Microsoft will invite select businesses to use GPT-3 as an Azure tool; Open AI will continue selling its own GPT-3 API with the latest upgrades
Context & Ripple Effects
A year after Microsoft's exclusive license of GPT-3, the arrangement is taking its real shape: rather than hoarding the model, Microsoft is opening gated access to select businesses through Azure, while OpenAI keeps selling its own API directly with the newest upgrades. That two-track split — enterprise distribution through the cloud, cutting-edge access through the lab — is the template both companies have been building toward since Microsoft first put GPT-3 to work internally in Power Apps' natural-language-to-code feature.
The invitation-only framing matters because it previews what came next: this gated Azure tooling matured into the broadly available Azure OpenAI Service carrying GPT-3.5, DALL-E 2, and eventually ChatGPT. This announcement is the hinge where GPT-3 stops being a research curiosity and starts being sold as cloud infrastructure.
First-order effects
- Select businesses gain GPT-3 access wrapped in Azure's enterprise terms, while OpenAI's direct API customers keep first claim on the latest model upgrades — two customer bases, one model.
Second-order effects
- Enterprise buyers must now choose between Azure's stability-and-compliance wrapper and OpenAI's faster upgrade cadence, forcing each channel to compete on terms rather than on the model itself.
Third-order effects
- If the pattern holds as it did through 2023, frontier models become a standard part of hyperscaler cloud catalogs — with the lab retaining a direct channel for early adopters and the cloud owning the enterprise mainstream.
The trend: Frontier AI models are being platformized into hyperscaler cloud services under exclusive licenses, splitting distribution between the lab's direct API and the cloud partner's enterprise channel.