This blog was co-authored by Michael Johnson, Cloud Solution Architect, EDA and Golan Abraham, Product Manager.
Semiconductor innovation is the invisible engine powering today's AI revolution, Internet of Things (IoT) expansion, and automotive transformation. Yet as chips become faster, smaller, and smarter, the electronic design automation (EDA) industry faces an unprecedented complexity explosion.
Each new generation of transistor technology pushes from larger to smaller nanometer processes. That puts more strain on the limited number of semiconductor engineers available and demands significantly more compute resources and storage capacity per project, raising costs. Meanwhile, consumer demand for cutting-edge devices compresses time-to-market (TTM) windows in a race between EDA companies.
Leading semiconductor companies have necessarily turned to the cloud to tackle these mounting pressures. The ability to burst workloads to virtually unlimited compute capacity has broken through the constraints of fixed on-premises infrastructure, with hybrid architectures enabling teams to maximize existing on-premises resources while tapping into the cloud when needed for intensive workloads.
With compute elasticity established, what's the next breakthrough for EDA teams to tackle the steep non-recurring engineering (NRE) costs that still challenge project profitability?
This article explores how Amazon FSx for NetApp ONTAP (FSx for ONTAP) transforms and accelerates EDA workflows, unlocking developer productivity and improving infrastructure utilization. We’ll also see some real-life customer success stories, where EDA companies have reduced job completion time by 50% and reduced storage costs by 35% with the help of FSx for ONTAP.
Find out how as we cover:
- The challenges driving EDA complexity and costs
- The operational challenges slow down engineer productivity
- Data barrier challenges for IT and storage admins
- FSx for ONTAP: The AWS-native storage foundation for EDA on AWS
- EDA customer success stories with FSx for ONTAP
- Arm Limited uses FSx for ONTAP to burst to AWS
- Annapurna Labs modernizes EDA with FSx for ONTAP
- Welcome the next generation of chip innovation with FSx for ONTAP
The challenges driving EDA complexity and costs
Here's a number that should stop every EDA team in their tracks: 70% of your project's NRE costs come from developer time and specialized EDA tools (20%). IT infrastructure only contributes about 10% of NRE. So, the bulk of NRE is not the million-dollar compute clusters (6%), not the petabytes of storage (4%)—instead it’s the hours engineers spend waiting, rebuilding, and working around infrastructure friction. With costs growing with every nanometer advance and the shortage of talent, optimizing developer and EDA tool resource productivity has a significant impact on lowering NRE and driving design closure.
Here you can see the empirical breakdown of the exponential NRE costs for EDA as process nodes shrink from 16nm to 3nm:
The operational challenges slow down engineer productivity
Why does developer time consume such an outsized portion of EDA budgets? The answer lies in the characteristics of chip design workflows that slow down the design and development engineers’ productivity.
- Data center capacity requirements keep growing. The transition from 5nm to 3nm processes drove a 4–6x increase in compute requirements, placing corresponding strain on storage and I/O performance. With another 4x increase expected moving to 2nm, combined with AI and GPU workloads competing for the same data center capacity, even well-funded companies face mounting constraints that keep engineers waiting instead of designing.
- Workflows span diverse and often shifting phases. A single design iteration might trigger synthesis, place-and-route, timing analysis, or verification. Managing these interdependencies consumes substantial developer time and leads to recurring compute bottlenecks from missing project resources alignment.
- Multiple projects compete for limited datacenter resources. Teams juggle multiple concurrent projects while critical requirements or competitor announcements instantly spawn urgent parallel tracks–requiring immediate access to complete design environments.
- EDA development is inherently iterative. Each design iteration requires checking out large datasets and rebuilding environments, which are processes that can consume hours per cycle. When checkout and build processes take hours instead of minutes, engineers either wait (a productivity loss) or make compromises (a quality risk).
These delays have an obvious drag on delivery. The faster results come back and the less time it takes to identify and fix bugs, the faster engineers can experiment and innovate, producing more design optimization iterations.
With companies turning to AWS to eliminate compute constraints, the leading cause of infrastructure inefficiency is now data friction—making data infrastructure optimization the most direct path to NRE costs reduction.
Data barrier challenges for IT and storage admins
EDA's unique characteristics create data bottlenecks that turn IT and storage administrators’ routine tasks into significant productivity drains, directly threatening TTM goals and increasing costs, as follows:
- Scale complicates data strategy.
EDA workloads generate massive volumes of data due to the accumulation of tool-generated artifacts, simulation outputs, and years of design libraries across tens of specialized EDA tools. Faced with dynamic workflows, teams often preemptively sync petabytes of datasets—consuming substantial time and budget moving data they might never access.
- Extremely demanding performance requirements for storage.
As the nanometer precision increases, EDA designs become increasingly compute-intensive, requiring high-performance storage with intense file I/O and sub-millisecond latency to support projects running in parallel.
