Answers for backers, programmers, FPGA engineers, and technical supporters.
Yes. The crowdfunding campaign is live now.
The crowdfunding campaign went live on 30 June 2026.
Join the community on Discord: Volatco.
The below expands on architecture, operation, workloads, and delivery expectations.
Volatco is a Forth-native, massively parallel computing platform built around GA144-class architecture for resilient robotics, edge AI, real-time control, and autonomous systems.
Volatco is designed for teams that want fast local decision-making, low-latency event response, and long-running operation without mandatory cloud runtime dependency.
Volatco is a hardware-first path for building understandable, durable, high-agency machine systems.
You can build systems that sense, decide, and act locally with tight timing and low energy overhead.
You can also prototype mesh-style compute behaviors, where many small F18A nodes coordinate state and decisions without a single central controller.
Typical uses include robotics control loops, drone navigation logic, distributed sensor coordination, edge inference pipelines, and long-running autonomous research platforms.
288 independent F18A processing nodes (2 x GA144), up to 192 billion operations per second in aggregate on the board, around 13 uW suspended chip power to about 972 mW typical all-nodes-running chip power, and workload-dependent per-instruction energy with measured examples around ~6 pJ.
Real-time sensor fusion, closed-loop control, distributed coordination, and local inference pipelines where deterministic response matters.
Terminology note: in this documentation, a node means a single F18A processor. A GA144 chip contains 144 nodes.
On VOL00, if you build a two-chip boot stream, both chips are initially full of ganglia by default unless the boot stream explicitly programs otherwise.
polyFORTH can use Snorkel and Ganglia to discover and reach unused nodes across both chips, enabling rich interconnection patterns for ambient phasic state meshes and contemplative decision-making programs.
Fast control data flow does not need to pass through polyFORTH, Snorkel, or Ganglia; high-rate loops can run directly between participating nodes.
In plain terms: Ganglia/Snorkel help orchestrate and reconfigure the system (for example, lower-frequency events and major algorithm/state transfers), while time-critical control can remain fully distributed.
Background synopsis:
Apeiron is the name used here for a learning-first agent architecture built on top of Volatco, not a separate operating system.
The name comes from the ancient Greek term for the boundless or unbounded, used here to describe an agent architecture that expands capability without fixing behavior into a closed policy.
Its center is an open-ended skill core that keeps composing new behaviors from experience. Fast control stays close to the hardware on direct node-to-node paths, while Ganglia and Snorkel support system-wide observability, supervision, discovery, and task redistribution.
The result is a machine architecture where learning remains primary, control remains local and enforceable, and identity can be treated as a versioned model of goals and capabilities instead of a fixed permanent policy.
Another way to say it is that Apeiron pushes hardware-software hybridization far enough that the user can concentrate on the idea at hand, instead of first asking whether the programming language will allow it.
Yes, Volatco is designed as a persistent computing platform that maintains state and continuity even when power is lost or systems are restarted.
No. Volatco is designed for local operation and does not require cloud lock-in for core workloads.
Cloud platforms can still be useful for optional collaboration and deployment workflows, but they should not be a mandatory runtime dependency for control-critical systems.
Volatco prioritizes systems that keep running when networks are unstable, vendors change terms, or remote APIs become unavailable.
Each F18A node can transition between active and suspended behavior on extremely small time scales, enabling rapid local response without constant full-power operation.
Event handling can propagate directly across connected F18A nodes for high-rate control, while Ganglia/Snorkel are typically used for lower-frequency coordination and major state changes.
Estimated shipping: November 2026.