Microsoft Copilot: goodbye OpenAI and Anthropic, MAI models arrive

What has changed: Microsoft cuts ties with OpenAI and Anthropic

Microsoft has kicked off a quiet yet major transition. Inside products like Excel and Outlook, OpenAI and Anthropic's AI models are being steadily swapped out for Microsoft's own MAI (Microsoft AI) models. According to reports by The Decoder , tens of thousands of weekly queries already flow through these new internal models.

Mustafa Suleyman, Microsoft's AI head, announced the intention to "permanently eliminate" the cost of external models. Therefore, the direction is clear: reduce dependence on third-party providers and lower the marginal cost of every AI interaction. However, this financial optimization raises a direct question for corporate users: will performance remain unchanged?

The MAI model: internal architecture and strategic positioning

MAI models are not totally new. Microsoft has been developing them in-house for a while, focusing on computing efficiency and native integration with the Microsoft 365 ecosystem. In fact, their adoption in products like Excel and Outlook points to specific optimization for structured tasks: data analysis, email summarization, and formula generation.

However, OpenAI's models — particularly GPT-4o — and Anthropic's like Claude 3 were chosen in the past precisely for their complex reasoning and natural language processing capabilities. In contrast, MAI models seem to prioritize speed and cost per query over processing depth. This trade-off is not neutral for companies using Copilot for high cognitive value tasks.

We at SHM Studio we carefully monitor these types of developments. In particular, we evaluate how infrastructural changes in AI models impact the artificial intelligence services that we integrate for our clients.

Immediate impact for companies using Copilot

For marketing and IT managers, the operational question is concrete. Those who have integrated Copilot into their workflows — from drafting reports to managing campaigns, from analyzing sales data to summarizing briefs — might notice qualitative variations in the outputs. Furthermore, there is currently no official communication from Microsoft guaranteeing equal performance between old and new models.

According to Gartner , the quality of the underlying model is one of the critical factors when evaluating ROI for enterprise AI tools. As a result, even a partial drop in Copilot's capabilities could lead to real inefficiencies, especially for companies that have built automated workflows around these tools.

Beyond this, there's a transparency issue. Companies pay for a Copilot subscription without knowing precisely which model is processing their requests. Therefore, the silent swapping of models creates an AI governance challenge that IT leaders cannot ignore.

Still a work in progress: Microsoft between cost savings and reputation

Microsoft's move should be viewed in a broader context. Last year, the race to invest in AI put pressure on the margins of all major players. Microsoft, having invested billions in OpenAI, now finds itself having to balance operating costs with shareholder expectations. Therefore, internalizing models is a financially rational response.

However, the reputational risk is real. Copilot was sold as a tool powered by the best models available on the market. If the perception of quality were to drop, companies might reconsider adoption or reduce purchased seats. Similarly, competitors — from Google with Gemini to Salesforce with Einstein AI — could leverage this moment to position themselves as more transparent alternatives.

According to an analysis by Harvard Business Review , trust in the AI vendor is a crucial asset for enterprise organizations. Because of this, Microsoft will need to clearly communicate what MAI models can do to keep from losing its Copilot customer base.

What to do now: operational guidelines for marketing managers

First of all, it is a good idea to run an internal check of the workflows that rely on Copilot. In particular, we suggest spotting the tasks where the quality of the AI output is critical — for instance, content generation, semantic analysis, or the summarization of complex documents.

Next, it's a good idea to compare Copilot's output from the last few weeks with previous ones, to spot any changes in quality. This kind of audit, even an informal one, gives you solid data before making decisions on renewals or alternatives. Also, it's worth checking out if Microsoft will offer options to choose the underlying model for enterprise plans, like they already do in other API setups.

• Audit of AI-dependent workflows: map critical tasks that use Copilot daily.
• Qualitative benchmark: compare recent outputs with those produced in previous months on identical tasks.
• Evaluation of alternatives: consider tools like digital marketing solutions that integrate selectable AI models.
• Monitoring Microsoft communications: follow official updates on the MAI models roadmap.
• Internal AI governance: update AI usage policies to include evaluation criteria for the underlying model.

For companies using Copilot in LinkedIn campaigns or google ads campaigns , the quality of text generation is a factor directly tied to creative performance. Therefore, active monitoring is recommended right from the start.

Outlook: towards a more fragmented AI ecosystem

This transition by Microsoft is not an isolated case. Google has also developed its own Gemini models to reduce dependence on third-party suppliers. Thus, the enterprise AI market is evolving toward a model where major players prefer vertical integration over open partnership.

For Italian SMEs, this situation has direct impacts. First off, picking AI tools can't just be about the vendor's brand anymore; you've got to look at the underlying model and whether it actually fits your specific use cases. Plus, relying on just one ecosystem—like Microsoft 365 with Copilot—turns into a strategic risk if model performance isn't guaranteed in the contract.

Finally, there is a growing need for internal skills to evaluate the quality of AI outputs. Companies that invest in AI-assisted copywriting or in SEO strategies powered by language models need to be able to tell quality output from mediocre stuff, no matter the tool being used. To dive deeper into these topics, you can check out the SHM Studio blog or contact our team for a dedicated consultation.