Microsoft launched a $2.5 billion business unit called Microsoft Frontier Company to help enterprises adopt multi-model AI strategies rather than locking into a single provider. The move follows Microsoft's own admission that binding Copilot exclusively to OpenAI models was a mistake. As models like DeepSeek and Gemini have caught up to GPT, enterprises increasingly route different tasks to different models based on cost, speed, context window, and data residency requirements. This shift is driving demand for AI gateways, orchestration frameworks, and routing infrastructure — tools like LiteLLM, Portkey, LangChain, and cloud-native platforms like Amazon Bedrock and Azure AI Foundry. The broader trend: models are becoming replaceable components behind an orchestration layer, and controlling that routing layer is the new competitive moat.

5m read timeFrom thenewstack.io
Post cover image
Table of contents
Betting $2.5 billion on flexibilityOne model no longer fitsAI gateways become core infrastructureThat changes what developers buildThe ecosystem is already respondingOrchestration is the new moat
155 Impressions