Your data stays on YOUR devices — never uploaded to the cloud.
Built on disaggregated storage-compute architecture
and EBOF all-flash storage for secure, efficient, scalable local AI.
AI SALES
From intelligent product selection and financing lease pilots to TCO analysis and partner pre-screening — visitors can complete their first round of evaluation and lead capture without needing a sales rep.
AI quality inspection, predictive maintenance, and edge inference deployment
Quickly set up AI teaching, training, and research platforms
On-premise data sovereignty and medical imaging inference scenarios
Submit a regional partnership application with automated pre-screening
PRODUCT OVERVIEW
WeCalc integrates compute, storage, and management into a single system, following the philosophy of “minimum viable, linear scaling, infinite possibilities” — helping enterprises start fast with a single unit and evolve smoothly to production-grade clusters.
Financing Lease
Equivalent to approximately ¥40,000 in ChatGPT token credits — ideal for enterprises and institutions looking to adopt local AI capabilities with minimal upfront investment.
Ships with fully integrated hardware and software — no specialized setup required.
One device delivers a complete computing service, ideal for quick pilots.
Power on and launch computing services with one click, lowering deployment barriers.
End-to-end turnkey service from deployment to operation, completed in 48–72 hours.
Supports hot-swap expansion, scaling smoothly from a single unit to a cluster.
Flexibly accommodates CPUs, GPUs, SSDs, and more.
Huawei Ascend & Kunpeng certified, compatible with 90%+ of domestic GPUs.
PRODUCTS
From a single unit to clusters of thousands — covering everything from teaching labs to large-scale model training
Best 1P StarterBasic
Designed for small-scale AI inference, data analytics, and training labs. A single unit delivers complete computing capability.
Reference Price
¥98,000
Lease from ¥2,000/month
Production WorkhorseProfessional
For mid-scale AI training & inference and industrial edge computing. The go-to production-grade model.
Reference Price
¥2–5M
Most popular production solution
Flagship Custom SolutionEnterprise
Built for large-scale model training and HPC with PB-level storage and 200G/400G high-speed interconnect.
Reference Price
¥5M+
Custom delivery by scale
TECHNOLOGY
The only known team in China accelerating compute through hardware-accelerated all-flash storage
NVMe-oF protocol fully decouples compute and storage at the physical level. RoCEv2 high-speed interconnect eliminates TCP/IP congestion bottlenecks.
Latency ≤100μs | Cluster Bandwidth ≥100Gbps
PCIe Gen5 gateway-free distributed storage pool. 8+2 erasure coding achieves high-grade data protection with only 20% redundancy overhead.
IOPS ≥1M | Bandwidth ≥56GB/s
The system is deployed inside the customer’s premises. All data stays on your own devices — never uploaded to the cloud — ensuring data sovereignty from the ground up.
Zero Leak Risk | Full Compliance
VERIFIED
Huawei benchmark at equivalent ¥1.2M cost — outperforming traditional solutions across the board
BUSINESS MODEL
Inspired by the sharing economy, transforming computing power from heavy-asset purchases into lightweight shared services
Launch quickly with a pilot deployment or a ¥2,000/month financing lease to gain 1P of on-premise computing power.
48–72 hour turnkey deployment. Data stays on local devices — more secure, more compliant.
From a single unit to multi-node clusters, compute and storage grow smoothly with business demand.
Combining partner networks, O&M support, and upgrade services to build a long-term sustainable computing infrastructure.
Partners & Clients
NEWS
The WeCalc core team was invited to the Huawei Developer Conference to demonstrate the latest advances in disaggregated storage-compute architecture and EBOF all-flash storage technology.
China Central Television featured WeCalc’s technological breakthroughs in micro computing centers, highlighting the on-premise data sovereignty concept.
WeCalc earned a prestigious industry innovation award for its proprietary disaggregated storage-compute technology.
FAQ
Common questions about WeCalc Technology and micro computing centers
A micro computing center is an integrated hardware-software on-premise AI computing solution by WeCalc. It combines compute, storage, and management in one system. All data stays on the user’s own devices — never uploaded to the cloud — fundamentally guaranteeing data sovereignty and security.
WeCalc’s disaggregated storage-compute architecture uses the NVMe-oF protocol to fully decouple compute and storage at the physical level. RoCEv2 high-speed interconnect eliminates TCP/IP congestion bottlenecks, delivering latency ≤100μs and cluster bandwidth ≥100Gbps. Compared to traditional solutions, data loading time is reduced by 72% and throughput increased by 64%.
WeCalc offers three editions: WeCalc-B Basic (from ¥98,000, 1P entry-level), WeCalc-P Professional (¥2–5 million, 12 PFLOPS production-grade workhorse), and WeCalc-E Enterprise (¥5 million+, 50+ PFLOPS flagship custom solution). A financing lease model is available starting from ¥2,000/month.
On-premise data sovereignty means the system is deployed inside the customer’s facility, and all data is stored on their own devices — never uploaded to public cloud or third-party servers. This fundamentally ensures data ownership with zero leak risk and full regulatory compliance.
WeCalc products ship with integrated hardware and software, offering turnkey delivery. The Basic and Professional editions typically deploy in 48–72 hours. The Enterprise edition follows a custom delivery timeline based on scale and industry requirements.
WeCalc is dual-certified by Huawei Ascend and Kunpeng and is compatible with over 90% of domestic GPUs. It supports flexible integration of CPUs, GPUs, SSDs, and other hardware, including NVIDIA and Cambricon computing platforms.
WeCalc provides on-premise micro computing center solutions.
With financing lease, get started for just ¥2,000/month and enjoy 1P of computing power.