It made for a fascinating story.
A large, slow-moving incumbent (in this case Google) was swept off its feet by a more nimble upstart named OpenAI. It was the type of story that would build a Silicon Valley legend.
In his first letter to shareholders after becoming CEO in 2016, Sundar Pichai said Google would move to an “AI-first world.” But in late 2022, a relatively unknown company suddenly pulled the rug from under Google’s feet. ChatGPT became an overnight sensation, and Google, which had long considered itself the leader in AI, seemed to be falling behind.
In the years since then, the search giant has taken swift action to turn the situation around. The company quickly restructured key parts of its business internally and aggressively pushed to bring generative AI into its core products.
The company’s latest AI model, Gemini 3, was launched this month to amazing reviews. Importantly, Gemini 3 launched on day one in Google’s most important business: Search. This was a strong sign of the company’s success after three years.
But there are big questions Google is still grappling with as it rebuilds its image in the age of AI. OpenAI’s first-mover advantage means ChatGPT has become the “Kleenex” of AI (a name everyone associates), and it will be difficult for Google to change that.
There are also more existential questions. First of all, how can Google maintain a healthy web ecosystem after AI eats up the internet?To keep pace with its rivals, Google is reimagining search not as a directory but as a conversation that saves scrolling and clicking. The question of what that means for the future of the Internet persists. Still, Google’s business may not only survive self-destruction, but thrive.
“Google has users, and as long as they continue to rush out new products, no one can match their distribution tentacles,” said a former Google executive. “No matter what the scenario is, they will emerge as winners.”
Google told Business Insider in a statement: “We’ve experienced platform transitions before, and each one has been a moment of expansion for Google and search. We move quickly, innovate, and create new opportunities for our products, partners, and users. And that’s what we’re doing with AI.”
Gemini Rising
Did Google foresee that happening? It depends on who you ask.
When OpenAI’s ChatGPT was rolled out, Googlers felt a mixture of frustration and relief. Some employees had worked on similar projects that had never made it outside Google’s walls, and it was frustrating to see a much smaller competitor get the glory. Others believe this is the starting pistol in the AI race and the green light for Google to finally unleash its own AI on the world.
Google has never lacked brains or technology. Employees at DeepMind, the AI research lab within Google’s parent company, had been tinkering with extensive language models before ChatGPT arrived, but leaders worried that releasing a chatbot that was prone to bias and mistakes would spark a backlash.
“This wasn’t meant to be released publicly yet,” said a former DeepMind employee who worked on Google’s own LLM technology. “We discovered the flaws and realized it was starting to get interesting, but we weren’t sure how to make it fully useful.”
To the surprise of some leaders within the company, and insiders say Google ultimately miscalculated, users seemed to care little about ChatGPT’s shortcomings.
In 2023, Google merged its two AI labs to focus on building Gemini. Glenn Chapman/AFP via Getty Images
Inside DeepMind, the team was rushing to launch an answer to ChatGPT, a chatbot called Sparrow, which, according to people involved in its development, would have had more secure guardrails than ChatGPT. The project quickly died down when Google merged DeepMind with its in-house AI lab Brain, after which the company shifted its full focus to Gemini.
At the same time, Google made an internal push to bring generative AI to all of its most important products, including search, YouTube, and Android. We have released the Gemini chatbot multiple times, the latest of which is Gemini 3.
Exploring the future of Google
For the past three years, Google has made it seem pretty easy to destroy itself. The company just reported that it hit $100 billion in its first quarter. The company’s cloud business has long been second-tier to Amazon and Microsoft, but it’s rapidly gaining momentum. The company’s specialized AI chips, which had been a slow internal effort for years, are suddenly seeing big demand from companies competing for more computing power.
Current and former employees say Google’s biggest advantage is that it owns the complete technology stack needed to respond to the AI boom. The company offers a number of products that are already used by billions of people, including Gemini, Search, and Maps. We have a group of researchers building cutting-edge AI models. And in addition to powering all of these internal efforts, the company has a cloud infrastructure that its biggest competitors pay to use.
But search remains Google’s core business, a ship it has spent decades trying not to rock. When co-founders Larry Page and Sergey Brin founded Google, it didn’t have a clear business model beyond building the world’s leading search engine, but the founders eventually leaned toward placing ads in search results. It became one of the most scalable businesses ever imagined in the digital age, and continues to provide the largest portion of Google’s revenue to this day.
Liz Reid, Head of Google Search. Sajjad Hussain/Getty Images
Now, Google needs to decide whether AI summaries (generated summaries that provide instant answers to users’ questions) and a more dramatic rethinking of search in AI mode will be just as effective when it comes to search advertising.
Google can’t afford to stand still. Industry analysts at Business Insider sister company EMarketer predict that Google’s share of the search advertising market will fall below 50% for the first time next year (a revision of an earlier prediction that it would happen in 2025). EMarketer predicts that by the end of 2026, Google’s market share will be 48.9%.
Oscar Orozco, senior forecasting analyst at EMARKETER, said the continued strong performance of Google’s search business suggests that advertisers’ wariness about AI overview is “exaggerated” by the market.
“However, this remains a long-term threat to Google’s search business, and we believe that as LLM monetizes its search capabilities, Google will continue to lose market share, albeit at a slower rate than we expected at the same time last year.”
Google is reimagining the web as we know it
Google doesn’t seem worried, at least not publicly.
In a recent interview with the Wall Street Journal, Google search chief Liz Reid said that while changes to search have reduced traffic to some websites, the overall number of search queries has increased as Google has made it easier for people to find answers through tools like AI Overview and Lens.
“The increase in searches will offset some of the impact on ad clicks and end up at about the same level,” Reid said. He also argued that chatbots cannot replace the products users are trying to buy, suggesting that transactional search, which is more likely to generate advertising revenue for the platform, is Google’s top priority.
Even that is changing at Google. As generative AI gets people to what they want faster, the search funnel (the time it takes to start researching a product and make a “purchase”) is shrinking.
Google has begun to show the world how it is adapting to protect its core revenue stream. But it’s unclear what the generative AI boom will mean for the long-term health of the web, and how it will affect websites that provide information rather than products.
Google says its new AI search feature helps users find more relevant content, and queries continue to increase year over year. But many publishers are already seeing a decline in search traffic, raising concerns that even if AI summaries increase overall searches, they won’t translate into clicks on their websites.
A recent Pew Research study based on the search history of 900 users concluded that users were 8% likely to click on a traditional link when presented with an AI summary, compared to 15% when the AI summary was not presented. Google said the study was flawed and not representative of search traffic as a whole.
Lily Ray, vice president at SEO research firm Amsiv, said the move to an AI-first Google will make it harder for publishers and content creators who build much of the web to make money. If people stop publishing, where will Google get the information it needs to build its AI summaries?
“I don’t know how they’re going to solve it. Their current excuses won’t solve it,” she says.
One idea she proposed was for Google and other AI companies that build large language models to pay commissions on the sites they use for training and referencing, as an incentive to continue development.
She’s not alone in feeling that something needs to change.
“The web business model is not going to survive without some change,” Cloudflare CEO Matthew Prince said in an interview with the Council on Foreign Relations earlier this year. Prince, whose business is a gateway between websites and the internet, calls AI answerbots an “existential threat” to the internet.
“If content creators can’t derive value from what they’re doing, they’re not going to create original content,” Prince said.
