Google built the tech behind ChatGPT, then watched OpenAI launch it

Google became one of the world’s most valuable companies by changing how people find information. Now AI rivals are challenging that dominance and pushing Google to spend heavily to secure its future. Can it embrace the disruptive technology without breaking the business that made it rich?
Marissa Mayer, a former Google executive, remembers the company as a scrappy startup in a small office above a bike shop in Palo Alto, California. She joined in 1999, fresh out of nearby Stanford with a master’s in computer science. “It’s pretty wild. When I interviewed at Google, there were just seven people there,” Mayer told Planet Money. By the time she was hired a month or two later, she was its 20th employee.
That year, Google brought in just over $200,000 in annual revenue. By 2022, the eve of the AI boom, the company — by then reorganized under a parent called Alphabet — was generating more than $280 billion. Google made its fortune selling ads next to search results, and its search business became an unstoppable money machine. Its share of the U.S. search-engine market grew from 80% in 2009 to nearly 90% in 2020.
In 2012, Mayer left to become CEO of Yahoo, trying to revive a competitor Google had blown out of the water. Neither Yahoo nor any other rival dethroned Google. For a time, it seemed only politicians or antitrust lawyers could cut the company down to size. But the biggest threat may have come from inside its own labs.
Google plowed search profits into AI research. In 2017, a team of eight Google employees developed the transformer, the breakthrough that made today’s AI chatbots possible. Mayer, now CEO of the AI startup Dazzle and still a Google shareholder, sees that work as reason for optimism. “This is not a technology that’s happening to Google,” she said. “They helped invent this new technology.”
Yet Google is now spending enormous sums to build its AI capabilities and fend off rivals, and overhauling search to answer questions directly with AI instead of just providing links. Wall Street is nervous. In late October, Alphabet released its latest earnings report, and its stock fell more than 12% the next day — the biggest single-day percentage decline since the company went public in 2004.
The late Harvard Business School professor Clayton Christensen laid out a classic theory in his 1997 book, The Innovator’s Dilemma: When New Technologies Cause Great Firms to Fail. He theorized that the very thing that makes a business wildly successful can lead to its downfall in the face of a disruptive technology. Embrace it, and you risk today’s profits. Fail to embrace it, and you may risk the company’s health tomorrow.
Google didn’t just invent the transformer. Its researchers made computers translate languages, recognize speech and images, drive cars autonomously, and beat one of the world’s best players at Go. It developed an internal chatbot called Meena that employees played with years before chatbots became a sensation. But it didn’t release it.
Then, on Nov. 30, 2022, OpenAI released ChatGPT, offering a new way to find information and a glimpse of a future where people didn’t need Google to find answers. Google scrambled to respond, reportedly declaring a “code red” as ChatGPT captivated the masses.
Why wasn’t Google first out of the gate? Christensen’s theory offers one explanation: when you’re making tons of money from search, why rush to release something that could undermine it? A company can get too focused on the best way to milk its cash cow, and that focus can create a kind of myopia.
Last year, Google’s chief scientist, Jeff Dean, offered an explanation for why Google hadn’t released something like ChatGPT sooner. Dean told podcaster Dwarkesh Patel that they looked at their existing chatbots through the lens of their search business. “From a search perspective, you want to get the right answer 100% of the time,” Dean said. “These models were not near that bar.” They were prone to hallucinating and saying offensive things.
ChatGPT had those problems too, making mistakes and saying racist, sexist and other incendiary things that made headlines. Yet it took off like a rocket ship. People didn’t just use it to find information; they used it for things like drafting a note or summarizing text. Christensen warned about this trap, writing that companies that keep disruptive technologies “bottled up in their labs” risk losing out to rivals willing to run with them, however imperfect.
Mayer, however, questions whether releasing a chatbot first would have gone as well for Google as it did for OpenAI. “The standards for an established company like Google in terms of


