Some people might be surprised that MS would pick the product with the best technological fit rather than the one they already have a deep business and financial relationship with.

Surely Microsoft's expertise these days is in cross-selling passable but not best-in-class products to enterprises who already pay for Microsoft products.

It says something about how they view the AI coding market, or perhaps the level of the gap between Anthropic and OpenAI here, that they've gone the other way.

▲

dijit 2 days ago | parent | next [-]

They are right to be surprised.

Why is Azure popular? Not on its own merits, it's because there is a pre-existing relationship with Microsoft.

Why is Teams the most widely used chat tool? Certainly not because it's good.. it is, again, pre-existing business relationships.

Seems odd for a company that survives (perhaps even thrives) on these kinds of intertwined business reasons to, themselves, understand that they should go for merit instead.

▲

mikestorrent 2 days ago | parent | next [-]

Yep. Similarly, Microsoft Entra... if you want Office, you're getting it anyway. Might as well use it for SSO, right? And here's your free Teams license... how can you justify paying for Slack when we've a perfectly good chat client at home?

	▲	dijit 2 days ago \| parent [-]
		I tried for a while to get Entra working with an external identity provider (Google Workspace). The other way around worked (Google could use Entra) but it was basically impossible to backend Entra from Google. Weird.

▲

vehemenz 2 days ago | parent | prev [-]

Except nobody chooses M365 Copilot over ChatGPT or Claude, so clearly the usual reasons aren't working. In this case, improving the product via integration is a last resort.

▲

thewebguyd 2 days ago | parent | prev [-]

> It says something about how they view the AI coding market

I think Microsoft views models as a commodity and they'd rather lean into their strengths as a tool maker, so this is Microsoft putting themselves into a position to make tools around/for any AI/LLM model, not just ones they have a partnership with.

Honestly I think this sort of agnosticism around AI will work out well for them.