Home/Pricing
Transparent pricingPublished rates, no discovery call required
Every list price Velyrix quotes is on this page. Large commitments, multi-year terms and wholesale colocation are quoted individually — and always below these numbers.
On-demand and reserved rates
| GPU | On-demand | 1 month | 6 months | 12 months | 36 months | Saving at 36 mo |
|---|---|---|---|---|---|---|
| GB300 NVL72 | $8.91 | $8.55 | $8.20 | $7.93 | $7.57 | 15% |
| B300 SXM (HGX) | $8.34 | $8.01 | $7.67 | $7.42 | $7.09 | 15% |
| B200 SXM (HGX) | $6.61 | $6.35 | $6.08 | $5.88 | $5.62 | 15% |
| H200 SXM (HGX) | $3.79 | $3.64 | $3.49 | $3.37 | $3.22 | 15% |
| H100 SXM (HGX) | $3.10 | $2.98 | $2.85 | $2.76 | $2.63 | 15% |
| H100 NVL / PCIe | $2.70 | $2.59 | $2.48 | $2.40 | $2.29 | 15% |
| A100 SXM 80GB | $1.90 | $1.82 | $1.75 | $1.69 | $1.61 | 15% |
| A100 SXM 40GB | $1.47 | $1.41 | $1.35 | $1.31 | $1.25 | 15% |
| RTX PRO 6000 Blackwell | $2.18 | $2.09 | $2.01 | $1.94 | $1.85 | 15% |
| L40S | $1.35 | $1.30 | $1.24 | $1.20 | $1.15 | 15% |
| RTX 5090 | $0.75 | $0.72 | $0.69 | $0.67 | $0.64 | 15% |
| RTX 4090 | $0.63 | $0.60 | $0.58 | $0.56 | $0.54 | 15% |
| L4 | $0.39 | $0.37 | $0.36 | $0.35 | $0.33 | 15% |
USD per GPU-hour, exclusive of tax. Multi-node InfiniBand pods add a fabric charge of $775 per node per month. No charge for standard egress. Velyrix list pricing sits 15% above the market median — see what the premium buys.
Whole-machine pricing
| Node | 1 month | 6 months | 12 months | 36 months |
|---|---|---|---|---|
| 8× B300 SXM 288GB | $44,800 | $41,200 | $39,900 | $38,100 |
| 8× B200 SXM 180GB | $38,000 | $35,000 | $33,800 | $32,300 |
| 8× H200 SXM 141GB | $23,900 | $22,000 | $21,300 | $20,300 |
| 8× H100 SXM 80GB | $18,400 | $16,900 | $16,400 | $15,600 |
| 8× A100 SXM 80GB | $10,400 | $9,600 | $9,300 | $8,800 |
| 8× A100 SXM 40GB | $8,000 | $7,400 | $7,100 | $6,800 |
| 8× H100 NVL 94GB | $14,700 | $13,500 | $13,100 | $12,500 |
| 8× RTX PRO 6000 Blackwell 96GB | $9,500 | $8,700 | $8,500 | $8,100 |
| 8× L40S 48GB | $6,000 | $5,500 | $5,300 | $5,100 |
| 8× RTX 5090 32GB | $3,800 | $3,500 | $3,400 | $3,200 |
| 8× RTX 4090 24GB | $2,700 | $2,500 | $2,400 | $2,300 |
| Compute — Dual EPYC 9654 | $1,490 | $1,370 | $1,330 | $1,270 |
| Storage — NVMe dense | $3,800 | $3,500 | $3,400 | $3,200 |
See the full dedicated server catalogue for configuration detail and market rate comparison.
Hosting and operating hardware you own
| Component | Unit | Price | Required? |
|---|---|---|---|
| Colocation (space, power, cooling) | per kW / month | $110 – $215 | Yes |
| Deployment & commissioning — air-cooled | per node, one-off | from $2,300 | Optional |
| Deployment & commissioning — liquid / rack-scale | per node, one-off | $4,600 – $11,000 | Optional |
| Fabric build & validation | per 64 nodes | from $66,000 | Optional |
| Managed operations (24×7 NOC, monitoring, RMA) | per node / month | from $500 | Optional |
| Spares depot management | per fleet / month | from $1,700 | Optional |
| Remote hands — scheduled | per hour | $230 | À la carte |
| Procurement assistance | per engagement | Disclosed before you agree | Optional — never required |
You are never required to purchase hardware through Velyrix. BYOG pricing is identical whether you buy direct from an OEM, through a distributor, or ask us to help. See how BYOG works.
