UpStation catalog, short version (https://ai.upstation.io/brief.txt). Map of every page: https://ai.upstation.io/llms.txt. Homepage with every table: https://ai.upstation.io/. JSON: https://ai.upstation.io/catalog.json.
UpStation sells pre-priced NVIDIA GPU servers, network switches, optics and cables, and colocation, at published prices, so buyers and their AI assistants can design and price a deployment without a sales call. Prices in USD as of 2026-09-09; final price confirmed on order. Continental US; Canada quoted case by case. Contact alec@upstation.io.
Rules: one priced base build per class; Intel/AMD equivalents at a host delta; resize memory/CPU/drives with the rate card; options have price deltas; anything marked quote is quoted by UpStation. Tier 1 = OEM factory build, Tier 2 = system builder, BYO = host with 8 slots and no GPUs.
Power: provisioned kW = nameplate W ÷ 0.8; A+B feeds. Colo on provisioned kW: $175/kW/mo 5-yr, $225/kW/mo 3-yr, $250/kW/mo 1-yr.

GPU servers (8 GPUs): price — GPUs · CPUs · memory · storage · network · nameplate kW → provision kW — part — lead — page
- NVIDIA B300 (HGX NVL8), Tier 1 OEM build: $572,364 — 8× B300 · 2× 64C Intel · 2TB · 1× 960GB M.2 · 8× 800G · 13.6 kW → 16.96 kW — Supermicro SYS-822GS-NB3RT-01-G2 — Ships 1-3 wk — https://ai.upstation.io/p/UPS-GPU-B300-8
- NVIDIA B200 (HGX): out of stock industry-wide, see B300
- NVIDIA H200 (HGX), Tier 1 OEM build: $319,421.64 — 8× H200 · 2× 48C Intel · 1TB · 1× 240GB SATA · 2× 25G · 9.1 kW → 11.34 kW — Supermicro SYS-821GE-TNHR — 6-10 wk — https://ai.upstation.io/p/UPS-GPU-H200-8
- NVIDIA RTX PRO 6000 Server Edition, Tier 1 OEM build: $191,888.10 — 8× RTX PRO 6000 · 2× 72C Intel · 1TB · 2× 480GB + 1× 3.84TB · 2× 10G · 6.8 kW → 8.47 kW — Supermicro SYS-422GA-NRT — 3-5 wk — https://ai.upstation.io/p/UPS-GPU-RTXPRO-8
- NVIDIA RTX PRO 6000 Server Edition, Tier 2 system builder: $163,909.60 — 8× RTX PRO 6000 · 2× 96C AMD · 768GB · 1× 3.84TB + 2× 1TB · 2× 25G · 6.7 kW → 8.37 kW — Octoserver 4U-E10G-GENOAX2 — 3-5 wk — https://ai.upstation.io/p/UPS-GPU-RTXPRO-8-T2
- NVIDIA RTX 5090 (passive server edition), Tier 2 system builder: $76,679.60 — 8× RTX 5090 · 2× 96C AMD · 512GB · 1× 3.84TB + 1× 1TB · 2× 25G · 6.4 kW → 7.99 kW — Octoserver 4U-E10G-GENOAX2 — 5-6 wk — https://ai.upstation.io/p/UPS-GPU-5090-8
GPU hosts, no GPUs (8 slots):
- $47,884.46 — no GPUs, 8 slots · 2× 72C Intel · 512GB · 2× 960GB · 2× 25G — Supermicro SYS-422GA-NRT — 3-5 wk — https://ai.upstation.io/p/UPS-GPU-BARE-ENT-8
- $37,937.10 — no GPUs, 8 slots · 2× 28C Intel · 512GB · 2× 960GB · 2× 25G — Supermicro SYS-521GE-TNRT — 3-5 wk — https://ai.upstation.io/p/UPS-GPU-BARE-ENT-8-X13
- $39,179.76 — no GPUs, 8 slots · 2× 32C AMD · 512GB · 1× 960GB · 2× 25G — Supermicro AS-5126GS-TNRT — 3-5 wk — https://ai.upstation.io/p/UPS-GPU-BARE-ENT-8-AMD
- $32,919.60 — no GPUs, 8 slots · 2× 32C AMD · 512GB · 2× 960GB · 2× 25G — Octoserver 4U-E10G-GENOAX2 — 2-3 wk — https://ai.upstation.io/p/UPS-GPU-BARE-T2-GEN5
- $15,146.20 — no GPUs, 8 slots, PCIe 4.0 · 2× 32C AMD · 512GB DDR4 · 2× 960GB · 2× 25G — Octoserver 4U-E10G-ROMEX2 — 2-3 wk — https://ai.upstation.io/p/UPS-GPU-BARE-T2-GEN4
Switches (FS, PicOS included; full table, cables and optics at https://ai.upstation.io/switches): ports — price — part
- 64× 800G OSFP — $44,600 — N8650-64OD, PicOS — 1,800 W
- 32× 800G OSFP — $37,499 — N8650-32OD, PicOS — 1,000 W
- 64× 400G QSFP-DD — $23,999 — N8610-64DB, PicOS — 1,000 W
- 32× 400G QSFP-DD — $19,999 — N8610-32D, PicOS — 600 W
- 24× 200G + 8× 400G — $12,400 — N8510-24CD8D, PicOS — 600 W
- 64× 100G QSFP28 — $11,900 — N5630-64C, PicOS — 550 W
- 32× 100G QSFP28 — $7,919 — N8560-32C, PicOS — 250 W
- 48× 25G + 8× 100G — $6,049 — S5470-48Y, PicOS — 150 W
- 48× 10G + 8× 100G — $4,569 — S5860-48SC — 120 W
- 48× 1G RJ45 + 4× 10G — $1,089 — S3410-48TS, PicOS — 60 W
