Skip to content

gpupartner.com — faq

Frequently asked questions

If your question isn’t here, email us — see the footer for contact.

What is GPU Partner?
A white-glove GPU compute brokerage. Tell us what you need; we source reserved capacity through deep partner relationships and respond with options within 24 hours.
What GPUs do you broker?
NVIDIA H100 (SXM/PCIe), H200, B200, GB200, GB300, the Vera Rubin generation, and the Rubin NVL line — plus Ampere A100 where still available. AMD Instinct MI300X and the MI3xx line. New SKUs are added as partner inventory becomes available.
Do you only broker reserved capacity, or whole systems too?
Both. We broker reserved GPU capacity from cloud and bare-metal partners, AND turnkey OEM systems — NVIDIA DGX, Supermicro HGX, Dell PowerEdge, HPE Cray, Lenovo ThinkSystem — plus custom rack-scale builds with storage and networking. If you need to own the hardware (not rent the slice), we source the system and coordinate the integrator who racks it.
How is this different from AWS, GCP, or Lambda?
We are a brokerage, not a hyperscaler. Major clouds list a price; we negotiate per deal across multiple providers and bring you the best fit for your requirements (term length, region, interconnect, transferability). We do not host the compute ourselves.
How do I get a quote?
Pick a model from the catalog, configure your allocation in the wizard, and submit. Our brokerage team responds within 24 hours with one or more sourcing options. No public list pricing — quotes are bespoke per deal.
How long are quotes valid?
The exact expiry is shown on each quote in your dashboard — typically a few days, since GPU pricing moves and partner inventory shifts. Once expired, we re-source on request.
Who are your partners? Will I know who is hosting before I sign?
We work with a network of cloud, bare-metal, and independent GPU operators. The specific partner is disclosed before you sign each contract, so your security and compliance team can review the operator on every deal. We do not publish the full partner list because some operators are private and our negotiated rates are confidential.
What is the minimum deal size?
We focus on reserved-capacity deals starting at 8 GPUs. Smaller asks (1–4 GPUs) are usually best served by hourly cloud providers; if that fits your workload we will say so rather than waste your time.
How do you make money?
We earn a margin on each contract, paid by the partner. Your quote reflects the all-in price you pay — no separate platform fees, no monthly minimums, no charges added to your invoice after the fact.
How do I accept a quote?
On the RFQ detail page, click "Accept quote" and confirm. We draft a contract immediately and email a per-deal addendum for you to sign. Acceptance requires a one-time signature on the Master Services Agreement (MSA).
What is the MSA?
A one-time master agreement that governs the relationship; you sign it once and it covers all subsequent contracts. Each accepted quote then has its own short addendum specific to that deal.
How do you accept payment?
We invoice directly per contract addendum — NET-30 standard, custom terms negotiable on multi-year deals. Wire and ACH supported.
Can I cancel an RFQ?
Yes — you can cancel any RFQ before accepting a quote. Once a contract is created (after acceptance), cancellation depends on the addendum terms.
What countries do you serve?
Most jurisdictions. AI compute hardware is subject to U.S. export-control rules; some country / SKU combinations require additional review or are restricted entirely. We will let you know if your request needs review.
How do I delete my account?
Visit /settings/privacy and request deletion. Per GDPR Article 17, your personal data is anonymized after a 7-day cancellation window. Transactional records (RFQs, contracts, audit log) are retained per legal obligation.
Where do I see analytics about my account?
There is no buyer-side analytics dashboard yet. Your RFQs, quotes, and contracts are visible in /my-rfqs and /my-contracts.

Last updated May 2026 · Browse the catalog