ASH A(π)
The Intelligent Computer.
Electron Revolution's first native ASH intelligence computer — workstation-class hardware integrated with the ASH mission environment. The only native ASH system in the launch catalog.
Not a bigger GPU. A different kind of system.
The other twelve ER systems are excellent, AI-capable computers. ASH A(π) is the only one built around native ASH from the platform up — the hardware and the intelligence layer were specified together, not paired after the fact.
Built on a workstation platform, not a consumer one.
NVIDIA RTX PRO 6000 Blackwell Workstation Edition 96GB ECC
128GB DDR5 ECC RDIMM
8TB Gen5 NVMe for models and data, plus a dedicated 2TB ASH/OS drive — the platform and your working data never compete for the same disk.
AMD Ryzen Threadripper PRO 9975WX
| Motherboard | ASUS Pro WS WRX90E-SAGE SE |
|---|---|
| Cooling | TR5 workstation liquid cooling |
| Power Supply | Seasonic PRIME TX-1600 Titanium |
| Chassis | ER Signature workstation chassis |
ASH is provisioned as part of the platform, not installed after.
Mission Computing
ASH A(π) is engineered around persistent missions rather than disconnected apps and files — see the Mission Computing section below.
Local & private model use
The 96GB workstation GPU is designed to support running models locally, on hardware you own.
Agents & tools
ASH is being developed to coordinate agents and tools against a mission, with the hardware to run that coordination locally.
Memory & provenance
Mission Computing is designed around a record of what happened and why — provenance concepts intended to keep execution accountable, not a black box.
Governance & permissions
Missions are intended to run within operator-set permissions rather than open-ended autonomy — governance is part of the design, not an afterthought.
Hybrid local + cloud
Local intelligence is the default; ASH is designed to reach optional cloud intelligence for workloads that exceed any local machine, not to require it.
ASH is under active development. Capabilities described here reflect the intended product direction for ASH A(π) — see docs/internal for current implementation status if you need the engineering detail behind this page.
A computer built around what you're trying to get done.
Apps, files, and disconnected tools. You are the one holding the whole task together across windows.
Mission-oriented computing — persistent missions, coordinated intelligence, and governed execution, engineered to hold the task together for you.
How the pieces connect.
- ASH is the central coordination point.
- ASH connects to: User, Missions, Agents, Models, Tools, Local Data and Memory, Governance, and ER Hardware.
- ASH also connects to optional Cloud and external APIs, shown with a dashed line to indicate this connection is optional, not required.
ASH coordinates missions, agents, models, and tools against ER hardware, with local data/memory and governance always in the loop. Cloud intelligence is optional, not required.
96GB of professional GPU memory, on hardware you own.
Model execution on owned hardware means your data doesn't need to leave the machine for the workloads that fit locally — reduced dependence on external APIs, lower latency for local workflows, and control over what stays private.
This is not a claim of complete offline operation for every workflow, and it does not mean every future ASH capability will run with zero cloud dependency. ASH A(π) is designed for hybrid use: local by default, with optional cloud intelligence for the workloads that exceed any local machine — including this one.
96GB of ECC GPU memory is a large, honestly-stated number — it is not a guarantee of which specific models it can run. Model support depends on the model, not just the hardware.
The hardware and the platform were specified together.
Hardware selected for the workload
CPU, GPU, memory, and storage sized specifically for running ASH and its models locally — not a gaming spec repurposed for AI.
ASH provisioned as part of the platform
The intelligence layer ships configured on the hardware, not left as a separate install step.
Storage separation
A dedicated 2TB ASH/OS drive keeps the platform independent from the 8TB working/model storage.
Workstation power & thermal envelope
TR5 liquid cooling and a Titanium-rated PSU sized for sustained workstation load, not burst gaming loads.
Validation
Every ER system is assembled, thermally validated, and burned in before delivery.
Serviceability
Standard workstation-platform components — built to be opened and serviced, not sealed shut.
Where ASH A(π) sits, honestly.
Full specification.
Compute & Memory
| CPU | AMD Ryzen Threadripper PRO 9975WX |
|---|---|
| GPU | NVIDIA RTX PRO 6000 Blackwell Workstation Edition 96GB ECC |
| GPU VRAM | 96GB ECC |
| System Memory (ECC) | 128GB DDR5 ECC RDIMM |
Storage
| Storage | 8TB Gen5 NVMe + 2TB dedicated ASH/OS SSD |
|---|
Platform & Chassis
| Motherboard | ASUS Pro WS WRX90E-SAGE SE |
|---|---|
| Cooling | TR5 workstation liquid cooling |
| Power Supply | Seasonic PRIME TX-1600 Titanium |
| Chassis | ER Signature workstation chassis |
| Networking | TBD |
| Operating System | TBD |
ASH-Native Status
This is the only native ASH intelligence system in the Electron Revolution launch catalog. All other systems (SKU01-SKU12) are AI-capable but not ASH-native.
Lead Time & Warranty
TBD — confirm before publishing storefront copy
TBD — confirm before publishing storefront copy
ASH A(π) is commissioned, not just shipped.
Every ASH A(π) begins with a consultation — configuration, ASH provisioning, and commissioning are discussed before anything is built.
Local development preview — no live payment processing is configured.