Microsoft hosts its annual software developer conference in Seattle on Monday, drawing thousands of coders looking to turn the past years of investments into artificial intelligence into profitable products and services for consumers and businesses.

The Redmond, Washington-based software giant, which is an investor and deep strategic partner with ChatGPT creator OpenAI, and has already spent $64 billion this year, much of it on data centers needed for AI-based services such as Copilot used in its popular Microsoft 365 applications.

But there are signs that Microsoft -- whose shares are up more than 30% this year, defying a broader Nasdaq decline -- is reworking its relationship with OpenAI and seeking to become a neutral arms dealer in the AI race, Reuters reported.

Earlier this year, Microsoft allowed OpenAI to branch out and work with Oracle on the massive "Stargate" data center project in Texas.

Meanwhile, CEO Satya Nadella has argued the company can get expenses down, saying that once it settles on an algorithm and begins to optimize it, Microsoft can obtain 10 times better performance for the same computing costs.

Demand for AI services in Microsoft's Azure cloud computing is also continuing to grow.

Thomas Blakey, an equity analyst with Cantor Fitzgerald, said that the company is increasingly keeping revenue-generating AI services inside its own data centers, where it can continue to tweak them for better cost.

It shifting to only using outside data center services such as CoreWeave, which is known as a "neocloud" that focuses on offering AI chips from Nvidia, when Microsoft needs short bursts of extra computing power for specific projects.

"If they have to flex up in some way, they've been consistently saying that they're going to shift away from buying more data centers and dirt and cement and they're going to leave that to the neoclouds," Blakey told Reuters.