White Paper

WHITE PAPER Bringing commercial technologies to defense in the age of AI mrcy.com 2 mrcy.com 2 PURPOSE-BUILT HARDWARE The proliferation of sensors, big data, increased computing resources and advanced algorithms have ushered in an era of increasingly power ful AI systems. These systems and their complex deep-learning neural networks continuously drive computational demands on hardware platforms. Optimizing hardware at the hear t of these processing-intensive AI systems to receive actionable information in real time necessitates assimilation of the latest computing components including motherboards, memor y, central processing units (CPU) processors, graphic processing units (GPUs), field programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs) and integrated systems in a package (SiP). Advances in AI will continue to be dependent on hardware optimized for AI algorithms, including novel designs for neuromorphic computing, hardware accelerators for machine learning, embedded systems, and parallel architecture research motivated by AI processing requirements. — whitehouse.gov/ai Equally impor tant is the need to process information as close to the data source, or "edge," as possible, with field-ready hardware that is not only agile but also reliable, affordable, upgradeable and easy to integrate. Close collaboration, strategy and roadmap alignment between technology suppliers focused on delivering edge solutions and industr y leaders, such as Intel Corporation, are pivotal in deploying relevant AI-enabling technologies to the field. Additionally, and specifically to meet the heightened technical, security and packaging demands of the defense industr y, hardware providers must invest time and resources to adopt appropriate design practices and business models to bring breakthrough AI technologies to the tactical edge. KEEPING PACE WITH THE AI EVOLUTION AI, deep-learning and neural networks rely on real-time communication and processing of data, but often this is impeded by inefficient hardware and technology. To rapidly launch, incorporate and sustain evolving AI workloads, system architects seek to deploy the most capable hardware based on technology built for the commercial sector. However, this technology is not inherently built for edge environments. Engineering rugged systems is a significant design and test effort, which can delay hardware deployment and result in proprietary, custom-built hardware that is difficult and expensive to upgrade or scale. To cost-effectively keep pace closely with technology leaders to develop commercial off-the-shelf-based products with an open, modular design approach. Poised to reshape the global order, artificial intelligence (AI) is rapidly transforming the defense industry by broadening the scope of machine applications and leveling the playing field for nation states looking to gain power on the global stage. The future soldier will likely operate on an entirely new battlefield, one where AI serves as a force-multiplier and where possessing superior AI technologies enhances human capabilities for the ultimate military advantage. Intel® 2nd Gen XEON® Scalable Processor Modular systems, such as OpenVPX subsystems and Mercur y 's flagship RES HD blade server platform, simplify integration and eliminates the need to "rip and replace" entire systems during technology upgrades.

• Cover