Home › Actions
Actions are instructions your AI assistant follows to do something with this catalog, not just read it. Each one reads and extends one design file (design.schema.json): Size → Network → Rack & power → Estimate → Proposal. Stop after any step and what you have is still useful. Everything is priced only from catalog.json; results are budgetary, never a quote. Prices in USD as of 2026-11-03; Supermicro lines at eStore B2B net (12% off list, list in brackets); the supplier confirms the final price on order. Continental US; Canada quoted case by case. Ready to buy: book a meeting and UpStation introduces the supplier (how buying works). Every page: llms.txt (map) · homepage (all tables) · catalog.json (data).
| Action | What it does | Status | Instructions |
|---|---|---|---|
| Design and estimate | requirements → sizing (GPU nodes, plus optional storage and cluster-services nodes) → networks (GPU fabric, front-end, OOB) → racks and power (U, kW, circuits, colo) → a budgetary estimate on UpStation letterhead (HTML/PDF/Markdown) with a design ID | live | /actions/estimate/SKILL.md |
| Network topology | sized nodes → GPU fabric (single switch or rail-optimized leaf-spine), front-end MLAG pair, OOB → topology diagram, fabric table, priced network bill of materials and a port map / cable schedule with cable IDs | live | /actions/network/SKILL.md |
| Rack elevation and power | devices → rack elevation (which device in which U), provisioned kW per rack, A+B circuits, load per feed, power cords per feed, colocation cost in every listed location | live | /actions/rack/SKILL.md |
| Proposal | the design written up for the person you are working with to take to their CFO, CTO or board: options, risks, timeline, total cost | coming | — |
Built by following the instructions with only this site's data; the arithmetic and catalog checks pass on each.
| Design | Hardware | Provisioned | Colo | Files |
|---|---|---|---|---|
| 2 PB usable archive on HDD storage nodes (RAID 6), Montana colo | $297,543.55 | 4.59 kW | $1,193/mo | Estimate PDF · HTML · MD; network PDF · MD · ports CSV; rack plan PDF · MD; design.json |
| 4× HGX B300 training cluster (32 GPUs) with NVMe storage and a services cluster, Montana colo | $3,307,252.84 | 82.51 kW | $21,453/mo | Estimate PDF · HTML · MD; network PDF · MD · ports CSV; rack plan PDF · MD; design.json |
| 4× 8-GPU RTX PRO 6000 inference farm (32 GPUs), Texas colo | $724,604.56 | 34.33 kW | $7,724/mo | Estimate PDF · HTML · MD; network PDF · MD · ports CSV; rack plan PDF · MD; design.json |