Dedicated GPU · No Hypervisor Overhead · Direct GPU Access

Taiwan GPU Cloud at
30% Less Than the Competition

Dedicated GPU instances deployed in under 2 minutes. No hidden fees, no shared tenancy. Priced transparently — targeting 30% below the competition.

< 2 min
Deploy time
30% cheaper
vs competition
99.9%
Uptime SLA
Taiwan DC
Low latency · Local
Local Co.
Mandarin · TW timezone
Tax Invoice
Academic billing

Trusted by researchers at

NTUT
NTU
NYCU
TSMC
Inventec
MediaTek
NCKU
NCHU
Taipei Medical University
NTUT
NTU
NYCU
TSMC
Inventec
MediaTek
NCKU
NCHU
Taipei Medical University
NTUT
NTU
NYCU
TSMC
Inventec
MediaTek
NCKU
NCHU
Taipei Medical University
NTUT
NTU
NYCU
TSMC
Inventec
MediaTek
NCKU
NCHU
Taipei Medical University

Real-time GPU Infrastructure

Prices benchmarked against the market · We target 30% below market rate

Live
High Usage

RTX PRO 6000 WS 96GB

Blackwell · 96 GB VRAM

NT$34.36/hr

Reference range: NT$27~42/hr

Rent Now
Live
High Usage

RTX 5090 32GB

Blackwell · 32 GB VRAM

NT$17.94/hr

Reference range: NT$10~27/hr

Rent Now
Live
High Usage

RTX 3090 24GB

Ampere · 24 GB VRAM

NT$7.19/hr

Reference range: NT$4~12/hr

Rent Now
High Usage

DGX Spark GB10 4TB

GB10 Blackwell · 128 GB VRAM

NT$26.00/hr

Exclusive to GPUtw.ai

Book
High Usage

RTX 3090 x2 48GB NVLink

Ampere · 48 GB VRAM

NT$3.84/hr

Reference range: NT$8~24/hr

Book
All machines busy

H100

Hopper · 80 GB VRAM

NT$53.44/hr

Reference range: NT$53~75/hr

Request this GPU
All machines busy

V100 x8 NVLink 128GB

Volta · 128 GB VRAM

NT$34.56/hr

Reference range: NT$32~52/hr

Request this GPU

Custom Configuration

Need a specific GPU, multi-node cluster, or long-term contract? Contact us to discuss your requirements.

Contact Us

How it works

From GPU selection to live SSH — three steps, under two minutes

01

Pick your GPU

Browse the catalog and choose the model and specs that fit your workload. Availability and pricing update in real time.

02

Deploy in under 2 minutes

Provisioning starts immediately after you submit. SSH access is ready before your coffee brews — no queues, no waiting lists.

03

Pay by the second, stop anytime

Billed hourly, charged per second. Terminate the moment your job finishes — no minimum fees, no penalties.

Made in Taiwan · Researchers First

Built Local, Built for Researchers

Our team comes from an academic research background — we know what researchers actually need. From official tax invoices to academic discounts, GPUtw.ai fills the gap that international cloud providers leave behind.

Taiwan Data Center

Data stays in Taiwan. Low latency, fully compliant.

Taiwan Company

Mandarin support, Taiwan-timezone service.

Official Tax Invoice

Issue triplicate uniform invoices for schools and companies.

Research Support

Academic backgrounds on our team — we understand research workflows.

Academic Discounts

Institutions and research groups can negotiate custom rates.

Team Accounts

Share GPU resources with your team

Convert a personal account into a team account, invite collaborators, and let member deployments bill to one managed balance with owner-controlled limits.

Shared billing

Member usage can draw from the team owner balance for cleaner accounting.

Member controls

Set monthly spend limits, instance caps, and deploy suspension per member.

Balance visibility

Choose whether members can see the available team balance.

Create team account

Ready to deploy?

Sign up free. Pay only for what you use — billed by the second, no minimum fees.

Create Free Account