cloud arm infrastructure in Riyadh, Tencent Cloud obtaining a Saudi operating licence, and Lenovo launching the first laptop bearing the “Made in Saudi Arabia” label.

Artificial intelligence resembles heavy industry more than mere software; it requires vast tracts of land, massive electrical power, advanced cooling technologies, and huge capital investments.Here, the competitive advantage of the Gulf states that possess these resources, together with geopolitical flexibility, becomes evident. Saudi Arabia’s strategy aims to compete not by creating a better chip or training a larger model, but by securing a growing share of the physical infrastructure on which everyone’s models run, and by leveraging access to these centres to attract investment, transfer knowledge and impose localisation requirements. This approach is embodied in DataVolt’s project, which is constructing a 100-megawatt facility costing roughly $1 billion in Neom, dedicated to exporting computing capacity, with plans to expand it to 360 megawatts.

Strategic calculations start from a realistic equation: U.S. restrictions push Riyadh toward Chinese alternatives, while Chinese shortcomings in delivering leading-model performance push it back toward Washington.In the end, the two sides compete for a resource increasingly controlled by the Gulf: infrastructure. While Washington treats chips as a tool of foreign policy and Beijing leverages open models to reach markets its hardware cannot access, the Gulf states act as pivotal swing nations that reject a forced choice and profit from international competition, replicating their expertise in turning conventional energy assets into geopolitical influence, this time through computing units and data centres, as Chinese chip firms rise to challenge Nvidia’s dominance, such as Shanghai Infinim Technology, whose IPO demand exceeded the offered shares by roughly 6,000 times.

The shift toward dual infrastructure has a direct impact on technology and cost decisions across the Arab region.For engineering teams and organisations in the Gulf, Egypt and the Levant, this model means that build-and-development decisions are no longer confined to a single Western computing environment; hybrid architectures that combine advanced Chinese models with major cloud-computing centres are now feasible. Regional CTOs must precisely grasp regulatory compliance requirements, as the strict separation between using American chips and training Chinese models demands systematic scrutiny of supply chains and software licences. Moreover, exporting computing capacity directly from large regional projects such as Neom enables companies to reduce latency and rely on nearby local data centres, while necessitating the upskilling of technical staff to handle multi-vendor cloud environments.

This trajectory reveals a new formulation of the technology-playbook, in which the world does not settle into two isolated blocs but opens the door for regional powers to employ their material resources and strategic location to link Western supply chains with Eastern capabilities without being subject to a single-party set of conditions.