Cables/optics (reach in brackets): 800G: 2 m DAC $239 (passive to 3 m), AEC 5 m $1,424 (max 5 m), AOC 10 m $3,011 (max 50 m), SR8 $879 (50 m) each, DR8 $1,319 (500 m) · 2×FR4 $1,869 (2 km), breakout 800G→2×400G 2 m $299; 400G: 2 m DAC $174 (passive to 3 m), Active DAC 5 m $611 (max 5 m), AOC 10 m $1,055 (max 30 m), SR8 $219 (100 m) each, DR4 $549 (500 m) · FR4 $719 (2 km) · LR4 $999 (10 km), breakout 400G→4×100G 2 m $199; 200G: 2 m DAC $120 (passive to 3 m), Active DAC 7 m $325 (max 7 m), AOC 10 m $1,034 (max 50 m), SR4 $299 (100 m) each, FR4 $1,249 (2 km) · LR4 $2,624 (10 km), breakout 200G→2×100G 2 m $119; 100G: 2 m DAC $54 (passive to 5 m), Active DAC 9 m $300 (max 9 m), AOC 10 m $221 (max 30 m), SR4 $99 (100 m) each, CWDM4 $209 (2 km) · LR4 $399 (10 km), breakout 100G→4×25G 3 m $136; 25G: 2 m DAC $39 (passive to 5 m), Active DAC 9 m $73 (max 9 m), AOC 10 m $73 (max 30 m), SR $49 (100 m) each, LR $74 (10 km); 10G: 2 m DAC $18 (passive to 7 m), Active DAC 10 m $105 (max 10 m), AOC 10 m $39 (max 30 m), SR $25 (300 m) each, LR $34 (10 km). PicOS licence included with each switch. 1-yr PicOS support: $369 at 100G, $439 at 400G, $659 at 800G. Same port counts on Dell/SONiC or Arista: quote.
Compute and storage nodes:
- $33,298.33 — 1U · 2× 32C Intel · 512GB · diskless, 12 NVMe bays · 2× 25G · 1.0 kW — Supermicro SYS-122C-TN — 3-5 wk — https://ai.upstation.io/p/UPS-CN-INT-1U
- $182,994.59 — 2U · 2× 32C Intel · 512GB · 24× 7.68TB NVMe, 184TB · 2× 100G · 1.3 kW — Supermicro SYS-222C-TN — 3-5 wk — https://ai.upstation.io/p/UPS-SN-INT-NVME
- $51,743.45 — 2U · 1× 16C AMD · 128GB · 24× 24TB HDD, 576TB · 2× 100G · 0.6 kW — Supermicro ASG-2015S-E1CR24H — 3-5 wk — https://ai.upstation.io/p/UPS-SN-AMD-HDD
Parts rate card: memory DDR5-6400 32GB $46.06/GB; DDR5-6400 64GB $39.53/GB; DDR5-6400 96GB $46.08/GB; DDR5-6400 128GB $47.23/GB; DDR5-4800 32GB (Octoserver) $39.53/GB; DDR4-3200 32GB (Octoserver Rome host) $13.44/GB; AMD EPYC 9005/9004, mainstream 16-96C (Supermicro) $31–95/core; Intel Xeon 6700P, mainstream 32-64C (Supermicro) $101–166/core; Intel Xeon 6900P, 72-128C (Supermicro) $58–82/core; AMD EPYC 9004/9005 at Octoserver $29–51/core; AMD EPYC 7003 at Octoserver (PCIe 4.0 host) $13–29/core; Nearline HDD 16-30TB SAS/SATA (Supermicro) $49–63/TB; NVMe Gen4 U.2 TLC 1.9-15TB (Supermicro) $690–1390/TB; NVMe Gen5 U.2/E1.S/E3.S (Supermicro) $715–1029/TB; SATA/SAS SSD (Supermicro) $678–1050/TB.
Not in this catalog: NVIDIA B200; GB200 / GB300 NVL72 rack-scale systems and liquid-cooled racks; AMD Instinct; 2-GPU and 4-GPU servers; Dell, HPE and Lenovo priced builds; InfiniBand switches, and Arista or Dell/SONiC Ethernet switches; Vendor-coded optics; Racks, PDUs, structured cabling and installation labour; Software licences other than the PicOS switch OS; managed services; GPU cloud or rental.
Sizing (full rules https://ai.upstation.io/sizing, worked designs https://ai.upstation.io/designs): 8 GPUs per node, nodes = ceil(GPUs ÷ 8); HGX (B300, H200) when the GPUs must act as one, PCIe (RTX PRO 6000, RTX 5090) when each GPU works alone. GPU fabric 1 port per GPU (H200 400G, B300 800G) at 1:1: one switch while the ports fit it, else on R-port switches leaves = ceil(ports ÷ (R/2)), spines = max(2, ceil(leaves × R/2 ÷ R)); front-end 2 ports per node at 3:1; OOB 1× 1G per node and per switch. Storage: read GB/s = GPUs × 0.15 (text) to 0.8 (vision), aggregate; write = max(½ read, checkpoint GB ÷ pause s); plannable TB = raw × efficiency (RAID 6 10+2 0.833, EC 4+2 0.667, replica 3 0.333; Ceph also × (N−1)/N) × 0.85; nodes = max(by GB/s, by TB, protection floor). Racks: 40 U and 40 kW unless the site says otherwise: 2× B300, 3× H200, 4× RTX PRO 6000 Tier 1, 4× RTX PRO 6000 Tier 2, 5× RTX 5090 per rack.
Reference: Four 8× B300 nodes, 5-year colo = hardware $2,289,456, 67.84 kW, colo $11,872/mo.