Dedicated capacity, thoughtfully planned

Your next deployment
starts here.

Reserve dedicated NVIDIA B300 HGX servers through a conversation—not a checkout. We’ll define the hardware, timing and service around your requirements.

Platform

NVIDIA B300 HGX

Dedicated servers with eight GPUs per server. Minimum reservation: one server.

Reservation term

36 / 48 / 60 months

Choose a term to discuss. Pricing and payment structure are set in your proposal.

Configuration

Built for your needs

Ethernet networking, with storage, networking and other configuration details agreed before signing.

Close-up reference image of an NVIDIA accelerator chip package

Defined around your workload

The details matter.

Your reservation is for dedicated B300 HGX servers—not just an hourly allocation. We’ll discuss the configuration and the infrastructure service together, so you know what you’re reserving.

Final server vendor, component specifications, storage and connectivity are set out in your quote and agreement. Reference imagery does not guarantee a particular OEM or configuration.

Discuss a custom configuration →

Deployment roadmap

64 B300 GPUs.
Each quarter of 2027.

We’re speaking with customers now to plan reservations ahead of deployment.

01

Q1 2027 · +64 B300 GPUs planned

64 GPUs cumulative target capacity.

02

Q2 2027 · +64 B300 GPUs planned

128 GPUs cumulative target capacity.

03

Q3 2027 · +64 B300 GPUs planned

192 GPUs cumulative target capacity.

04

Q4 2027 · +64 B300 GPUs planned

256 GPUs cumulative target capacity.

These are deployment targets, not guaranteed delivery dates or live inventory. Timing is subject to final configuration, procurement and facility readiness; your agreement will define acceptance and service commencement.

From first conversation to deployment

A clear path forward.

01

Share the workload.

Tell us your GPU or server count, workload, target start date and preferred term. Add storage, network or support requirements if you know them.

02

Shape the proposal.

We’ll work through configuration and provide a quote covering pricing, payment terms, connectivity, support scope and separately scoped custom work.

03

Plan the deployment.

After agreement, we coordinate procurement, provisioning and acceptance. Availability and all commitments are confirmed in the signed reservation agreement.

Good questions

Why request a quote instead of choosing a public rate?

A useful comparison includes the hardware, term, prepayment, connectivity and support—not just the GPU-hour. We’ll give you a proposal for the deployment you actually need.

Can I start with a single server?

Yes. The minimum reservation is one dedicated eight-GPU B300 HGX server. You receive the same personal attention whether you reserve one server or plan a larger deployment.

Can the configuration evolve?

We can discuss storage, networking, additional compute and future hardware as your requirements evolve. Changes depend on technical feasibility and availability, and are separately scoped and priced. Future upgrades are not automatically included.

What is included in infrastructure service?

Power and cooling, an agreed connectivity allocation, infrastructure monitoring, hardware maintenance and direct operator support. Your quote defines bandwidth, any transfer-related charges, support hours and service commitments. Software and model administration are not included unless separately agreed.

Start with a conversation

Your workload.
Our full attention.

Tell us what you want to build. We’ll work through the configuration, timing and service scope with you.

Discuss your deployment