Tom Warren

I think like the Microsoft AI stuff, they obviously had seven models and again, they were kind of like the smaller models, but then they had the reasoning one, which I think that's definitely a step forward from what they've been doing before because typically it's been very focused on specific tasks, whether it be like

images or like transcription, all that sort of stuff.

But this reasoning and the thinking model is a lot about math and code, right?

Which is where they've kind of been falling behind.

So like cloud code and cursor and all the rest of those coding tools.

So I think

They've definitely put a stake there.

They want to go aggressively after coding.

Yeah, exactly.

So I think they're trying to reverse that trend.

They've successfully done that, I think, the first step.

It's just the question of where do they go next.

And obviously Mustafa said that they want to be the top four.

How they go about that is going to be very interesting.

He actually said also to me that they're going to lean on OpenAI for the next few years, right?

Because obviously they've got access to their models for quite some years yet.

So they can lean on them for certain stuff whilst they basically...

compete right and then build their own and so that's that's kind of interesting um perspective which is it's unusual to hear that from uh an executive to be honest um yeah but it's kind of what we thought a lot long anyway really isn't it so um but yeah i think they've they've definitely they've definitely started to shift the this this sort of message around they are now funny thing out the whole keynote i think co-pilot was only mentioned five times

Without it being GitHub Copilot.

Yeah, that's the one thing out of the keynote.