Meeting such demands requires a storage layer with a sustained high input/output per second (IOPS) and throughput characteristics. If the storage layer isn’t up to that, there will be performance gaps that can delay design iterations and impact turnaround time.
- Operational processes introduce development friction.
Environment creation is a bottleneck in EDA processes because it requires copying very large datasets, often taking hours and sometimes days. Given the iterative nature of design workflows, this friction doesn't just slow individual tasks—it influences how engineers approach problems, sometimes forcing compromises to avoid provisioning delays.
- Global collaboration faces synchronization challenges.
The multiple stages of EDA workflows are often handled by specialized teams across geographies, making real-time data collaboration essential. However, traditional replication approaches create version conflicts and access delays that undermine collaboration effectiveness. Teams often work with outdated information or wait for synchronization cycles that interrupt critical feedback loops.
- Cost pressures accumulate across the workflow.
Repeated data duplication, high-performance infrastructure requirements, and persistent storage of large datasets—often for temporary or redundant processes—combine to significantly increase both cloud and on-premises spending.
- Data protection and security requirements add infrastructure constraints.
EDA workflows demand enterprise-grade data protection features, including snapshots, back-ups, disaster recovery, and security controls. Integrating these features into cloud and hybrid storage layers is complex, increasing risk and management burden.
- Hybrid management and data migration introduce architectural challenges.
Hybrid environment management and data migration pose significant data mobility and synchronization challenges, as EDA teams must effectively handle large-scale data across on-premises and cloud infrastructures all while maintaining rapid provisioning, high throughput, and seamless global collaboration.
The question for EDA architecture teams becomes clear: How do you build a data infrastructure that’s optimized for EDA workflows and reduces TTM?
You do that with FSx for ONTAP.
FSx for ONTAP: The AWS-native storage foundation for EDA on AWS
FSx for ONTAP is a fully managed AWS storage service based on NetApp® ONTAP® enterprise capabilities. FSx for ONTAP delivers a high-throughput, hybrid, and elastic infrastructure that naturally fits chip design workflow characteristics.
FSx for ONTAP does that with its five main feature pillars:
FSx for ONTAP inherently addresses each data friction point through technologies that understand EDA's unique demands enabling engineers to build, test, and collaborate faster and more efficiently across regions and in hybrid architectures.
- Solving scale through intelligent data movement and scale-out architecture
NetApp FlexCache® technology directly tackles the “sync everything or sync nothing” dilemma behind cloud-bursting strategies by creating intelligent, writeable caches that only pull accessed data from on-premises origins for use via FSx for ONTAP. In a typical electronic design flow, where just 10% of the data sets are accessed during active projects, caching with FlexCache reduces data transfer volumes by up to 90%.
The scale-out Gen-2 FSx for ONTAP architecture supports up to 12 high-availability (HA) pairs per Availability Zone (AZ), enabling incremental capacity and throughput scaling up to 1 PiB in a single namespace as projects grow, as showcased in the diagram below.
FSx for ONTAP gives customers true elasticity beyond traditional scale-out:
- Volumes automatically grow to meet rising demand, and SSD capacity can be decreased after project completion—eliminating fixed infrastructure costs from unused storage.
- FSx for ONTAP manages your volumes via the FlexVol® and FlexGroup technologies: FlexVol volumes are standard, dynamic volumes; FlexGroup volumes are scale-out volumes that distribute data across multiple nodes for massive capacity and performance. With these technologies, FSx for ONTAP unlocks access to massive namespaces reaching up to 1 PB in size, with billions of files—eliminating the complex mount-point management that traditionally fragments large EDA datasets across multiple volumes.
Finally, FSx for ONTAP includes automatic data tiering that moves cold data to a capacity tier, freeing up SSD capacity without manual intervention and keeping high-performance space available for active workloads. In a typical flow where 60–80% of EDA data is cold, data tiering gives customers out-of-the-box scalability for all projects by freeing up SSD storage capacity.
- Delivering high-performance storage that naturally fits EDA workloads
Performance specifications directly target EDA's file-intensive requirements: Up to 2.4 million IOPS with sub-millisecond latency supporting verification jobs that query thousands of files per second. Throughput scales to 72 GBps across scale-out configurations, so multi-terabyte builds don't become infrastructure bottlenecks.
FSx for ONTAP delivers this performance across both the single-AZ and multi-AZ HA configurations, ensuring consistent performance regardless of your deployment model.
Also, FlexCache technology unlocks local-like performance at a global scale by bringing frequently accessed data closer to users and compute resources.
- Unlocking operational agility that reduces engineering waiting time and speeds up TTM
NetApp FlexClone® technology creates thin, writable clone data copies of full environments for use in development, testing, simulation, analysis, etc. These copies are spun up in seconds rather than hours, regardless of the environment size.
