UpStationTalk to an Engineer

Home › Actions

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).

ActionWhat it doesStatusInstructions
Design and estimaterequirements → 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 IDlive/actions/estimate/SKILL.md
Network topologysized 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 IDslive/actions/network/SKILL.md
Rack elevation and powerdevices → 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 locationlive/actions/rack/SKILL.md
Proposalthe design written up for the person you are working with to take to their CFO, CTO or board: options, risks, timeline, total costcoming—

Filled examples

Built by following the instructions with only this site's data; the arithmetic and catalog checks pass on each.

DesignHardwareProvisionedColoFiles
2 PB usable archive on HDD storage nodes (RAID 6), Montana colo$297,543.554.59 kW$1,193/moEstimate 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$3,307,252.8482.51 kW$21,453/moEstimate 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$724,604.5634.33 kW$7,724/moEstimate PDF · HTML · MD; network PDF · MD · ports CSV; rack plan PDF · MD; design.json