We're excited to introduce PhoenixAI Anywhere, our self-managed platform for enterprises that need to run analytics entirely on their own infrastructure. PhoenixAI Anywhere runs the same sub-second PhoenixAI engine as PhoenixAI Cloud, deployed and operated entirely inside your own environment — your cloud VPC, your data center, or a network with no outbound connectivity at all.
Why PhoenixAI Anywhere — and why now
Not every workload can move to the cloud, and that gap is widening. In Barclays' Q4 2024 CIO Survey, 86% of CIOs said they planned to move at least some workloads from public cloud back to private or on-premises infrastructure — the highest figure the survey had recorded. AI accelerated it: once enterprises moved from pilots to production, the data feeding those systems became the thing that needed protecting, and regulators followed. Gartner predicts that by 2028, 65% of governments worldwide will introduce some technological sovereignty requirements to improve independence and protect against extraterritorial regulatory interference.
For some organizations, that requirement reaches past the data to the control plane itself. Most are well served by PhoenixAI Cloud (BYOC), which keeps data in their own account while we carry the operational load. But a subset operates under mandates where nothing crosses the boundary at all — the same principle behind private and sovereign AI deployments.
The reasons are specific, and any one of them is enough on its own:
- Air-gapped or zero-egress networks — no outbound path to any vendor
- Data residency and jurisdiction — data that has to stay in a specific country
- Compliance and audit — every access path accounted for
- Control-plane ownership — not just where data sits, but what crosses the boundary
- Existing infrastructure investment — teams already operating their own platform at scale
- Vendor lock-in avoidance — increasingly a board-level position
Until now, meeting any of them meant choosing between a managed database that can't go where the data lives, or open source that goes anywhere but leaves your team to build the operational tooling themselves. PhoenixAI Anywhere exists to close that gap.
That same AI shift reshaped us, too. The rename from CelerData to PhoenixAI reflects where analytics has landed: the database is now the serving layer for AI and agentic workloads running against live operational data — which makes a self-managed form factor core to the product, not an edge case.
What PhoenixAI Anywhere is
PhoenixAI Anywhere is a platform, not just a database: the PhoenixAI engine together with the Anywhere Console that manages it. Anywhere deploys your PhoenixAI database clusters, keeps them healthy, and gives your team visibility and control over what's running — the operational surface a self-managed deployment has historically had to assemble for itself.
The difference is what surrounds the database. Running an analytics engine yourself is a problem at both ends: getting it deployed and configured correctly in the first place, and then the months afterward — knowing whether the cluster is healthy, finding out why a query slowed down, working out what to send us when something breaks.
That burden is what Anywhere is built to absorb — the aim being that you set it up and get on with your work, rather than acquiring a second job running a database. The August release starts on that with health checks, diagnostics for both the database and the infrastructure around it, and a guided way to hand the full picture to us when you do need help. The operational surface grows from there.
What ships in the August Public Preview
Anywhere's August release is Public Preview, with GA following a standard hardening cycle. The theme this month is operational confidence: enhanced troubleshooting and support tools, and better visibility into usage and licensing. Here's what's included today:
- Automated deployment. Anywhere stands up a fully featured cluster (console, warehouses, and storage) on your own Kubernetes, in any cloud or your data center, with an upgrade path in from existing self-managed deployments.
- Automated lifecycle. Declare the shape you want and Anywhere reconciles the cluster to match, then keeps it there: scaling, failover, upgrades, volume changes, password and SSL rotation, and backup and restore.
- Query Insight. Query insights and profiling, plus catalog browsing: visibility into what's actually running against your data, and the detail to debug and optimize a specific query when one isn't performing the way you expect.
- Usage visibility. CCU metering with a full audit trail and export, so you can see and account for what you're actually consuming.
- Self-serve troubleshooting. Automated health checks surface problems directly instead of leaving you to infer them from logs. Key metrics are collected and displayed in the Anywhere Console. Diagnostic command presets cover both halves of a self-managed deployment for quick diagnostics (SQL commands for database insight, shell commands for the infrastructure around it).
- Scenario-aware support bundle. Pick what you're seeing (compute node crash, coordinator stall, or failing SQL) and the Anywhere Console preselects the artifacts support usually needs for that situation, with every item adjustable before you generate. What comes out is a single archive, ready to send: no collecting logs by hand, no back-and-forth about what we'll ask for next.
Built for zero outbound dependency
Anywhere was engineered around a single question: what does it take for an enterprise database to run somewhere that can't take an outbound connection to a vendor at all?
- No outbound path required. Anywhere's control plane runs entirely inside your cluster and needs no connection back to PhoenixAI, the architectural reason it works where a split control plane can't.
- Licensing that validates offline. Entitlement checks and expiry are enforced with no dependency on internet connectivity.
- Metering that stays local. Usage is measured and exported inside your environment, with nothing reported back to us automatically.
- Diagnostics without a connection. Support bundles are files you generate and transfer deliberately, so getting help works the same in a fully disconnected network as anywhere else.
Most commercial products assume a call home somewhere in that chain, whether for licensing, usage reporting, or support. Anywhere assumes the opposite.
In production
"Coinbase's blockchain dataset spans hundreds of billions of rows across more than 300 normalized tables. Our fraud and compliance workloads require complex joins, and blockchain network analytics demand low-latency query performance that our previous query engines could not provide. 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. This level of performance at scale fundamentally changes what data teams can do."
— Xinyu Liu, Senior Staff Software Engineer, Coinbase
What's next on the roadmap
Anywhere follows PhoenixAI's standard release framework — Public Preview, then GA after a hardening cycle. On the roadmap:
- Console write capabilities — a deployment UI and a full SQL editor
- Self-service alerting and AI-assisted support
- Broader Kubernetes distribution support
- Connectivity conveniences for limited-connectivity environments
- Unified identity management
Get started today
PhoenixAI Anywhere's August Public Preview is available now for teams ready to run PhoenixAI fully inside their own environment. Contact us to start a 30-day free trial or talk through your requirements.