This instant cloning capability increases the speed of the continuous integration and continuous delivery (CI/CD) pipeline for EDA workflows by directly reducing developer wait times for environment copies. It also indirectly reduces costs by maximizing developer productivity and accelerates TTM by eliminating the infrastructure friction that otherwise slows DevTest cycles.
- Enabling global collaboration without synchronization friction
FlexCache gives distributed EDA teams a way to collaborate on shared datasets with local-like performance. FlexCache creates intelligent cached copies of data volumes in remote locations, retrieving only the data blocks that teams actually access rather than replicating entire datasets.
This eliminates the synchronization delays and version conflicts that traditionally interrupt critical design feedback loops, allowing global design teams to work collaboratively on the same projects without waiting for complete dataset transfers.
- Optimizing storage costs while accelerating engineering productivity
Data written into FSx for ONTAP undergoes inline footprint reduction through the NetApp storage efficiency features, including thin provisioning, data compression, deduplication, and compaction. These efficient technologies directly translate to engineering productivity gains by reducing the time developers spend waiting for data operations and environment provisioning.
Combined with data tiering, data cloning, elasticity, the ability to decrease SSD capacity on demand, and pay-as-you-consume models, these efficiencies directly minimize fixed storage costs, aligning infrastructure expenses with actual project phases rather than peak capacity demands. The result is both faster development cycles and optimized cost structure that scales with project demands.
- Safeguarding EDA workloads with enterprise-grade data protection
FSx for ONTAP provides enterprise-grade data protection through cross-AWS region replication with point-in-time recovery capabilities.
NetApp Snapshot™ technology creates instant, space-efficient, point-in-time copies without performance impact. These copies are created on a configurable schedule and used in case of a local error or failure. Customers can select a Snapshot copy to quickly and easily recover lost data.
NetApp SnapMirror® technology replicates data changes incrementally and continuously to a secondary FSx for ONTAP system. SnapMirror consumes space and bandwidth efficiently and provides the basis for reliable, cross-Region backup and disaster recovery copies. FSx for ONTAP disaster recovery capabilities automatically handle failover and failback to secondary systems, maintaining business continuity.
Security features, including isolated data per tenant, Active Directory authentication, immutable storage, malware protection, and ransomware protection, safeguard EDA intellectual property.
- Streamlining cloud adoption without workload disruption
For existing NetApp ONTAP customers, migrating EDA workloads to FSx for ONTAP requires minimal disruption as you can leverage proven ONTAP data replication capabilities like SnapMirror alongside AWS DataSync to move workloads gradually, allowing teams to maintain productivity during the transition.
This approach preserves existing ONTAP management skills while enabling hybrid operation during migration, ensuring consistent performance and familiar workflows throughout the cloud adoption process.
EDA customer success stories with FSx for ONTAP
Leading semiconductor companies are already achieving measurable gains with FSx for ONTAP on AWS—turning data infrastructure into a productivity engine.
Arm Limited uses FSx for ONTAP to burst to AWS
Arm Limited leveraged the hybrid capabilities of FSx for ONTAP to create an intelligent cloud bursting solution to support peak demand by using FlexCache as an in-cloud cache for their on-premises NetApp systems.
The results speak directly to EDA productivity challenges: There was a 50% reduction in job completion times with over 10 million jobs successfully processed daily, supported by FSx for ONTAP’s scale-out configuration and a data footprint in the tens of petabytes. The hybrid approach accelerates Arm Limited’s chip-design process and improves speed to market without requiring changes to application code or data management practices.
Annapurna Labs modernizes EDA with FSx for ONTAP
Annapurna Labs, Amazon's chip design division, chose FSx for ONTAP to modernize its EDA infrastructure completely. This migration delivered a 35% storage cost reduction, maintained 100% availability for business-critical EDA workloads, and doubled storage capacity per volume while maintaining workload performance. This infrastructure foundation helped the company eliminate downtime and accelerate design for their latest machine learning chips—directly connecting storage efficiency to innovation velocity.
When high-performance, cloud-native storage aligns with EDA workflows, the results are clear: faster design cycles, lower infrastructure costs, and scalable collaboration. FSx for ONTAP gives semiconductor teams the cloud storage speed and scale to meet today’s toughest EDA demands.
Welcome the next generation of chip innovation with FSx for ONTAP
For efficient engineering, EDA teams fundamentally need the familiarity and performance of their on-premises experience with the elasticity and collaboration capabilities that only AWS and NetApp infrastructure can provide.
With FSx for ONTAP, you get the elastic and performant infrastructure that accelerates EDA innovation.
Ready to learn more? Visit the FSx for ONTAP homepage or request a meeting with a specialist.
Find out more about NetApp Workload Factory, the free service that helps you plan, optimize, and automate FSx for ONTAP operations to meet EDA storage needs.
