ALCHIP Technologies scales cloud infrastructure to support advanced AI chip design

Semiconductor design teams rely on compute-intensive workloads to implement increasingly complex chip architectures. As a leading dedicated AI and HPC ASIC company, ALCHIP supports advanced design programs that require high levels of performance and reliability for AI chip development.

During the final design phase, chip design workloads can require several months of continuous computation as teams approach key program milestones. For ALCHIP, consistently supporting these workloads is essential to meeting customer delivery timelines across the complex semiconductor supply chain.

Fixed on-premises infrastructure needed greater flexibility when program demands peaked. The ability to rapidly provision additional high-performance compute and storage capacity, without disrupting active workloads or requiring large upfront capital investment, became a strategic priority as ALCHIP’s design programs grew in scale and complexity. The company sought an approach that could extend its existing environment for long-running chip design workloads during key delivery milestones.

To support growing infrastructure requirements during peak design phases, ALCHIP expanded its AWS Hybrid Cloud architecture, supplementing its on-premises environment with cloud-based capacity during periods of increased demand. This approach allows ALCHIP to scale resources as needed while continuing to run active chip design workloads within a familiar operational framework.

The AWS Asia Pacific (Taipei) Region plays a key role in this architecture by providing low-latency connectivity between ALCHIP’s on-premises environment and AWS compute resources. This connectivity allows ALCHIP to access on-premises storage systems from cloud-based compute environments, reducing the need for large-scale data duplication and minimizing delays associated with transferring large design datasets.

For compute-intensive chip design workloads, ALCHIP uses Amazon Elastic Compute Cloud (Amazon EC2), including high-memory Amazon EC2 x2iedn instance types optimized for memory-intensive simulations and design verification tasks. These instances allow ALCHIP to scale compute capacity during the final design phase, supporting workloads that require extended execution times.

The hybrid architecture also supports NetApp SnapMirror, enabling synchronized data replication between on-premises storage and AWS environments. This ensures data consistency across environments and allows engineers to shift workloads as needed while protecting design integrity throughout long-running execution cycles.

“When design workloads run for months at a time, interruptions can have significant cost and schedule implications. AWS gives us the confidence to scale compute capacity as needed while supporting long-running jobs during our critical design phases,” says Johnny Shen, chairman, CEO, and president at ALCHIP Technologies.

By expanding its hybrid cloud architecture on AWS, ALCHIP is establishing a scalable foundation for increasingly complex chip design requirements. The company can increase compute and storage resources by more than 100x on demand, aligning infrastructure usage with extended execution cycles and key delivery milestones.

The shift from a capital-intensive infrastructure model to a more flexible operational framework enables ALCHIP to manage costs more efficiently across the program lifecycle. Rather than maintaining fixed capacity sized for peak demand, ALCHIP can now scale resources dynamically as design requirements evolve, optimizing total cost of ownership while ensuring compute capacity is available on demand throughout critical design phases.

With scalable compute and storage available on demand, ALCHIP plans to shorten development timelines over time while reducing schedule risk for advanced AI chip design. The initiative also supports ALCHIP’s ongoing collaboration with AWS on AI processor development and reinforces its role within the global semiconductor supply chain.