OpenAI announces multi-year AWS partnership ($38B compute commitment); Microsoft details $15.2B UAE investment plan through 2029
November 3, 2025000
- OpenAI–AWS: multi-year strategic partnership announced; OpenAI to run advanced AI workloads on AWS infrastructure starting immediately
- Agreement size: $38B compute commitment over ~7 years; capacity targeted to be fully deployed before end of 2026 with ability to expand into 2027+
- Infrastructure scope: Amazon EC2 UltraServers with clusters using NVIDIA GB200 and GB300 GPUs; “hundreds of thousands” of chips initially; ability to scale to tens of millions of CPUs; workloads cover ChatGPT inference and next-gen model training
- Context: follows OpenAI’s availability of open-weight foundation models on Amazon Bedrock; cited customer adopters include Comscore, Peloton, Thomson Reuters, Triomics, Verana Health, among others
- Microsoft–UAE investment: Microsoft discloses $15.2B USD total spend in the UAE from 2023–2029 tied to AI/cloud expansion and G42 partnership
- Breakdown to end-2025: just over $7.3B spent comprising $1.5B equity investment in G42; >$4.6B capex for advanced AI/cloud datacenters; >$1.2B local opex/COGS
- 2026–2029 plan: >$7.9B additional spend including >$5.5B capex for continued AI/cloud build-out and almost $2.4B local opex/COGS
- AI infrastructure detail: licensed and accumulated in UAE the equivalent of 21,500 Nvidia A100 GPUs (mix of A100/H100/H200 previously); new U.S. export licenses approved in September enable shipment of the equivalent of 60,400 additional A100-class units using Nvidia GB300 GPUs
- Talent & operations: nearly 1,000 Microsoft UAE employees; ~100 engineers; Emirati partner ecosystem ~1,400 firms employing ~45,000 professionals; Global Engineering Development Center opened in Abu Dhabi; AI for Good Lab established
- Skilling: initiative to skill 1,000,000 people in UAE by end-2027; recent programs include upskilling 120,000 federal and local government employees; plans to skill 175,000 students and 39,000 teachers