Dedicated servers .
Every server is hand-built on Supermicro, Tyan, ASRock Rack, ASUS or Gigabyte platforms, burn-tested for 48 hours and racked by our own staff. Pick a platform, choose your RAM and storage — we handle the rest.
| Processor | Cores / Threads | Clock | RAM | Storage | Price | |
|---|---|---|---|---|---|---|
| Ryzen™ 9 5900XZen 3 · AM4 | 12C / 24T | 3.7–4.8 GHz | 32–128 GB DDR4 | 2× NVMe Gen4 | $250/mo | Configure |
| Ryzen™ 9 5950XZen 3 · AM4 | 16C / 32T | 3.4–4.9 GHz | 32–128 GB DDR4 | 2× NVMe Gen4 | $285/mo | Configure |
| Ryzen™ 9 7950XZen 4 · AM5 | 16C / 32T | 4.5–5.7 GHz | 32–192 GB DDR5 | 2× NVMe Gen4 | $350/mo | Configure |
| Ryzen™ 9 9950XZen 5 · AM5 | 16C / 32T | 4.3–5.7 GHz | 32–192 GB DDR5 | 2× NVMe Gen5 | $395/mo | Configure |
| Processor | Cores / Threads | Clock | RAM | Storage | Price | |
|---|---|---|---|---|---|---|
| Core™ i9-9900KCoffee Lake | 8C / 16T | 3.6–5.0 GHz | 32–128 GB DDR4 | 2× NVMe | $250/mo | Configure |
| Core™ i9-10900KComet Lake | 10C / 20T | 3.7–5.3 GHz | 32–128 GB DDR4 | 2× NVMe | $265/mo | Configure |
| Core™ i9-11900KRocket Lake | 8C / 16T | 3.5–5.3 GHz | 32–128 GB DDR4 | 2× NVMe Gen4 | $275/mo | Configure |
| Core™ i9-12900KAlder Lake | 16C / 24T | 3.2–5.2 GHz | 32–128 GB DDR5 | 2× NVMe Gen4 | $310/mo | Configure |
| Core™ i9-13900KRaptor Lake | 24C / 32T | 3.0–5.8 GHz | 32–192 GB DDR5 | 2× NVMe Gen4 | $365/mo | Configure |
| Core™ i9-14900KRaptor Lake Refresh | 24C / 32T | 3.2–6.0 GHz | 32–192 GB DDR5 | 2× NVMe Gen4 | $385/mo | Configure |
| Platform | Cores / Threads | RAM | Storage | Network | Price | |
|---|---|---|---|---|---|---|
| EPYC™ 4004 seriesSingle socket · AM5 | Up to 16C / 32T | 32–192 GB DDR5 | NVMe Gen4/5 arrays | 1–25 Gbps | $295/mo | Configure |
| EPYC™ 9004 “Genoa”Single / dual socket · SP5 | Up to 192C / 384T | 64 GB–1.5 TB DDR5 | NVMe / NVMe-oF arrays | 10–400 Gbps | $495+/mo | Configure |
| EPYC™ / Threadripper™ customAny chassis, any vendor we carry | Custom | Up to 1.5 TB | Custom | 10–400 Gbps | Quote | Request a build |
| GPU platform | VRAM | Host CPU / RAM | Use case | Price | |
|---|---|---|---|---|---|
| NVIDIA® RTX™ 5070Blackwell · 12 GB GDDR7 | 12 GB | Ryzen™ 9 · 64–128 GB | Dev inference · render | $395/mo | Configure |
| NVIDIA® RTX™ 5090Blackwell · 32 GB GDDR7 | 32 GB | Ryzen™ 9 / EPYC™ · 128–256 GB | LLM inference · training | $595/mo | Configure |
| NVIDIA® H100 80GBHopper · SXM/PCIe | 80 GB HBM3 | Dual EPYC™ · up to 1.5 TB | Serious training · HPC | $1,995/mo | Configure |
| AMD Instinct™ MI210CDNA 2 · ROCm stack | 64 GB HBM2e | EPYC™ · 256–512 GB | ROCm workloads · HPC | $695/mo | Configure |
| NVIDIA® DGX™ SparkGrace Blackwell · desktop AI node | 128 GB unified | Grace CPU · self-contained | Edge AI · prototyping | $349/mo | Configure |
| Custom GPU buildsChina-sourced · 40 GB VRAM class cards | 40 GB | EPYC™ / Ryzen™ · as spec'd | Multi-GPU density · budget AI | Quote | Request a build |
Multi-GPU chassis (2–8×) available on all platforms. GPU servers include the same 333 TB transfer, IPMI and hardware SLA as every Cyberheat build.
We take cards, ACH/wire — and crypto, settled at invoice.
Installed your way at deploy time — automated reinstalls and custom ISOs available over IPMI.
Xeon®, Threadripper™ PRO, custom storage arrays, GPU density — if it fits in a rack, we'll build it.