Deploy and own NVIDIA H200 servers
Hopper with 141 GB of memory per GPU. Proven, lower power, and quick to ship.
HGX H200 at a glance
- GPUs
- 8x NVIDIA H200 (Hopper) with NVLink
- GPU memory
- 141 GB per GPU, 1.1 TB per server
- CPUs
- 2x Intel Xeon or AMD EPYC
- Networking
- Usually 8x ConnectX-7 at 400 Gb/s
- Form factor
- Air-cooled, typically 8U
- Power
- About 10 kW per server
- Ballpark price
- $340K per server
Availability
H200 is the easiest current GPU to get, and servers usually ship in weeks. At about 10 kW, they also fit more data centers.
What you need to run H200s
Power
About 10 kW per server. Most high-density data centers can host them.
Networking
For a cluster, a 400G network of switches, optics, and cables.
Compliance
NVIDIA’s partners check the buyer, the site, and the use before they ship.
Setup
Racking, cabling, and testing before handover.
We manage. You own.
Scope it
Tell us your budget, timeline, and power. You get a plan and ballpark costs.
One call
We procure it
No preferred vendors. We source servers, networking, and data center space at today’s best price, in your name.
60-90 days typically
We run it
Operations and maintenance (O&M), software, and customers, on an annual contract.
Ongoing
What it costs
Flat fees, not revenue-share.
Paid to vendors, in your name
Hardware
Servers, networking, shipping, and installation, with nothing added.
Colocation
Space, power, and cooling, usually on a 3-year contract.
Parts, bandwidth, and insurance
Spare parts, internet, and property insurance on the hardware.
Paid to us, flat
Setup fee
A flat $25K for 1-7 servers, quoted for 8 or more.
Management fee
From $1K a month per 4 servers.
Want a customer for the GPUs? We charge a small percentage of the contract, only if it signs.
Tell us what you’re planning.
- A call with our CTO about what you need
- Ballpark costs and timelines for your budget
- A straight answer if it’s not a fit
Prefer email? hello@amcompute.com
NVIDIA H200 servers: questions
- How much does an H200 server cost?
- Ballpark $340K for a new 8-GPU server, before networking, shipping, and the data center.
- H200 or B200?
- H200 costs less, draws less power, and ships sooner. B200 has more memory and much more compute.
- HGX H200 or H200 NVL?
- HGX H200 links all eight GPUs with NVLink, and it’s what we deploy for clusters. H200 NVL is a PCIe card version.
- Is H200 still worth buying?
- For inference on models that fit its memory, yes. The software is mature and resale values have held up.
Further reading
More hardware and hosting
- NVIDIA B300 serversBallpark pricing, power, and lead times for HGX B300 servers.
- NVIDIA B200 serversBallpark pricing, power, and availability for HGX B200 servers.
- NVIDIA RTX PRO 6000 serversBallpark pricing and power for RTX PRO 6000 servers, the lower-cost way in.
- NVIDIA GB300 NVL72 racksWhat it takes to own a liquid-cooled GB300 NVL72 rack.
- GPU colocationA U.S. data center for the GPU servers you own.
- GPU colocation directoryHigh-density and liquid-cooled data centers by metro: power, kW per rack and liquid cooling, each with a source.
- GPU server vendorsWho sells GPU servers worldwide, from the public partner lists of NVIDIA, Supermicro, Dell, HPE, Lenovo, Cisco, Gigabyte, ASUS and AMD, with server models by maker.
Who it’s for
- AI startupsOwn your GPU servers instead of renting cloud GPUs.
- Universities and research labsA private GPU cluster in your institution’s name, run for you.
- Companies and enterprisesRun AI models on GPU servers your company owns, hosted and run for you.
- Family offices and fundsOwn GPU servers in your own entity and rent them to AI companies.
- Individual investorsOwn one or a few GPU servers through your LLC, rented out and run for you.
