Sundar Pichai says Google’s AI models are competitive, but it’s the product that is the problem.

At The New York Times’ DealBook Summit, Google CEO Sundar Pichai spoke with host Andrew Ross Sorkin about the company’s AI work. At one point, Sorkin asked Pichai how Google compares to its competition, especially Microsoft. Previously, Microsoft’s CEO, Satya Nadella, had said that “Google should have been the default winner in the world of big tech’s AI race.”

The implication is that Google is behind despite having both the talent and the technology. “They’re the vertically integrated player in this,” Nadella said. “They have everything from data to silicon to models to products and distribution.”

Of course, Google isn’t the default winner. I’m not sure anyone can say who will win or who is winning. Still, I don’t think Google would be at the top of anyone’s list, even though Google played a significant role in developing much of the technology used today in generative AI chatbots.

“You guys were the originals when it comes to AI.” Sorkin pointed out. “Where [do] you think you are in the journey relative to these other players?”

Pichai told Sorkin he would love to see “a side-by-side comparison of Microsoft’s models and our models any day, any time.” Then, Pichai pointed out that Microsoft is “using someone else’s models.”

With those four words, Google’s CEO took an uncharacteristically large shot at one of his biggest competitors. That said, it’s not an entirely invalid point. Microsoft has partnered with OpenAI and incorporated its generative AI product into its software. In many ways, it has tied its entire business strategy to what is arguably a pretty chaotic partner.

You’ll remember that it was just over a year ago that OpenAI fired CEO Sam Altman before Nadella got involved and the board rehired Altman. The company is now trying to shift from its current non-profit structure to a for-profit corporation. Along the way, it has lost significant talent and is embroiled in a lawsuit with co-founder Elon Musk, who is working on building his own generative AI company.

It’s not that Microsoft isn’t working on its own models, but it partnered with OpenAI to jumpstart its efforts. I don’t know whether that’s paid off yet, but the company has a lot riding on someone else’s success. That seems a bit risky, and it’s a valid criticism from Pichai.

Of course, the thing about generative AI isn’t necessarily the models. The problem is the product. Look no further than the fact that Anthropic’s models are widely considered better than OpenAI’s GPT 4o, but ChatGPT is a far more widely used product.

Google’s biggest problem isn’t that it can’t build competitive models; it’s that it hasn’t figured out how to build compelling products that won’t destroy its existing search business. Microsoft doesn’t have that problem. Sure, Bing exists, but it’s not a significant enough business to matter, and Microsoft is happy to replace it with whatever its generative experience might look like for search.

All of that is to say that Pichai might be right that Microsoft is using someone else’s models, and he might be right that Google’s models stack up just fine, but the reality is none of that matters. Microsoft may be putting all of its eggs in Sam Altman’s slightly chaotic basket, but Google’s eggs are already baked into a search engine-shaped soufflé.

Google hasn’t figured out the product part. It hasn’t figured out how to turn Gemini into a product at the same scale as search without killing its real business. Until it does, it doesn’t matter whether the competition uses someone else’s models.