PhoenixAI is a real-time analytical database purpose-built for AI agents and customer-facing analytics that cannot tolerate the latency of batch ETL pipelines. It unifies streaming ingestion from Kafka, Flink, and Spark with direct query access to Apache Iceberg and Delta Lake lakehouse tables, so agents run sub-second SQL over both live and historical data without copying or transforming it first.

The platform ships as PhoenixAI Cloud, a fully managed bring-your-own-cloud deployment, or as PhoenixAI Anywhere for self-hosted operation. It includes SOC 2 Type II governance, fine-grained access controls, an MCP connector for agentic integration, and the Agent Fawks SQL assistant.

The emergence of agentic AI is driving demand for a new category of real-time analytical database that can serve live, fresh data to autonomous agents at agent speed and scale. Conventional batch-oriented data warehouses were not built for this workload, expanding the addressable market beyond traditional business intelligence toward operational and agentic data serving.

As enterprises move AI agents from pilots into production customer-facing applications, the requirement for sub-second, high-concurrency analytics over streaming and lakehouse data is growing. PhoenixAI's eighty million dollar Series B in June 2026, led by Sky9 Capital, reflects investor conviction that this agentic AI database category will expand alongside broader production AI agent deployment.

PhoenixAI delivers sub-second SQL across both streaming data and Apache Iceberg or Delta Lake lakehouse tables within a single unified analytical database, eliminating the batch ETL and copy lag that traditional warehouses impose on AI agents. The engine combines SIMD vectorized execution, a massively parallel processing architecture, and a cost-based optimizer to sustain tens of thousands of concurrent queries at low latency.

Governance is enterprise-grade with SOC 2 Type II compliance and fine-grained access controls. An MCP connector plus the Agent Fawks SQL assistant make the platform natively usable by autonomous AI agents rather than only by human analysts.

PhoenixAI enters the agentic-AI database category as a recent rebrand of CelerData, a company whose standalone brand recognition still lags established incumbents such as Snowflake and Databricks, which hold far larger installed bases, partner ecosystems, and marketplace integrations. Published pricing is not disclosed on its pricing page, which directs buyers to a workload-dependent sales quote, raising procurement friction for self-serve teams that expect transparent consumption pricing.

The platform's dependence on the StarRocks open-source core means its differentiated engineering reputation is shared with a community project rather than fully proprietary, and the BYOC and self-managed deployment models shift infrastructure and operational burden onto the customer's own cloud and data teams. Enterprise buyers requiring a turnkey fully-managed single-tenant SaaS without shared responsibility must weigh that operational tradeoff against the latency and governance benefits.

PhoenixAI uses a quote-based, workload-dependent pricing model rather than published per-unit rates, with PhoenixAI Cloud (bring-your-own-cloud) and PhoenixAI Anywhere (self-managed) as its two deployment offerings and a 30-day free trial with no credit card required on both. Because the platform deploys inside the customer's own AWS, Azure, or Google Cloud account under a shared-responsibility model, pricing is tailored to query volume, data scale, deployment region, and support tier rather than fixed tiers.

The strategy targets enterprise buyers who value data residency and governance over self-serve consumption pricing, replacing a published price card with a sales-led evaluation against the customer's actual data profile. This friction-accepting approach suits regulated and large-scale agentic workloads where deployment control and security outweigh transparent unit economics.