01
Transparent deposit billing
Place a deposit, then pay as you go — down to the hour. No contracts, no surprise charges, no hidden fees.
A flexible, transparent VPS service built for AI practitioners and creative studios. You deposit, we meter — and the remaining balance is always yours to withdraw.
Unused deposit balance returned in full, any time.
Engineer-led monitoring, day and night.
Per-project builds — no preset plans.
From payment to SSH, for standard builds.
CCX Asia Limited is a Hong Kong–based VPS provider serving creators, researchers, and growing studios across Asia. Rather than offering one-size-fits-all plans, we let you assemble the CPU, memory, storage, and GPU configuration that fits your project.
We believe in a simple principle: use only what you need, only when you need it. You start by placing a deposit, we draw from it as your resources run, and you can withdraw the remaining balance whenever you choose. No long-term contracts. No hidden costs.
We don't sell servers. We provide the quiet space your creative work needs to breathe.
A transparent deposit model where you pay for exactly what runs — and nothing more. No monthly lock-in. No cancellation fees.
Place any amount you like. We are flexible on both minimum and maximum balances.
Design your CPU, RAM, SSD, GPU and network profile together with our team.
Resource usage is metered and automatically drawn from your deposit balance.
Unused balance is refunded to your account in full — on cancellation or on request.
Deposits are held separately. Withdrawal requests can be made any time — funds are returned to your designated account within 3–5 business days of final settlement.
We focus on honest infrastructure that fits your actual workload — not glossy overprovisioning. Six reasons our customers stay with us.
Place a deposit, then pay as you go — down to the hour. No contracts, no surprise charges, no hidden fees.
Your unused balance is always refundable. "Cancel when you don't need it" should be the default — and here, it is.
Configure CPU, RAM, storage and GPU to match the project. From a small staging box to a GPU training rig.
Standard configurations deploy within minutes of payment. Bespoke builds are typically ready within one business day.
Based in Hong Kong with low-latency reach to Tokyo, Singapore and Shanghai — ideal for Asia-facing products.
Real engineers watch the metrics and tune the hosts — especially for the odd load patterns of AI and rendering.
The three profiles below are representative examples. Actual builds are fully tailored around your workload, duration and network requirements. We propose a configuration with nothing wasted.
For dev environments, small bots and static hosting.
For always-on production, mid-scale inference and build environments.
For AI training, large-scale rendering and video pipelines.
Dual-GPU rigs, low-latency networking only, massive cold storage — unusual requirements welcome. Builds range from vCPU 1 / RAM 1 GB up to multi-GPU configurations, all in one place.
Whether you are running language models, training variants, or rendering the next film — CCX Asia is built around the soft elasticity these workloads actually demand.
Run open-source or private language models on dedicated hosts. Handles both batch jobs and interactive APIs well.
Long-running browser agents, social ops bots, scrapers. Persistent sessions and ample memory guaranteed.
Spin up GPU capacity for LoRA, QLoRA or SFT runs. Scale back down afterwards and keep the unused deposit for next time.
Build private retrieval systems for your internal documents. Embedding models and vector DBs can live on the same host.
Scraping, transforming, batch aggregation — scale up a host just for the run, then shrink when the job is done.
Reproduce the latest papers in a clean, isolated environment — without cluttering your workstation with CUDA trees.
If anything isn't covered here, reach us directly — we prefer honest conversation over polished sales.
No fixed minimum. We are flexible for small trial projects and scale up comfortably for large studios. Talk to us about your expected usage and we'll suggest a reasonable starting amount.
Unused balance is returned to your designated account within 3–5 business days after final settlement of any running resources. You can request a refund at any time.
Yes. Resize CPU, RAM, storage, or swap in GPUs at short notice. We meter based on what's running, so temporary bursts are economical.
Our primary data presence is in Hong Kong, with strong routing to Tokyo, Singapore, Shanghai and Seoul. Clients worldwide are welcome.
Yes. We offer NVIDIA L4, A100 and similar GPUs for inference, fine-tuning and rendering. Multi-GPU configurations available on request.
Bank transfer in HKD, USD, JPY and major currencies. Crypto options available for select clients. We are happy to discuss what works best.
No need to have the requirements fully nailed down. We're happy to scope the configuration with you from the earliest planning stages.
Use case, duration, spec needs — we propose a fit.
We walk the procedure together once approved.
Minutes to one business day after payment.