Citi Analysts: H2 Tech Investment Focus Shifts to the Data Layer
0xBroomberg
Citi AI-investment head Heath Terry says the top tech opportunity for 2H 2026 is the data layer — the software-stack tier that cleans and organizes data for AI models — with Snowflake, MongoDB and peers whose revenue scales directly with data consumption as core picks.
What exactly is the "data layer"?
The data layer sits between storage and applications in the software stack. Its job: clean, format, and organize raw data so AI models can consume it.
In plain terms = data is AI's fuel; the data layer is the refinery. The more AI consumes, the bigger this business gets.
Terry argues that AI is driving a massive surge in data consumption, making the data layer the most reliable investment theme for the second half.
Which companies benefit most directly?
Terry names four: Snowflake, MongoDB, Datadog, and Elastic. All share one trait — revenue tied directly to how much data customers actually use.
This means → the more a customer consumes, the more they pay. It is a consumption-driven model — stronger AI demand feeds straight into faster growth.
Valuations are already steep: Snowflake reported 33% annual revenue growth, trading at roughly 112× forward P/E; Datadog reported 25% growth, at about 98× forward P/E.
What is the "model routing" opportunity?
Model routing — technology that assigns each enterprise task to the best-suited AI model — is another area Terry highlights.
This reflects a shift: AI deployment has entered a multi-model phase, where companies run several models at once. Who dispatches, and how to cut cost, is now a real problem.
The leading players — Not Diamond, Martian, OpenRouter — are all private, so public-market investors cannot access this theme yet.
Is the SaaS "apocalypse" over?
Terry notes that software stocks went through a broad valuation reset the market informally dubbed the "SaaS apocalypse."
As investor confidence in data-consumption demand recovers, the sector is shifting from a blanket sell-off to stock-by-stock divergence.
This means → the second-half software opportunity is no longer "buy the basket." Picking the right names is what drives excess returns — in Terry's words: "We're getting back to a better place where you can differentiate winners from losers."
Content is for reference only, not financial advice.