Following the success of the 4th AI Hardware Summit that took place 13-16th September in Mountain View, US, Kisaco Research is proud to announce the annual 2022 Executive Outlook. The Executive Outlook provides leading industry executives, and key decision makers, an exclusive chance to network, discuss and share their views and predictions on the market from a strategic viewpoint, and, collectively define a vision for the future of AI systems. Our Executive Outlook is an annual, end-of-year executive-level summit and awards ceremony. It will recognize both engineering and business achievements during the year and gather the very best and brightest in the AI acceleration world, to pause and reflect on the year and set a vision for the following 12 months. The forum will consist of analyst overviews and predictions, executive keynotes, and an awards ceremony.

Getting the software right is important when developing machine learning models, such as recommendation or classification systems. But at eBay, optimizing the software to run on a particular piece of hardware using distillation and quantization techniques was absolutely essential to ensure scalability. "[I]n order to build a truly global marketplace that is driven by state of the art and powerful and scalable AI services," Kopru said, "you have to do a lot of optimizations after model training, and specifically for the target hardware." With 1.5 billion active listings from more than 19 million active sellers trying to reach 159 million active buyers, the ecommerce giant has a global reach that is matched by only a handful of firms. Machine learning and other AI techniques, such as natural language processing (NLP), play big roles in scaling eBay's operations to reach its massive audience. For instance, automatically generated descriptions of product listings is crucial for displaying information on the small screens of smart phones, Kopru said.

MOUNTAIN VIEW, Calif., Aug. 8, 2021 -- As machine learning models continue to grow in size and complexity, and more and more models enter production in enterprises worldwide, the acceleration of these workloads is becoming more complex and compute-intensive. At the front end, data-centricity is taking precedence over model-centricity, while at the back end, AI practitioners increasingly want systems that are performant and efficient, but also sustainable, explainable and accountable. From massive research models like GPT-3, to day-to-day models deployed by enterprises around the world, the AI Hardware Summit will bring together people focused on making AI fast, efficient, and affordable. In previous years the AI Hardware Summit has been the site for announcements and product reveals from the likes of Habana Labs, Graphcore, Intel, Qualcomm and many others. As the nexus of the AI acceleration community, the summit has also helped ground-breaking hardware reach people who are using AI to change the world.

With the popularity of AI growing over the past few years, so too have the number of conferences either dedicated to machine learning and artificial intelligence or at least discussing the topics as a major portion of the agenda. However, the inaugural AI Hardware Summit stood out from the crowd last year. At the conference, not only did attendees hear from the major technology players, some start-ups like Habana emerged from stealth mode with innovative solutions. The second annual AI Hardware Summit presented by Kisaco Research promises more of the same. This year's conference promises to offer a wide variety of solutions for AI, not just at the chip level but also some of the supporting products and services for AI solutions.

This author account is for Partner Content with Emerj advertising and content partnership clients. In 2018, James Kobielus wrote an article on the AI market's shift to workload-optimized hardware platforms, in which he proposed: Workload-optimized hardware/software platforms will find a clear niche for on-premises deployment in enterprises' AI development shops. Before long, no enterprise data lake will be complete without pre-optimized platforms for one or more of the core AI workloads: data ingest and preparation, data modeling and training, and data deployment and operationalization. We are seeing Kobielus' words come true. In the past year, nearly 100 companies have announced some sort of AI-optimized IP, chip, or system optimized, primarily for inferencing workloads but also for training.