Microsoft Frontier Company will embed 6,000 industry and engineering experts to co-design, deploy and keep improving AI systems — measured on business outcomes, not billable hours. Early work spans LSEG, Unilever, Novo Nordisk, etc.

This is now an industry-wide arms race — embedded delivery is table stakes. Every hyperscaler and frontier model companies is building AI deployment capacity.

The real difference is the role partners play

• AWS: $1B into a direct, in-house engineering organization. Small pods of 5-6 embedded with customers. Partners come in for model expertise, vertical depth and capability gaps. Two weeks earlier (mid-June), AWS cut Marketplace fees for professional services from 2.5% to 0.5%, nudging more partner-delivered work through AWS.

• Google Cloud: $750M into its 120,000-member partner ecosystem for agentic AI. The center of gravity is the ecosystem — SIs, software and channel partners — with Google engineers embedded alongside major SIs.

• Microsoft: $2.5B into a new operating business. It goes direct with customers while naming Accenture, Capgemini, EY, KPMG and PwC as the partners it will scale through.

Microsoft may have a commercial twist:

CNBC reports the unit also folds in salespeople with industry experience, and the leader chosen to run it, Rodrigo Kede Lima, spent 6 years running enterprise transformation as a sales leader across the Americas and Asia.

The ambition may be bigger than deployment: help customers build with AI, then grow with it.

Across these programs customer keeps the IP, and the data stays put

AWS runs the work inside the customer’s own account, so data never leaves their governance.

Microsoft promises a customer’s data and IP never train models that would commoditize their edge — and stays model-neutral, running OpenAI, Anthropic, its own or open-source models so customers are not locked.

This is a new phase.

Hyperscalers are competing to pull enterprises into production AI faster, and to leave the intelligence with the customer.

Whoever helps them get there — SIs, software or channel partners — will get closer to these accounts.

Lessons for alliance leaders:

1. Co-build is now core AI go-to-market across every cloud — from in-house to partner-led

2. Engineering, cloud sales, SIs and marketplace routes now meet in the same account. Your product or service has to be easy to attach in that room

3. Build relationships with the people deploying AI, not only the people buying or selling it

Which approach shapes enterprise AI implementation fastest — direct, partner-led, or hybrid?

Source: Microsoft