Pathways and Strategies for Migrating EDA Design Environments to the Cloud

In the early stage of cloud adoption, chip‑design companies fall into two groups: large enterprises seeking elastic compute and agile operations, and start‑ups looking for a complete, standard design environment to accelerate development. 2022 is expected to be the first wave of domestic chip‑design cloud migration.

EDA Cloud Migration Pathway – The migration follows a sustainable process of research & planning → construction & execution → operation & optimization . The planning phase must evaluate cloud technologies and public‑cloud commercial terms.

Five decisive factors for an EDA cloud solution :

High elasticity – scale compute resources on demand to meet short‑burst simulation peaks.

Low data sensitivity – prioritize less‑sensitive design data (e.g., RTL, IP) for cloud placement.

Low interaction – minimize data transfer between on‑premise and cloud to reduce bandwidth costs.

Easy collaboration – centralized cloud platforms simplify multi‑site team coordination.

User habits – maintain familiar workflows (e.g., LSF job submission) to ensure a seamless transition.

Based on these factors, two mainstream cloud models emerge:

Hybrid‑cloud solution – optimal for medium‑to‑large design companies that already own on‑premise data centers but need elastic compute during peak periods.

Full‑cloud solution – ideal for start‑ups with limited capital, distributed teams, and a need for rapid environment provisioning.

Scenario analysis demonstrates how a typical 14 nm digital IC design project can be evaluated against the five factors, with detailed workload curves for compute and storage.

Case Study 1 – STA on the cloud – Static timing analysis (STA) offers high elasticity, low sensitivity, low interaction, easy collaboration, and familiar LSF‑based job submission, making it a safe first‑step for cloud migration.

Case Study 2 – Full‑cloud for start‑ups – A start‑up can obtain a complete three‑tier secure network, elastic compute clusters, and storage with minimal upfront investment, enabling a design environment to be ready within a week and costing only a few hundred thousand yuan per year.

The article concludes that successful EDA cloud migration requires thorough research, detailed planning, and a customized path that aligns with the company’s size, workload characteristics, and user habits.