As with all tradeshow events this year, ISC High Performance 2020 took on a very different look and feel. Overall, the event received high praise for adapting to a virtual environment, and the news coinciding with the event continued to generate headlines. One of the more interesting announcements was from NVIDIA and Mercedes Benz launching software-defined, intelligent vehicles using end-to-end NVIDIA technology.
High performance computing has had a long history in the design and development of automobiles. Verne Global has been assisting advanced manufacturers like BMW and VW with their HPC powered computer-aided design, modeling and simulation since 2012. The convergence of HPC, artificial intelligence (AI) and the automotive industry is well underway as HPC enables faster data processing for simulate vehicle crash tests, wind tunnel tests, etc.
That same data processing capability also comes into play for autonomous vehicles as neural networks are used to train object recognition and avoidance in the vehicle response systems. The multiple terabytes of data are gathered from autonomous vehicles every day. Neural networks amalgamate these large quantities of data and produce a single result or action that can be carried forward by the vehicle all through machine learning.
According to NVIDIA, this new announcement with Mercedes Benz will see cars equipped with multiple processing engines for high-performance, energy efficient compute and AI. “Starting in 2024, every next-generation Mercedes-Benz vehicle will include this first-of-its-kind software-defined computing architecture that includes the most powerful computer, system software and applications for consumers, marking the turning point of traditional vehicles becoming high-performance, updateable computing devices.”
Industry analysts see this announcement as having a two-fold impact on the automotive industry. One, it establishes an ecosystem, per se, of companies that excel in their individual fields coming together to compete against Tesla and the disruptive impact it's had on the automotive industry. According to Rob Enderle, president and principal analyst of the Enderle Group, “So if you take the leader of car intelligence, one that understands concepts like PCs as a service (PaaS), and Software as a Service (SaaS), you have the potential to redefine the car market with a more engaging solution than Tesla now offers -- and you have the potential increased economies of scale provided by the combined Mercedes/NVIDIA partnership.” Enderle sees this evolving into a Car as a Service (CaaS) model.
The other impact is on the car itself. The fact the announcement referred to “updateable computing devices” means that this is not a typical car feature. Patrick Moorhead of Moor Insights, noted “that putting autonomous hardware into a vehicle is like making a smartphone on wheels.” However, smartphones and cars are built to have very different lifespans. A smartphone as a device is outdated within a few years. Cars can be purchased with a decade of ownership in mind. This could have significant implications for the automotive business model - will cars hold more value with a continuously updated computer on board? Only time will tell.
So what does this mean for the data center infrastructure required to process this likely exponential growth in data coming in from these new vehicles? As Rich Miller from Data Center Frontier pointed out last year, we are already getting a glimpse of the data center impact of connected cars via ride-sharing apps. Both Lyft and Uber were planning big IT infrastructure spends prior to the pandemic in order to manage the infrastructure volume of their businesses. Uber alone was exploring a 5 megawatt data center deployment with up to 15,000 square feet of data center capacity and a “first right of refusal” on another 5MW.
In the race to develop and scale AI and machine learning applications, car manufacturers industry-wide shouldn't be limited by data center capabilities, or the underlying energy and utility infrastructure needed to support advanced supercomputing platforms. Iceland is a prime location to efficiently manage these types of applications. Verne Global’s industrial scale campus is designed and engineered to cater specifically for high performance computing and is powered by one of the world's most reliable, scalable and lowest cost energy profiles. Autonomous vehicles may change the automotive industry, but with careful planning and choosing the right partners, automobile manufacturers can start preparing now their data center infrastructure now.