AI and search: Could AI kill Google?

“Google has long been the gateway to the internet and ChatGPT has shown us that might not always be the case. At Balderton, we believe that Search is one of the most exciting areas in AI today.”

That was our take in December 2022, a month after ChatGPT’s release.

Since then, AI has evolved at a rapid pace – transforming industries like sales and coding, with AI agents poised to tackle an even broader range of use cases. When it comes to search, we’re now seeing evidence of structural shift – and for the first time in 15 years, Google’s market share has dropped below 90%.

This is Part 1 of a series where we’ll explore the dynamics of the search market and how AI is changing it. In Part 2, we’ll take a closer look at some of the opportunities we’re seeing for startups in the market.

“Search is a behaviour, not a channel”

Over the last 20-25 years, Google has dominated the world of search so much that it has literally become a verb of the same meaning.

But search actually refers to a user behaviour, and it tends to follow a three-step process:

1. The user starts out with an intent, which can be categorised into one of three types: Navigational – the user is looking to go to a specific site Transactional – the user is looking to purchase something Informational – the user is looking become knowledgeable about something
2. The user collects information according to his intent
3. The user makes a decision to end the search (” I’m at the website i was looking for”, ”successful purchase” or “i don’t want to buy something”, “I’ve learned enough”)

Search engines therefore must support the user throughout this process by:

1. Understanding the user’s intent
2. Providing accurate and relevant information according to the intent
3. Presenting this information in a clear and actionable manner

So how can AI improve existing search?

If we look closely at each of the key steps performed by search engines, it’s clear that there are numerous ways in which AI has the potential to transform search as we know it.

1. AI can improve understanding of the user’s intent:

AI can understand user intent far more accurately than traditional keyword-based search, which often struggles with complex or ambiguous queries. This is because AI-first search engines leverage richer context from individual user interactions. In the past, Google and other search engines had to infer intent from short, three-to-four-word queries. Now, users engage in longer, more natural conversations with follow-up questions, allowing AI to refine its understanding. Additionally, AI can process multiple input types—text, voice, and visual.

Understanding intent is probably the most important piece of the search puzzle. If you fail to truly understand what the user is looking for from the outset, it’s unlikely you’ll be able to meet their needs throughout the search process. We’ll dive a little deeper into this shortly.

2. AI can provide more relevant and accurate answers to queries:

One of AI’s greatest strengths is the ability to process and synthesise vast amounts of information. Unlike traditional search engines that return a list of links, AI-powered search engines like ChatGPT and Perplexity generate direct, contextualised responses tailored to user needs.

Anyone who has used ChatGPT over Google for a search query will have felt this first-hand and can attest to just how much better the responses are. It’s not surprising that consumers are already increasingly using ChatGPT (which sees over 1bn queries a day) over traditional players, particularly when it comes to informational queries.

3. AI can offer better interfaces, presenting information in a clearer and more actionable way:

The way search results are presented dramatically impacts usability. AI can enhance interfaces by making interactions more conversational, dynamic, and personalised. A key part of this is AI’s ability to return context-aware follow-up questions, guiding users to refine their queries or explore relevant related topics.

What’s more, the rise of AI agents unlocks new opportunities at this stage, with autonomous web browsers now able to perform multi-step searches on behalf of users. For example, instead of manually clicking through multiple links when looking to compare products, an AI assistant could go through these steps for you – entering your specific requirements to obtain personalised quotes from each provider, looking at relevant reviews, and providing a full summary of the most suitable options.

Over the past decade, innovations focused on just one of these pillars have created large category winners. For example, social feeds created entirely new interfaces (pillar #3), allowing Meta, Tiktok etc to build global giants. Meanwhile, other players focused on providing better, more relevant answers (pillar #2), creating a generation of vertical search winners – for example companies such as Zillow ($14B market cap), which was one of the first search engines that had access to real estate listings across the US.  As AI enters the scene, we can expect to see a total transformation across each of these key stages of search, fundamentally reshaping how people discover and interact with information online. And it all starts with pillar #1, intent…

Intent – Is AI just making search incumbents better?

The reason for the dominance of search giants like Google and Meta is their deep understanding of user intent. It is fundamental to their business models.

If you look at Google’s model, in particular, revolves around two key tasks:

1. Delivering the most relevant content based on user intent.
2. Connecting businesses with high-intent users through advertising.

While only the second pillar is monetised, the first is crucial—it fuels Google’s massive user base, which in turn attracts advertisers. Notably, 80% of Google searches display no ads, highlighting the company’s priority on user experience to maintain engagement.

Because of this, Google has invested vast resources to stay ahead in understanding intent. They have an immense data advantage—processing over 8.5 billion searches per day—along with unparalleled computational power, making the giant nearly untouchable.

Today, AI can unlock more nuanced, context-rich search interactions, leading to significantly more accurate interpretations of user intent. Incumbents like Google have been quick to react to this step shift (for example with the launch of Google’s AI Overviews and the newly announced Google AI mode) and with their existing data, vast compute resources and own proprietary models, the incumbents seem to be in an ideal position to leverage this new way of ‘harvesting’ intent. Not only can AI Overviews provide more tailored responses suited to the intent of the user, it can also be used to increase the intensity of a user’s buying intent, in turn increasing the potential value of each click (unlocking yet more advertising spend).

Let’s look at how this works in practice:

In the example above, based on my search, Google’s AI Overview highlights key criteria for buying shoes, and then goes one step further and links to sellers that offer shoes matching those criteria.

Notice how I started by searching for what to consider when buying shoes—an indication of low to medium purchase intent. AI Overview then argued that shoes with high brand reliability would be the best option and immediately suggested a relevant website.

If I click on that website, my intent to buy will be stronger than if I had simply clicked on a standard text link without the AI Overview. Advertisers will be willing to pay more for such a click.

How do you make money with search?

Broadly speaking, informational searches generate revenue through subscriptions (e.g., ChatGPT, Perplexity) or not at all (Google), while transactional searches are monetised primarily through ads (Google, Meta, TikTok).

Since the early days of internet search in the 1990s, advertising models have continuously evolved—from Yahoo’s early display ads to Google’s search ads and, later, feed-based ads on platforms like Instagram and TikTok. Over time, ads have become increasingly context-rich: display ads lack context, search ads are intent-driven, and feed ads offer the highest level of personalisation and contextual relevance.

AI is set to disrupt this landscape once again: Richer contexts, new interfaces, reduced reliance on clicks, and AI-driven search experiences could challenge existing ad models, giving founders a great opportunity to innovate here.

This innovation of ad models could be a challenge for Google. Their core revenue model is built on search ads, earning money every time a user clicks on an advertised link. While AI can increase the value of each individual click, it could also dramatically reduce click volume—or even eliminate clicks altogether, as seen with ChatGPT-style interactions. Whether the greater value per click can offset this decline is unclear and Google might end up in a dilemma where they need to adapt to new AI search and revenue paradigms without cannibalising their current way of making money.

Conclusion

AI will cause a lot of turmoil in search, one of the largest industries of the global economy. While incumbents have certain advantages, there is potential for a truly category defining company based on the outlined innovator’s dilemma. In part two of this series, we’ll look a little closer at some of the specific companies and opportunities we’re seeing in the market.

If you’re a founder building in the space we’d love to hear from you, feel free to reach out to [email protected]