Moyan Technology
Broker — Taiwan & China direct

Sourced by spec.
Quoted by RFQ.

The GPUs, memory, networking, power and cooling that build AI clusters — bought direct from the factory floor, priced across your whole bill of materials.

Request a quote Browse catalog 140 SKUs · 10 categories
Indicative market readout — as of 2026-08 · confirm by RFQ
GPU / Accelerator
NVIDIA H100 SXM5 80GB
$35,000 – $40,000
High-Bandwidth Memory
HBM3E 8-Hi 24GB stack
On request
InfiniBand
NVIDIA Quantum-2 MQM9700-…
$25,500 – $44,462
Liquid Cooling
GPU direct-to-chip cold p…
$300 – $500
Why a broker

Access and clarity in a market that hides its prices

01

Factory-direct

The Taiwan and China ODMs that build the world’s AI servers — not a reseller three steps removed.

02

One RFQ, whole BOM

Quote GPUs, memory, networking, power and cooling together — not across ten vendors.

03

Honest pricing

Dated market ranges, confirmed live at RFQ — never a stale list price.

04

Allocation-aware

Sold-out parts — HBM, latest GPUs, cold plates — with real lead times and second sources.

FAQ

Common questions

Do you publish fixed prices?

No. Prices for AI datacenter hardware change weekly and most high-value parts have no public list price. We show indicative market ranges with an "as of" date and confirm the current figure by RFQ.

Do you sell single units or only bulk?

Both, depending on the part. Some categories have low minimums; rack-scale systems and allocation-controlled parts are quoted per project.

Where do you source from?

Direct from Taiwan and China factories, ODMs and authorized channels. GPU availability can depend on the destination jurisdiction, which we confirm per quote.

Can you quote a whole cluster?

Yes. Send a target spec — GPU count, memory, storage, fabric, cooling — and we return an indicative BOM you can refine.

Ready to price your build?

Send your part list or a target cluster spec for an indicative full-BOM quote.

Request a quote