Frequently Asked Questions

TW Compute is Taiwan's full-stack sovereign AI platform and a GPU-as-a-Service platform that provides dedicated AI computing infrastructure for enterprise clients. As Taiwan's leading AI data center service provider, we offer 400,000+ GPUs across 93 countries and 200+ locations, enabling enterprises to access bare-metal GPU clusters configured to their specifications, deployed in approximately 48 hours, with no egress fees and no vendor lock-in.

Our GPU resource network covers 93 countries and 200+ locations globally. We don't own data centers or hardware. Instead, we operate an asset-light marketplace platform, providing flexible GPU computing services to enterprises through partnerships with a distributed GPU network.

TW Compute is designed for the AI era, suitable for AI model training, inference, and large-scale data processing. We provide high-performance GPU clusters, supporting enterprises in configuring dedicated resources with no vendor lock-in, enabling your AI business to scale flexibly.

Our pricing is transparent and predictable with no hidden fees. We offer flexible billing options including hourly and monthly plans. Contact our sales team for a customized quote based on your specific GPU requirements, locations, and scale.

Yes. We provide customized trial engagements for prospective customers. Talk to our sales team about your needs and we will design a trial that lets you evaluate the platform before commitment.

Absolutely. Our service is highly elastic: you can change GPU type, region, and scale on demand. We support fast expansion so your AI business can grow without limits.

No. Unlike major cloud providers, we do not charge egress fees for data transfer. This makes costs more predictable for AI workloads and allows you to move data freely without worrying about unexpected charges.

We support a wide range of NVIDIA GPUs including H100, A100, H200, and other enterprise-grade options. Availability varies by location. Contact our sales team for specific GPU availability in your preferred regions.

We can configure and deploy dedicated GPU clusters within 48 hours. This is significantly faster than the 12-week average waiting time with major cloud providers like AWS and GCP.

We provide multiple connectivity options, including high-bandwidth fiber, BGP networking, and dedicated lines, so your business gets the best network performance and reliability.

Bare-metal GPU access means you get 100% of the GPU's performance without any virtualization overhead. Unlike cloud providers that use virtualization layers (which can reduce performance by up to 20%), our bare-metal approach delivers maximum GPU potential for your AI workloads.

We implement multi-layer security architecture including physical security (Tier 3/4 data centers), network security (enterprise-grade configuration), security operations (continuous monitoring and incident response), and access control (strict role-based and key-based access enforcement). All data centers are verified at every location.

Yes. We guarantee 99.99%+ uptime via N+1 redundancy, redundant power and cooling, and 24/7 monitoring to keep your AI workloads running without interruption.

Yes. With our global network covering 93 countries and 200+ locations, you can choose where your data is processed and stored. This is particularly valuable for organizations with data sovereignty requirements or latency constraints.

We offer 24/7 engineer support with SLA-guaranteed uptime. Our team provides continuous cluster monitoring, allowing you to focus on your models rather than infrastructure management. Enterprise clients also receive dedicated account management.

For urgent issues, we commit to responding within 15 minutes; for general questions, we respond within 1 hour. Our goal is to resolve every customer issue as quickly as possible so your business runs smoothly.

Yes. We provide comprehensive technical training for our customers, covering GPU cluster management, security best practices, and resource scaling. Our technical experts guide your team personally to make sure you get the most out of our service.

You can reach our support team through multiple channels: email, live chat, or through your dedicated account manager for enterprise clients. We also provide comprehensive documentation and API guides for self-service support.