Meta Is Using Microsoft Foundry at Massive Scale, and the AI Cloud Race Just Got More Complicated
Meta is using Microsoft Foundry at massive scale to access outside AI models, according to a new Bloomberg report, even as it keeps building increasingly capable AI systems of its own.
The more interesting question is what happens when one of Microsoft’s biggest AI customers is also laying the groundwork to become a rival cloud-model provider.
➜ The real story: Meta using Microsoft infrastructure isn’t a contradiction.
It’s a preview of an AI market where even the companies building frontier models still shop around when another platform has the model, capacity or speed they need.
This guide breaks down what Bloomberg reported, what the report suggests about Meta’s development strategy and why Microsoft Foundry is becoming a key piece of the wider AI stack.
The article says Meta developers are using trillions of tokens each week through Microsoft’s cloud platform, including access to outside models that can help evaluate Meta’s own AI output.
Why Meta Is Using Microsoft Foundry Even While Building Its Own AI
Meta already develops its own AI models, but building models and using models are two different jobs.
Bloomberg reported that Meta taps multiple platforms depending on model availability and what works best for a given task, which gives engineers more freedom instead of forcing every workload through one internal system.
See more AI and Technology coverage.
That also lines up with how Microsoft positions Foundry.
The Azure service is designed as a broad model marketplace and development platform, so companies can move between different AI providers without rebuilding their entire workflow each time.
For Meta, that can mean using outside models as a second opinion while its own systems are being tested.
What Meta and Microsoft Foundry Reveal About the New AI Cloud Race
The slightly awkward part is that Meta is also exploring its own API business for giving developers access to AI models.
In other words, Microsoft may be serving one of the companies that could eventually compete with Foundry for the same developers.
Tech partnerships have always had a little “we’re friends until the product roadmap says otherwise” energy, and AI is making that overlap much more common.
Meta has already been using AI to speed up software development and launch new products faster, a strategy Spacelab covered in its look at Meta’s wave of AI-built apps.
Microsoft, meanwhile, is pushing Foundry and Microsoft Copilot deeper into the AI platform layer.
The practical takeaway is simple: the AI cloud race isn’t settling around one model or one provider. It’s turning into a flexible stack where companies mix their own models with competitors’ tools whenever that gets the job done faster.
FAQs
Why is Meta using Microsoft Foundry if it builds its own AI models?
Meta can use outside models for testing, evaluation and specific workloads while continuing to develop its own systems. A multi-model approach gives its developers more flexibility.
What is Microsoft Foundry?
Microsoft Foundry is an Azure platform for building AI apps and agents while accessing models from multiple providers through one development environment.
Where can you learn more about Meta AI?
You can follow the company’s latest AI products, research and platform updates directly from Meta.
