Skip to content

From search to answers: How LLMs are rewiring the Internet’s business model#

The content wars, episode 2

Read also:

The way we use the internet is undergoing a seismic shift. We are moving from a “search era” — where users actively seek out information — into an “answer era” defined by passive consumption through large language models (LLMs).

janus

This change, already visible in user behavior, will have profound consequences on the internet’s economic and structural foundations.

1. From Search to Answer: The Behavioral Revolution#

Internet users are increasingly turning to LLMs like ChatGPT, Claude, and Gemini not just to search for information, but to receive answers.

Rather than navigating a labyrinth of links, users are now offered synthesized, contextually relevant responses in natural language. The traditional search experience — type, click, skim, repeat — is being replaced by a streamlined Q&A paradigm. This evolution is not merely technological; it marks a behavioral transformation.

2. The Advertising Disruption#

This shift threatens the foundation of the internet’s dominant economic model: search-based advertising. Search engines like Google have long thrived by monetizing user intent — serving ads alongside search results. But in the answer era, users no longer see search result pages. Instead, they engage in conversational interfaces where traditional ads have no place.

This compels a fundamental shift: advertising dollars will inevitably follow attention, migrating from search engines to LLM platforms, even if it is not clear yet today.

3. The Great Content Absorption#

LLMs are not only changing how we find information — they’re also absorbing the web itself. Through continuous training, these models are ingesting and synthesizing vast swathes of existing content.

sea-llm-eating.png

As LLMs learn from everything that has already been published, traditional websites face a chilling reality: their old content, now part of a model’s knowledge base, may no longer generate traffic. What matters now is not what has been published, but what will be published — and who owns that future content.

4. The Rise of Sources of New Content (SNC)#

In this new paradigm, Sources of New Content (SNCs) become the internet’s most valuable assets. Since users won’t consult the original sources anymore, SNCs won’t survive on traffic alone. Instead, they must be remunerated directly by LLM providers — much like how search engines pay licensing fees for syndicated data. The only way for these content generators to thrive is if their outputs are part of LLMs’ training streams and monetization loops.

5. Strategic Advantage: Owning the Firehose#

LLMs integrated with continuous, high-volume SNCs (e.g., social platforms like X or Facebook) will have a decisive advantage. These platforms host a constant stream of fresh, diverse, real-time content — a goldmine for model fine-tuning.

The catch: this content is noisy, unstructured, and often irrelevant. The real differentiator will be the ability to clean, filter, and prioritize this stream effectively for high-quality model training.

Understand now why Musk bought Twitter to put Grok inside? Understand why Facebook is hiring highly-skilled profiles for its LLM?

6. Journalism in the Crosshairs#

The implications for media are existential. Already, many newsrooms rely on AI tools to generate or augment content. But in the answer era, even that intermediary step may disappear.

Much of journalism — especially the regurgitation of press agency notes — becomes redundant. As LLMs draw directly from press wires or primary sources, they can bypass the editorial layer entirely. Journalists who merely repackage agency content will be replaced. Only those producing original investigations or perspectives will retain relevance — and only if their output is monetized as SNC.

7. Goodbye UI, Hello API#

As humans retreat from the frontlines of content consumption, user interfaces become less important. SNCs no longer need to attract or retain human eyeballs; instead, they must serve clean, structured content to machines.

This suggests a future where SNCs operate through APIs designed for LLM ingestion, not human interaction. For social platforms — which are SNCs that also gather human behavior — maintaining a UI makes sense. But for others, simplicity will rule.

Note: Fortunately, this site did not invest a lot on UI!

8. Emails: The Overlooked Content Frontier#

One final — and critical — SNC stands out: emails. Private, rich in context, and constantly updated, email threads represent a trove of new, valuable information.

If LLMs are to become truly personal assistants, they must integrate deeply with email engines. Control over the inbox — and over how its data is processed — will become another battleground in the fight for SNC dominance.

Conclusion: A New Internet Business Model#

The transition from search to answers is not just a UX improvement — it is a reordering of the internet’s entire value chain.

In the answer era:

  • Search-based ad models erode.
  • SNCs gain economic and strategic value.
  • LLM providers become the new gatekeepers of attention.
  • APIs replace websites.
  • Monetization moves upstream — from traffic to training data.

In this brave new web, those who control the flow of new content — and those who can structure, clean, and monetize it — will define the next digital economy.

As an enterprise, if your SNC is integrated in the Cloud with AI, you're just multiplying the number of chains of your digital slave status.

Original version here.

(July 13 2025)

Navigation: