Microsoft and OpenAI are one of the hottest power couples in tech right now. For those who don't remember, Microsoft invested in OpenAI, the company behind ChatGPT, and OpenAI helped Microsoft power its AI search engine and other programs. But people familiar with the matter say the relationship has its conflicts behind the scenes. WSJ reporters, Tom Dotan and Deepa Seetharaman, wrote about this, and Tom joins me now. Tom, can you tell us a bit more about this partnership?

Tom Dotan: OpenAI is this company that was launched in 2015 that has been focused on building safe artificial intelligence. Microsoft, in 2019, first put in a billion dollars and has since put in billions and billions more to help this company develop its artificial intelligence technology, including generative AI, which is this technology that we know through ChatGPT in which you can take human prompts and spit out generative answers, things like text or images and things like that. Microsoft has taken this tech and incorporated into a ton of its products, so we've seen Bing. They're now incorporating it into a software that will sit across all of its office software, like Microsoft Office, Word, Excel, things like that. You've seen Microsoft basically take the technology that was developed by OpenAI and reorient the entire company around this tech to enliven all of its software products. In the midst of using its technology, there's also been this interesting relationship between these companies.

Zoe Thomas: Well, let's talk a little bit more about that because every relationship, friendship, partnership always has some struggles. So from the Microsoft perspective, what are those?

Tom Dotan: Well, I want to say at the outset here that, by and large, these guys at this current moment are getting along pretty well. It has benefited both the companies. OpenAI has gotten billions of dollars. Microsoft has gotten early access to what is being hailed as transformative technology. At the same time, Microsoft doesn't own the company. It owns only 49% of it, and that's a bit complicated. You don't see things like that happening too often in the business world. Typically, when a big company wants to utilize this transformative, groundbreaking technology, you would just absorb the company. You would just acquire it. But for kind of complicated reasons, partly the fact that Microsoft, if it were to acquire this company, could trigger some regulatory scrutiny. Also, OpenAI doesn't really want to sell itself. It doesn't own the company, which means it can't control all the products that it's working on. It can't control all of the initiatives that they're involved with. So in one sense, it's a huge part of Microsoft's business, but it also can't really tell OpenAI what to do.

Zoe Thomas: Tom, you've been speaking with people familiar with the matter. Can you give us an example of where this tension between the two companies has come to the forefront?

Tom Dotan: When Microsoft began incorporating and building the tools to incorporate the generative AI GPT technology into Bing and basically revamping Bing, there were some warnings from the OpenAI folks that were saying, "Look, you guys are using this underlying technology called GPT-4. It hasn't been fully trained yet in terms of dealing with all the different kinds of responses that this generative AI tech needs to function a lot more reliably." They brought up to them the possibility that this tech could give you some weird answers until it's more fully trained. Ultimately, Microsoft heard their warnings and said, "Yeah, but we feel good enough about where this is right now." An argument could be made that that OpenAI was right about a lot of things. If you followed the coverage, a couple of days after Bing launched its new chat tool, some reporters and users were able to get it to say some pretty bizarre stuff. These are the kind of responses that the OpenAI people were saying were possible until the technology was more fully trained.

Zoe Thomas: Microsoft has its own AI team. What has this partnership meant for them?

Tom Dotan: Once the OpenAI investment started growing and growing, Microsoft needed to redirect more of its resources to making sure that OpenAI's tech was given priority, because it is the priority right now. So if you were someone within Microsoft's research division and you were working on artificial intelligence, you're basically playing, if not second fiddle to OpenAI, then at least sort of subservient to the products that they're working on. So there's been a not small number of grumbles from people inside Microsoft's research team saying, "We don't have the resources that we used to get now because a substantial portion of that has been redirected toward OpenAI."

Zoe Thomas: One of the challenges to me seems to be that both Microsoft and OpenAI have similar products that they're selling to customers. Is there a lot of overlap between what these two partners are trying to make money off of?

Tom Dotan: Totally. Microsoft, in addition to being this infrastructure layer within the internet, they have Azure, which is a cloud service provider, which OpenAI runs on top of, they make software. OpenAI is a creator of technology that it sells to companies that compete with Microsoft that sell the same software. So in the story, we have this example of Salesforce, which is this big customer relations management software maker, and their software, one of their new features is powered by OpenAI. But Microsoft also sells this kind of software, and it has a feature very similar to what Salesforce has put out there that is also powered by AI.

Zoe Thomas: What happens next for these two partners?

Tom Dotan: You could imagine a scenario in the next couple of years where OpenAI software is the preeminent and predominant AI foundational layer for a lot of companies out there, and it's continuing to empower more and more of Microsoft's competitors. Or Microsoft decides to build its technology in-house. I want to be clear, we're not saying that that's the case right now. But this is an unusual partnership, and typically unusual partnerships end up being very complicated as time goes on.