Skip to main content
Customer story

Continental speeds AI learning for safer vehicles

With help from Equinix, auto parts manufacturer improves AI training time by 70%

Continental Speeds AI Learning for Safer Vehicles | Equinix customer story

Continental Speeds AI Learning for Safer Vehicles | Equinix  logo

70%

faster AI training

14x

more deep learnings per month

100%

clean and renewable energy coverage

Faster AI learning for safer connected vehicles

Making vehicles safer saves lives. That’s why auto parts manufacturer Continental set out to accelerate its AI learning. Interconnecting NVIDIA AI and IBM storage clusters using Equinix AI-ready data centers now speeds Continental’s AI data access and time to market for connected and autonomous vehicles.

By placing our IBM storage and NVIDIA GPU cluster in an Equinix ‘AI-ready’ data center in just two weeks, we had the infrastructure and interconnection we needed to increase the number of AI experiments by 14x, speeding our time to market.

Robert Thiel Principal Architect and Guild Master Computer Vision & Artificial Intelligence, Continental AG, Business Area Autonomous Mobility

The challenge: Processing distributed data at massive scale

Continental’s Advanced Driver Assistance Systems (ADAS) team needed to process more than 150 terabytes (TB) of data to inform design decisions that would increase the safety of connected and autonomous vehicles and, ultimately, save lives. 

Doing this required the team to pull image and sensor data from vehicles in different geographies, including Europe, America and Asia, and store and process it in a connected central repository that could be quickly accessed by hundreds of engineers anywhere in the world.

Taking neural networks to the next level

Continental leveraged its NVIDIA DGX and IBM Elastic Storage System (ESS) 3000 to accelerate the development and training of neural networks. The data is collected in different locations around the world, then processed, labeled, selected and made ready to be used for AI trainings and uploaded to the IBM ESS 3000, which (like part of the raw data) is in an Equinix data center in Frankfurt. 

Once the data was placed in the Equinix data centers in Frankfurt, Germany, and uploaded to the Continental cloud, thousands of Continental engineers around the world worked collaboratively to tag, structure, analyze and process huge amounts of data and develop new AI learning models on demand. This process must be extremely efficient; otherwise, car manufacturers can’t begin vehicle production due to software delays.

The Continental AI supercomputer cluster is based on NVIDIA DGX enterprise-class servers that consist of high-performance GPUs. In addition, the IBM ESS 3000, which is highly compatible with NVIDIA DGX servers, is an enterprise-grade parallel file system that provides Continental with superior data resiliency, scalability and control. Systems integrator SVA implemented the overall integration of the IBM storage solution in the cluster. Equinix Smart Hands® remote management installed the hard disks that captured the data from the test drives.

For this type of project, you need an end-to-end balanced system and a very flexible environment. Equinix provided the environment, infrastructure and expertise from its architects and engineers we required to build the Continental solution in a very professional way.

Jens Hoschke Sales Manager Storage Solutions, Automotive and Industrial Accounts, IBM Deutschland GmbH

Rapidly creating new value

Equinix IBX data centers and personnel provided the infrastructure and expertise to deploy Continental’s supercomputer cluster infrastructure in just a couple of weeks. Equinix Global Solutions Architect (GSA) expertise helped Continental design and deploy its supercomputer cluster and associated IBM storage on Equinix, including optimizing its cabling paths for lower latency and increased performance.

Continental leveraged Equinix Cross Connect to interconnect its cluster cages to extend already existing LANs, and Fiber Connect to connect the infrastructure between Equinix and Continental data centers. As connectivity was a top priority, housing the cluster at the Equinix FR6 Frankfurt data center was appealing because it was next door to Continental’s existing ADAS infrastructure. Continental’s corporate infrastructure division leveraged Equinix regional hubs by colocating in five facilities in EMEA, Asia and the U.S. to securely deploy and effectively scale its infrastructure to meet the demands of its connected and autonomous vehicle testing environment. This novel approach resulted in a much faster innovation through accelerated AI training. “With the performance, flexibility and scalability of the new GPU cluster, Continental improved AI training time by 70% using IBM Spectrum® Scale and NVIDIA DGX systems” hosted in Equinix data centers, says Balázs Lóránd, PhD, Head of AI Development Centre at Budapest, Continental AG, Business Area Autonomous Mobility.

The solution implemented Equinix Internet Access and Equinix Fabric® software-defined interconnection for on-demand, global connectivity to internet service providers (ISPs), IT infrastructure, and network and cloud providers.

Sustainability was also an important criterion for Continental when selecting Equinix as its partner. By 2050, Continental aims for its entire value chain, including suppliers, to be carbon neutral. As the world’s digital infrastructure company®, Equinix has the responsibility to harness the power of technology to drive sustainable growth and transformative social impact. “Having 100% certified green energy was a priority for us, especially for this type of high-performance, supercomputer/AI environment,” explains David Enenkel, Head of IT Operations, Continental/Deputy Local IT Manager, Continental AG, Business Area Autonomous Mobility. “Following Continental’s sustainability framework, the project team was very happy to see that Equinix could provide us with the necessary resources to run AI trainings much faster with greater optimization and sustainably.”

Benefits

Faster deployment time

All the infrastructure components and connected ecosystems of business and technology partners, such as OEMs and companies that provide innovative services, were already on Equinix, enabling the entire Continental deployment to be completed in just two weeks.

Improved AI training time

AI model training that used to take weeks now takes only days—a 70% reduction. Continental can run 14 times more deep learning trainings per month, in less time per training.

Room to scale

Continental’s supercomputer/storage cluster provides enormous compute capacity with several hundred GPUs and more than 150 TB of distributed storage.

Globally interconnected portfolio

With Equinix, Continental leveraged high-bandwidth, low-latency interconnection for its supercomputer cluster infrastructure, connectivity to its backbone network, and easy access to cloud providers and smart mobility ecosystem partners.

Infrastructure committed to sustainability

Continental understands that being a sustainable and responsible business increases its ability to innovate and meet the future—adding value to the company. This is why sustainability is firmly integrated in its corporate strategy and is a key driver of innovation. Equinix can provide Continental with a global footprint to build its digital infrastructure more sustainably and support its goal of a carbon-neutral supply chain.

Build what comes next on Equinix

With 281 data centers across 77 markets in 36 countries, Equinix puts your infrastructure next to the clouds, networks and partners you depend on.