Microsoft turns Middle East AI build-out into infrastructure policy

September 24, 2026

Microsoft says it plans to spend more than $10 billion through 2030 on cloud and AI infrastructure, subsea and terrestrial connectivity, and regional AI partnerships across Kuwait, Qatar, Saudi Arabia and the United Arab Emirates.

This looks less like another datacentre expansion than a commitment that ties AI capacity to cables, national partners and a multi-country timetable. That matters beyond Microsoft because it gives governments and large buyers in the region something they can plan around now: where workloads may run, which cloud footprint may exist, and which supplier is willing to build with state-backed institutions rather than just sell models into the market. For Microsoft, the move also makes visible that competition in generative AI is being fought with financing and physical build-out as much as with models.

OpenAI says the Government of Ukraine will get access to its Daybreak programme to help defend civilian infrastructure by finding software vulnerabilities and testing fixes faster. We read this as another step in turning frontier AI access into a supervised security relationship rather than an ordinary software subscription: the provider is choosing a counterparty and a use case, and the effect reaches other governments and critical-infrastructure operators that now have a clearer example of how cyber-capable models may be distributed.

OpenAI and Airbnb agreed to broaden Airbnb's internal access to OpenAI frontier models, including GPT-6 Astra, through OpenAI APIs and Amazon Bedrock. Amazon upgraded Seller Assistant into an always-on agentic system for merchants and added a plugin that brings its data into Amazon Quick and, in beta, Anthropic's Claude. Meta announced tethered VR glasses for spring 2027 with Meta AI built into the operating system.