Meta Platforms has emerged as one of Microsoft’s largest AI customers, reportedly spending hundreds of millions of dollars annually on AI models through Microsoft’s Azure cloud service. The social media giant processes trillions of AI tokens through Azure each week, underscoring the substantial investment in external AI infrastructure and the concentrated demand for AI among major technology companies.
This engagement is part of Meta’s broader strategy to leverage external AI models alongside developing its own proprietary systems. Meta procures access to various models through multiple platforms, prioritising availability and cost.
Its developers have also utilised OpenAI technology via Microsoft’s Foundry marketplace to benchmark the performance of Meta’s in-house models, according to individuals familiar with the matter. Andrew Bosworth, Meta’s Chief Technology Officer, has affirmed that the company rents leading AI models as an integral part of its development process, even as it builds its own models and AI infrastructure.
Microsoft’s Foundry marketplace, which offers models from multiple providers, reported 100,000 customers as of July. Despite its efforts to serve diverse sectors like manufacturing and transportation, a significant portion of its largest AI customers remain technology firms.
ByteDance, the owner of TikTok, has generally been Foundry’s largest spender, with Adobe and AI companies like Perplexity and Sierra also being major clients. Microsoft’s AI business shows a notable concentration, with its long-standing partner OpenAI accounting for approximately 70% of its total AI revenue in the most recent fiscal year.
Beyond Foundry, Microsoft also generates AI revenue from products such as Copilot and by providing computing capacity for AI workloads.
Concurrently, Meta is developing its own API business designed to provide customers with access to various AI models. This service could eventually position Meta as a direct competitor to Microsoft Foundry, potentially reducing Meta’s reliance on external platforms for its AI needs.
Meta has historically relied on Microsoft technology for certain product components before transitioning to in-house systems, such such as Bing supporting Facebook web searches before Meta ceased its use by late 2014, and Microsoft supplying computing capacity for Meta’s AI development prior to the current AI boom.
Published On Aug 22, 2026 at 06:25 AM IST
Join the community of 2M+ industry professionals.
Subscribe to Newsletter to get latest insights & analysis in your inbox.
Get updates on your preferred social platform
Follow us for the latest news, insider access to events and more.