Colin Zima spent a decade inside the enterprise analytics engine room. As an early engineer and later director at Looker, he watched companies pour millions into business intelligence, only to find that every dashboard, every report, every ad-hoc query could define the same metric differently. Revenue meant different things to sales, finance, and product. The same data source produced different answers depending on who asked and which tool they used.
When Google acquired Looker in 2019, Zima and his co-founders Christopher Merrick and Jamie Davidson saw the next problem coming before most of the market did. AI was about to flood the enterprise with natural-language queries. And if humans couldn't agree on what "revenue" meant, AI would hallucinate through the same ambiguity at machine speed.
They founded the company in 2022 to build the missing layer.
The round doubles the company's valuation from $650M in March 2025 and includes a $30M employee tender offer. Revenue grew 4x year-over-year, and the company reached profitability in early 2026, uncommon for a fast-growing AI startup.
The core insight: the semantic layer that made its BI consistent is now the critical infrastructure for trustworthy enterprise AI.
The timing is telling. Enterprise AI adoption is accelerating faster than the governance around it. Gartner projects that by 2027, 75% of hiring processes will include AI proficiency testing. IDC forecasts global AI spending to exceed $300B by 2026. But the same Gartner data shows that most AI analytics projects fail because the underlying data lacks consistent business context. That is the semantic layer problem the platform was built to solve.
From Founding DNA to the Semantic Layer Thesis
The company didn't start as a funding story. It started as a technical bet: the semantic model, a governed layer that sits between raw data and every query, defining metrics, dimensions, joins, and permissions in one place, would become the default architecture for both BI and AI.
The bet was not obvious in 2022. The market was still fragmented. Snowflake and Databricks were competing on warehouse performance. The semantic model approach was powerful but tied to a specific analytics workflow. Most companies treated the semantic layer as a nice-to-have documentation exercise rather than the execution layer for all data access.
Zima and team went the other direction. They built the platform around a semantic model from day one, not as a metadata catalog but as an active query governor. Every dashboard, spreadsheet, SQL query, and AI chat goes through the same semantic definitions. The system doesn't just know what data exists. It knows what each metric means, who can see it, and how it should be calculated.
The architecture proved prescient. When generative AI hit the enterprise in 2024, the first wave of chatbots and copilots crashed against the same problem the platform had already solved: AI could generate fluent SQL, but it had no idea which table held "active customers" or whether "ARR" included expansion revenue.
As we wrote in June, Golden Analytics raised $14M for a similar thesis, but the startup's approach is architecturally distinct. Golden builds AI-native analytics as a standalone product. The startup embeds the semantic layer as infrastructure underneath existing BI, AI agents, and embedded analytics.
The $120M Inflection
The Series C round announced on April 23, 2026 marks a clear inflection point. ICONIQ led the round, with participation from existing backers Theory Ventures, First Round Capital, Redpoint Ventures, and GV. The $30M employee tender offer signals confidence in the company's trajectory.
The numbers justify the conviction:
- 4x year-over-year revenue growth
- Revenue tripled year-to-date in 2026
- Company reached profitability in early 2026
- Valuation more than doubled from $650M to $1.5B in 13 months
- Customers include BambooHR, Checkr, Cribl, dbt Labs, Guitar Center, Mercury, Pendo, and Synthesia
Profitability at this stage is rare. Most AI infrastructure startups burn through capital scaling compute and headcount. The startup's path suggests product-market fit that doesn't depend on infinite venture subsidies, a signal investors are increasingly demanding in the 2026 funding environment.
"AI isn't replacing analytics, it's expanding it," Zima told Fortune in an exclusive interview. "Dashboards and spreadsheets aren't going away, but now anyone can get instant answers without technical expertise."
Where the Semantic Layer Fits
The competitive field is shifting. Sigma Computing raised $80M in Series E at a $3B valuation in May 2026, pivoting toward agentic analytics. ThoughtSpot has raised over $674M but relies on search-driven AI without deep semantic modeling. Databricks and Snowflake are building semantic capabilities into their platforms, but as features within a broader data stack, not as the governing layer.
The differentiation is architectural. The semantic model is not a bolt-on. It is the platform. Every query goes through it. Every AI agent inherits its definitions and permissions. This means the platform can serve as the context graph not just for its own UI but for external AI tools through its MCP server, APIs, and connectors to Claude, ChatGPT, and Cursor.
ICONIQ partner Matt Jacobson described the thesis: "The barrier in data has shifted from access to understanding. Everyone can ask questions, but without a shared layer of business context, the answers can break down. We believe the real problem was never accessibility — it was trust."
What Comes Next
The roadmap points toward deeper agentic capabilities. The company's MCP server already lets external AI agents query the semantic layer directly. The acquisition of Explo in October 2025 added embedded analytics for customer-facing products. The Databricks ISV Emerging Partner of the Year award in June 2026 signals platform-level integration with the broader data ecosystem.
The fundamental question the company is testing is whether the semantic layer becomes a new category of enterprise infrastructure, as essential as the data warehouse itself, or remains a feature inside larger platforms. The $1.5B valuation suggests investors are betting on the former.
For the enterprise buyers watching this space, the practical takeaway is simpler. Every company struggling with AI hallucinations, inconsistent metrics, or data access bottlenecks is facing the semantic layer problem. It is the most well-capitalized independent bet that solving it at the architecture level beats patching it at the prompt level.
TIMELINE: Omni Analytics
─────────────────────────────────────────────────────────────
2022 ──── 2023 ──── 2024 ──── 2025 ──── 2026
🏢 🧪 🚀 📈 ◉ NOW
Founded Product First Acquires Series C
by ex- launch major Explo, $120M
Looker + seed custo- reaches at $1.5B
team mers profit valuation
Omni's journey from founding to Series C in four years. Source: company disclosures, Fortune, BusinessWire.