Serving AI teams worldwide, from Singapore.
GPU cloud and AI inference on AMD Instinct GPUs. Headquartered in Singapore, powered by regional data centres.
Three ways to work with us.
One AMD based platform. Choose the option that fits how your team builds.
Dedicated or on demand AMD Instinct capacity for training, fine tuning and inference.
- Bare metal or virtualised
- Reserved or on demand terms
- ROCm software stack
Open weight models served through an API and billed per token, so you pay for what you use.
- OpenAI API compatible
- Usage based billing
- Served on ROCm
Dedicated deployments scoped around your security, data location and procurement requirements.
- Dedicated clusters
- Deployment location options
- Direct engineering contact
Headquartered in Singapore. Powered by regional data centres.
Compute hosted in the region, with a Singapore base for governance and support.
Singapore
Company headquarters, customer contracts, platform operations and engineering.
Regional data centres
GPU capacity hosted in regional data centre facilities, with dedicated options for teams that need separation.
Have compliance requirements?Share them early. We will walk your security and compliance teams through how a deployment would be set up.
Why AMD Instinct.
A platform chosen for memory capacity and open software.
Large GPU memory
High bandwidth memory well suited to large language models and long context work.
Open software
The ROCm ecosystem works with popular frameworks such as PyTorch.
Portable by design
OpenAI API compatible endpoints and open standards help keep your code portable.
Engineer access
Work directly with the engineers who run the platform.
How we work together.
Tell us the workload
Models, scale, timelines and requirements.
Get a proposal
GPU Cloud, Token Factory or both, scoped and priced.
Onboard
Access and setup with our engineers.
Grow
Expand capacity as your needs change.
Let's plan your AI capacity.
Talk to us about GPU capacity, Token Factory access or partnership opportunities.