How AI is Transforming the Huge Search Market

Original Title: The Fragmentation of Search

Author: Grace Carney

The market structure of search is currently being reshaped. Horizontal and LLM-based "search agents" like ChatGPT and Perplexity are putting unprecedented pressure on search engines like Google. Cracks are beginning to form, and new search possibilities are rapidly emerging.

USV anticipates that this fragmentation will be an ongoing process, and as a result, a multitude of new opportunities will arise in the search domain. We are particularly interested in opportunities that empower consumers and web publishers, who have long been dissatisfied with the unfair Google model, or may be uneasy about search agents becoming the new gateways to our information.

Before ChatGPT, Google was synonymous with search. Its concentrated market power meant it could make unilateral decisions that could adversely affect consumers, web publishers, and advertisers.

We are actively seeking companies that have a clearly opposing stance to Google. DuckDuckGo is one such example, having built a search business that protects consumer privacy, currently generating over 3 billion searches per month. However, the Google search engine still dominates the market.

Now, LLM-based search agents are triggering various new transformations. By providing users with faster and better answers to their questions, without the need to scroll through pages or click on deceptive ads, search agents unlock a shortcut for consumers to precisely find information, completely obscuring web publishers. This is undoubtedly stripping query business from search engines like Google and Bing, while also replacing web publishers that provided training data for ChatGPT but did not receive traffic from it.

Perplexity, Arc Search, Pi, and many other similar companies are built on different LLM products, all vying to become trusted brands in the realm of general search agents. In the past, a single tab was enough to satisfy my internet search, but now I often open three or more tabs.

Before ChatGPT, we were like this:

Since using ChatGPT, we are like this:

This fragmentation of search raises some key questions:

• What kind of business models will general search agents devise to cover the financial deficits accumulated from computation and reasoning costs? Especially once the AI hype bubble bursts, the billions of dollars in venture capital supporting these platforms will also fade.

• Are consumers willing to pay for search? If the answer is no, how long will it be before targeted ads quietly appear in the responses of these search agents?

• If search engines disappear, along with SEO and advertising revenue, will web publishers and content creators face livelihood issues? Will we see the same problems that "print newspapers" faced when digital publications first emerged?

• If general search agents refuse to include web publishers in their economic models, what will their ultimate fate be?

• Even if search agents agree to pay web publishers AI royalties, if no one actually sees them (agents typically extract information from various web pages, summarize it, and provide the summary directly to users, thus obscuring the original pages), why would web publishers bother creating new sites? Will the introduction of economic incentives change the types of content we are accustomed to seeing on the internet?

All startups engaged in the search business will face these questions, and as a result, will continue to fragment into various new behaviors and platforms. Like DuckDuckGo, we are excited about those that can create business models that Google and Microsoft cannot compete with (unless they undergo self-revolution), thus forming opposing platforms.

We can envision these platforms rewarding web publishers and content creators to keep internet content vibrant and diverse. This could be achieved through economic rewards like subscriptions, although we believe that crypto has a greater opportunity to create new monetization models for AI-native content. Cryptocurrencies, NFTs, and low-cost or even free minting will serve as sponsorship mechanisms. These rewards may also not be economic but could involve building a community-based network of content creators based on social tokens.

In addition, there will be more consumer-centric models. For example, personal knowledge agents that enable consumers to curate their unique datasets, which they can revisit, mine, and explore over time.

In the future, different types of AI-native vertical search platforms will emerge, relying on the randomness of LLMs and treating exploration as a feature. They will particularly shine in areas like travel, music, books, fashion, second-hand goods, and interior design. Today, text-based chat has become the primary user interface for Google search. However, we expect it to evolve into more diverse forms, such as photos, audio, video, and even biometric data. Just as iNaturalist allows us to query bird songs, new platforms will leverage advanced technologies to interpret richer data forms, providing a more personalized and immersive search experience.

In summary, it is an exciting time to be in the search business. LLMs are changing the game and unlocking new ways to access and interact with information. We are also deeply excited about the new species that may soon be discovered.