Colocation
| Product | Density | Per kW / month |
|---|---|---|
| Retail cabinet | 10–20 kW | $215 |
| High-density cabinet | 20–40 kW | $190 |
| RDHx cabinet | 40–60 kW | $170 |
| Liquid-cooled pod | 60–160 kW | $155 |
| Private cage | 250 kW – 2 MW | $145 |
| Private suite | 500 kW – 6 MW | $130 |
| Wholesale hall | 2 – 40 MW | $110 |
Storage & network
| Service | Unit | Price |
|---|---|---|
| Parallel FS (WEKA/VAST) | per TB / month | $42 |
| Parallel FS — capacity tier | per TB / month | $22 |
| NVMe block storage | per TB / month | $100 |
| Object storage (S3 API) | per TB / month | $20 |
| Archive / cold object | per TB / month | $5.75 |
| 100G unmetered uplink | per port / month | $1,600 |
| Private cloud interconnect | per 10G / month | $545 |
| Single-mode fibre cross-connect | per connection / month | $315 |
| InfiniBand pod networking | per node / month | $805 |
Deployment, commissioning and LLM engineering
| Service | Unit | Indicative fee | Typical duration |
|---|---|---|---|
| Cluster design & site readiness review | per engagement | from $15,500 | 2–3 weeks |
| Deployment — air-cooled node | per node | from $2,300 | Scheduled |
| Deployment — liquid / rack-scale | per node | $4,600 – $11,000 | Scheduled |
| Fabric build & validation | per 64 nodes | from $66,000 | 1–2 weeks |
| Commissioning & acceptance testing | per cluster | from $34,500 | 1–2 weeks |
| Cluster health audit | per engagement | from $21,500 | 1–2 weeks |
| Private LLM pilot | per engagement | from $27,000 | 2–3 weeks |
| Private LLM production deployment | per engagement | from $83,500 | 6–10 weeks |
| Managed LLM endpoint | per month | from $10,500 | Ongoing |
| Managed Slurm / Kubernetes | per cluster / month | from $2,400 | Ongoing |
| Engineer day rate (T&M) | per day | $2,400 | As required |
Pricing terms
All prices are in US dollars and exclude VAT, GST, sales tax and any applicable withholding. Local currency invoicing is available in EUR, GBP, CAD and MXN.
Prices on this page are indicative list rates published for orientation. They are not an offer. Binding pricing is issued in a written quotation valid for thirty days and is subject to availability, region, term, configuration and credit approval.
List pricing is reviewed quarterly and may change without notice. Rates contracted for a committed term are fixed for that term.
On-demand usage is billed per second in arrears each month. Reserved and dedicated capacity is billed monthly in advance. Professional services are billed on milestone acceptance.
Capacity, delivery dates and regional availability are indicative until confirmed in a signed order. Certain products are export-controlled and are supplied only where permitted following screening.
Failure to meet a contracted availability target results in service credits applied automatically against the next invoice, per the service level agreement.
Own the GPUs. Let us run them.
Buy your NVIDIA servers from any OEM or distributor you like, ship them to a Velyrix hall, and we handle the rest — deployment, fabric, cooling, monitoring and support. Or rent ours. Either way, you get a plan in one business day.
Frequently asked questions
How much does an NVIDIA H100 cost per hour?
Velyrix lists NVIDIA H100 SXM at $3.10 per GPU-hour on demand, falling to $2.76 on a twelve-month reservation and $2.63 on thirty-six months. The list rate is set at 15 percent above the market median of about $2.70, and remains inside the prevailing market range of $1.90 to $3.50.
Is there a discount for long-term commitments?
Yes. Reservations reduce the effective rate by roughly 6 percent at one month, 11 percent at six months, 15 percent at twelve months and up to 20 percent at thirty-six months, with further discounts available on large-volume commitments.
Are there data egress charges?
No. Standard plans include unmetered egress on the included uplink. Dedicated high-capacity transit ports and private cloud interconnects are priced separately.
What is included in the GPU price?
Compute, the node's local NVMe scratch, standard networking and egress, out-of-band management, 24x7 NOC support and the pre-handover burn-in and validation. Parallel file system capacity, object storage, InfiniBand pod fabric and managed services are priced separately so you only pay for what your workload needs.