Cambrian explosions · data infrastructure

Data and analytics mints $20B+ companies every time the infrastructure shifts. It's shifting again.

Every eight to ten years, an infrastructure shift breaks the incumbent stack and opens a short window. A few companies built for the new paradigm capture most of the value created inside it. The movement to the web and then to the cloud have generated ~$500b in enterprise value. The agentic revolution has opened the window for the largest evolution yet.

Three windows · where the value landed
Wave I Web 1.0 · ~2000s >$50B from two names

The original data explosion.

Companies suddenly held more data than a spreadsheet could show, and no way to read it. Two products defined the answer — and sold for prices that looked absurd at the time.

Examples

Tableau ACQ. SALESFORCE · 2019 $15.7B
Splunk ACQ. CISCO $28B
Wave II Cloud · ~2010–2018 >$200B value created

The rise of the modern data stack.

When data moved to the cloud, the vendors built for data centers and DBAs lost. The winners were cloud-native, priced by usage, and queryable by analysts instead of specialists. Every one of them started out pitching against Oracle.

Examples

Databricks FOUNDED 2013 $134B
Snowflake IPO 2020 $70B
ThoughtSpot SERIES F $4.2B
Sigma SERIES E $3B
Looker ACQ. GOOGLE $2.6B
Domo IPO $2.3B

Fivetran and dbt Labs, both born in this wave, merged in 2025–26 at a combined valuation near $10B — the plumbing of the modern data stack consolidating as the era closes.

Wave 2.5 The bridge · ~2019–2023

Modern BI, built for the last paradigm.

A bridge generation for the analyst-as-engineer era: collaborative notebooks, version-controlled SQL, data science folded into BI. They read the trend right but built for a paradigm already being replaced. This is the last BI made for humans running the analysis by hand.

Examples

Omni FOUNDED 2022 $1.5B
Hex SERIES C $420M
Mode ACQ. THOUGHTSPOT $200M
Wave III Agentic AI · Now >$500B up for grabs

The analyst was always the inference layer. Now the agent is.

The agent breaks that contract. The agent is the new inference layer — which means everything the human used to contribute on arrival has to be encoded in the foundation. Business logic. Metric definitions. Institutional knowledge. Trust signals. The judgment that used to happen at the end of the pipeline now has to happen at the beginning.

This paradigm shift requires a fundamentally new foundation. The companies that built the last one are not positioned to build this one.

Wave III · the agentic opportunity

The agentic shift defines this new chapter for data software.

The cloud shift moved data and widened access, but SQL stayed the interface and the dashboard stayed the output. Agents don't query that way. They need a layer the modern data stack was never built to provide.

The cloud shift

Professional-grade tools for SQL wizards.

Moved data and widened access, but the interface stayed SQL and the output stayed the dashboard.

The agentic shift

Context-driven, autonomous AI analysts.

Needs a governed semantic layer, live warehouse access, an API an agent can call, and a trail the data team can audit — none of it native to the modern data stack.

Why this wave, why now
01 · Inevitability

Not an "if AI takes off" bet.

Every large enterprise is already pointing agents at its data. The open question is who supplies the layer underneath: incumbents retrofitting their platforms, or a company built for it from the start. History tells us that new, native platforms outperform over and over.

02 · The timing signal

The Fivetran–dbt merger is the starting gun.

When the defining tools of an era merge to protect their IPO odds against platform encroachment, the era is consolidating. Incumbents are consolidating their existing market share, as opposed to pursuing the 5-10x TAM expansion opportunity.

03 · The size of the prize

The TAM expands beyond BI buyers.

Analytics and BI should clear $50B a year by the end of the decade. The larger prize sits outside it: any workflow where an agent reasons over enterprise data, which is most of them. The buyer widens from "BI teams" to "anyone putting agents on their data."

04 · The Supper position

Not a retrofit. Built native.

Supper owns the governed semantic layer between the agent and the warehouse. The question comes in plain language; Supper maps it to the right tables, applies the business logic the data team already defined, and returns an answer with its reasoning attached. Our platform is completely AI-native, built specifically to supercharge the needs of agent-driven businesses.

The Web produced Tableau and Splunk. The cloud produced Snowflake and Databricks. The agentic shift produces the next generation.