Massive parallelism
144 autonomous cores execute concurrent workloads without a central clock bottleneck.
Crowdfunding Launch: May 2026
Asynchronous computing for systems that must endure.
Volatco is a Forth-native, massively parallel platform designed for real-time control, robotics, edge AI, and long-lived autonomous systems.
- 144 independent F18A cores
- Up to 96B ops/s
- ~13 uW idle to ~972 mW peak (~650 mW typical upper draw)
- One-gate start/stop latency (~100 ps)
- ~7 pJ per instruction
- Event-driven compute - power used only when necessary
Why Volatco
Volatco avoids opaque cloud lock-in and fragile abstraction layers. It keeps computation close to hardware and close to human reasoning.
Ultra-low energy profile
Built for always-on systems, resilient edge devices, and power-constrained deployments.
100-year ethic
Repairable, teachable, and understandable computing designed against planned obsolescence.
Technical Snapshot
- Architecture: ISLP GA144-based
- Cores: 144 x 18-bit F18A processing nodes
- Peak compute: Up to 96 billion ops/s
- Energy is used only when necessary
- Power per F18A: 90 nW idle leakage to 6.8 mW active
- Power range per chip: 13 uW to 972 mW
- Start and stop: One-gate delay (100 ps)
- Per-instruction energy: ~7 pJ
- No OS to waste available time with unnecessary housekeeping
- Say goodbye to garbage collection!
Built for control and adaptation
Volatco is intended for machine-intelligent cybernetic systems where deterministic behavior, low latency, and graceful degradation are more valuable than cloud-scale dependency, adding innumerable points of failure.
Each F18A core can move between active and inactive states in a single gate delay, so systems can allocate compute only when needed instead of burning power continuously. That makes the platform a practical fit for long-running autonomous controllers, distributed sensing nodes, and embodied AI experiments that must remain responsive under tight energy limits.
Research Spotlight
Agentic AI: Embodied Learning Model
The ideal research path frames agency through embodied interaction, not passive observation. In this view, an agent does not treat inputs as direct world representation; it runs experiments, receives results, and updates behavior from those interactions. This is the essence of a truly self-tuning controller, not algorithmic.
The foundational module describes a cycle of anticipation, result comparison, and intrinsic states such as frustrated, self-satisfied, and bored, which drives experiment-switching and developmental learning over time.
See the source pathway and branches: github.com/cartheur/ideal.
Where It Fits
Robotics
Low-latency multi-sensor coordination and closed-loop control for embodied systems.
Drones
Fast distributed control loops for navigation, sensing, and mission autonomy at the edge.
Energy monitoring
Adaptive monitoring for microgrids, remote sites, and safety-critical power systems.
Edge AI
Parallel, local inference pipelines without forced cloud round-trips.
Off-grid and survival use
Can run in field setups powered only by a small solar panel and compact battery.
Autonomous research systems
Supports long-running experimental platforms that must keep operating without cloud dependency.
Pre-Launch FAQ
Can I buy right now?
Not yet. Purchases will open through the crowdfunding campaign in May 2026.
Is this a finished webshop?
No. This site currently focuses on education and campaign conversion before full ecommerce rollout.
Where can I ask technical questions?
Join the community on Discord: Volatco.