Key Highlights

  • Similarweb's core business has historically centered on web traffic and digital intelligence data for marketing and competitive benchmarking purposes.
  • A new thesis positions large LLM data contracts as a transformative revenue stream, monetizing Similarweb's existing data assets in a fundamentally new way.
  • This model differs from many speculative AI beneficiary narratives because it relies on monetizing already-collected data rather than requiring new product development.
  • The durability and scale of these LLM data contracts, rather than the concept itself, will determine whether this becomes a lasting pillar of Similarweb's growth story.

Every 'AI beneficiary' stock story deserves the same basic scrutiny: is the AI connection a genuine new revenue mechanism, or is it a narrative loosely draped over a company that happens to use the word 'AI' somewhere in its marketing? Similarweb's emerging thesis around large LLM data contracts is worth examining specifically because it's a case where the underlying business logic actually holds up reasonably well under that scrutiny.

Similarweb's traditional business is built around collecting, structuring, and analyzing web traffic and broader digital intelligence data — information about website visits, audience overlap between competing sites, digital market share, and related metrics. Historically, the customers for this data have been marketers, competitive intelligence teams, and business strategists trying to understand digital market dynamics for their own companies or industries.

What's changed is that large language models require enormous quantities of high-quality, structured data for training and for what's often called 'grounding' — supplementing a model's base training with more current, verified real-world information to improve accuracy and reduce the kind of outdated or hallucinated responses that plague models trained on static, aging datasets. Web traffic and digital behavior data, particularly structured and cleaned data of the kind Similarweb has spent years building infrastructure to collect, represents exactly the sort of large-scale, real-world dataset that LLM developers need and, critically, that would be extremely costly and time-consuming for those developers to independently replicate from scratch.

This is the structural reason the LLM data contract thesis is more credible than many adjacent 'AI beneficiary' narratives: Similarweb isn't pivoting into a new business or promising to build new AI capabilities from scratch. It's identifying a new customer category — LLM developers — for data assets the company has already spent years and substantial capital building. That's a fundamentally lower-execution-risk story than a company promising to develop entirely new AI products or capabilities to capture the same opportunity.

The genuine open question, and the one that will determine whether today's market enthusiasm is well-placed, is durability and scale. LLM developers' appetite for licensed third-party training and grounding data is itself an evolving market, with pricing, contract terms, and long-term demand still being established across the industry. A single large contract announcement is meaningfully different from a durable, recurring revenue stream, and the degree to which Similarweb can convert initial LLM data relationships into multi-year, expanding commercial agreements — rather than one-off licensing deals — will ultimately determine whether this becomes a genuine new growth pillar or a one-time revenue bump that gets fully priced in and then fades from the narrative.

For now, the combination of analyst target increases from Citigroup and Oppenheimer and reported insider buying suggests market participants closest to the story see genuine substance behind the LLM data contract thesis rather than pure speculation. But as with any new and rapidly evolving revenue category, the real test will come through subsequent quarters of disclosed contract value and renewal rates, not through a single day's enthusiastic price reaction.