PhoenixAI's $80M Bet on a New Database for Autonomous AI Agents

MENLO PARK, CA – June 11, 2026 – In a move that signals a critical infrastructure shift in the artificial intelligence landscape, PhoenixAI today announced it has closed an $80 million Series B funding round. The investment, led by global venture firm Sky9 Capital, is a significant validation of the company's mission: to solve the growing data bottleneck that threatens to stall the deployment of advanced AI in the enterprise.

PhoenixAI, formerly known as CelerData, is not building another general-purpose database with AI features tacked on. Instead, it has engineered what it calls an "Agentic AI Database" from the ground up, a platform purpose-built for the unique and demanding behavior of autonomous AI agents. This funding will fuel the development of its AI-native database, expand its market presence, and critically, strengthen governance capabilities for enterprises navigating the complexities of regulated industries.

The Agentic Shift: Why Old Databases Are Failing AI

The buzz around AI is moving beyond chatbots and copilots into a new, more active phase: agentic AI. These are not passive assistants; they are autonomous systems designed to execute complex, multi-step tasks—managing supply chains, serving customers, and optimizing internal workflows in real time. But as enterprises move from prototyping to production, they are hitting a wall. This wall, it turns out, is the very foundation of their data stack.

"Today's agentic landscape has moved quickly from planning and prototyping to full-on production for mission-critical work," said Rick Underwood, President of PhoenixAI, who previously helped steer Snowflake through its IPO. "Agents now fire off thousands of unplanned, real-time queries, often swarming systems with questions that weren't anticipated when the data stack was designed... which strains existing data stacks."

This strain arises from a fundamental mismatch. Traditional databases were designed for human users, who query data in relatively predictable, sequential patterns. AI agents operate differently. A single high-level question from a business user can trigger a "query explosion," where an AI agent issues dozens of concurrent, iterative queries in milliseconds to understand context, join disparate datasets, and formulate an answer. Furthermore, these agents demand data that is fresh to the second, as decisions based on hours-old, batch-processed information can be dangerously inaccurate in a live operational environment.

Industry analysis confirms this burgeoning need. The market for agentic AI is projected to skyrocket from under $8 billion in 2024 to nearly $200 billion by 2033. This explosive growth is predicated on the ability of enterprises to provide AI with the fuel it needs: real-time, governed, and accessible data. Without a new class of database, the promise of autonomous enterprise AI remains just that—a promise.

PhoenixAI's Solution: A Purpose-Built Engine for AI

PhoenixAI aims to be the solution to this architectural impasse. The company's platform is designed to unify real-time streaming data and historical data-at-rest within a single, high-performance engine. This allows it to handle the chaotic, high-concurrency query patterns of AI agents without breaking a sweat, delivering sub-second responses even when sifting through massive datasets.

Early adopters are already seeing the impact. Market-leading companies like Coinbase, Conductor, and Demandbase have deployed PhoenixAI in production, moving it from a theoretical solution to a proven enterprise tool.

At Coinbase, the platform has transformed their data capabilities. "PhoenixAI changed the equation: streaming updates from Kafka become queryable within seconds, analysts get sub–second responses on live normalized data, and our AI agents operate on the same real–time dataset," noted Xinyu Liu, Senior Staff Software Engineer at the cryptocurrency exchange. "This level of performance at scale fundamentally changes what data teams can do."

This sentiment is echoed by others who are leveraging the technology to power next-generation workloads. "Production queries scanning hundreds of millions of rows return in under a second, and this is the architecture we need for the next generation of agentic workloads," stated Wei Zheng, Chief Product Officer at Conductor. The ability to create a "fast, isolated warehouse for agent workloads" directly on existing data formats like Apache Iceberg, as highlighted by Demandbase's Senior Engineering Manager, Ryan Nowacoski, points to the platform's flexibility and its ability to integrate without requiring a complete data overhaul.

The Investment Thesis: Betting on AI's Foundational Layer

The $80 million investment, with participation from Atypical Ventures and Olive Technology Ventures, is more than just capital; it's a strong bet on which layer of the AI stack will generate immense value. While much of the public focus has been on large language models, savvy investors are looking at the less glamorous but equally critical infrastructure required to make those models useful in an enterprise context.

"The move to agentic AI is one of the largest infrastructure shifts we've seen, and the database is at the center of it," explained Ron Cao, Founder and Managing Partner of Sky9 Capital. "Enterprises can't put agents into production until something can serve them live data at the speed and scale agents now demand. That's the problem PhoenixAI was built to solve."

This investment places PhoenixAI in a competitive but rapidly expanding market. Major cloud providers are retrofitting their existing database services with vector search and other AI-related features. At the same time, a new cohort of specialized companies, like real-time analytics database Rockset (recently acquired by OpenAI) and open-source powerhouse ClickHouse, are also vying to become the data backbone for AI. PhoenixAI's strategy is to differentiate through its singular focus on the unique demands of autonomous agents, a niche it hopes to define and dominate.

Beyond the Hype: Governance and Real-World Deployment

For any enterprise CTO, the prospect of autonomous agents firing off thousands of queries against production systems is both exciting and terrifying. This is where PhoenixAI's emphasis on governance becomes a crucial differentiator. The move from supervised models to autonomous agents necessitates a new paradigm for security, access control, and auditability.

Traditional governance models, built for predictable human access, are insufficient for managing AI agents that can dynamically explore data schemas and formulate novel queries. A robust framework is required to grant agents the access they need while implementing guardrails to prevent unintended consequences, ensure data integrity, and provide a clear, auditable trail of every action taken.

By earmarking a portion of its new funding to "deepen governance for regulated industries," PhoenixAI is directly addressing the primary barrier to adoption for many large enterprises. This focus on building a safe, controlled environment for AI agents to operate is the critical bridge between experimental sandbox projects and mission-critical, revenue-generating deployments. The funding and the firm's clear focus signal that the foundational tools for the agentic era are now being built in earnest, aiming to turn the immense potential of autonomous AI into a practical business reality.