Last week in Reuters: Microsoft is investing $2.5 billion in ...people.
With Microsoft Frontier Company, 6,000 employees will be deployed directly at customer organisations to help them select the right AI technologies and actually generate returns on their AI investments.
It sounds familiar, because that is exactly what we are seeing in practice.
The problem is rarely the model. It is the integration with a company's own data, which is often still of limited quality, or with applications and systems that are difficult to integrate. That is where many AI projects get stuck, not on model quality.
The gap between pilot and production also remains significant. PoCs are quick and easy to set up, but industrialising AI and embedding it into a day-to-day process is a different story.
And finally, companies do not want to be locked into a single vendor. Microsoft itself acknowledges that, three years ago, it was a mistake to tie Copilot exclusively to one model provider (OpenAI). Today, it is about flexibility: being able to switch between models depending on what works best.
So the business model is now shifting: no longer "here is the tool", but "we will build it with you". Amazon did the same last week with a $1 billion investment, and Anthropic and OpenAI have also set up similar teams this year.
Major players are acknowledging what everyone involved in AI implementation has known for some time: technology is not the bottleneck. Sufficient AI expertise, data and processes are.
Article via Reuters: https://www.reuters.com/business/retail-consumer/microsoft-launches-firm-help-companies-adopt-ai-with-25-billion-2026-07-02/