Data has become both the most abundant and the most valuable resource in today’s business world. According to a recent IDC study, the world generated or replicated an estimated 64.2 zettabytes of data in 2020, with a forecasted compound annual growth rate of 23% through 2025. Data comes in all shapes and sizes, from social media posts to readings from sensors on industrial machinery. The deluge of data available to decisionmakers can be noise that drowns out important information, or an invaluable source of insight. Artificial intelligence and machine learning are becoming increasingly important for businesses to distinguish between the two.
Digital advancements like the rise of 5G and the process of digitalization have also been accelerating in a post-pandemic world. These advancements all require high performance computing (HPC) to advance to the level that businesses need to be competitive in today’s marketplace. They also need hardware capable of enabling the HPC applications that increase a business’s agility and keep it competitive.
Unlocking A.I.’s potential
The trend of HPC becoming a necessity is particularly true for enterprises (and even traditional businesses) that are using big data, machine learning and A.I., or those who plan to. A recent IDC study predicted that 75% of enterprise applications will use A.I. by the end of 2021. For example, Alaska Airlines has turned to A.I. to plan flight routes, as a way to save fuel and time, two of the most valuable commodities in the airline business. No doubt Alaska will not be the only airline to implement this system, and given the amount of data processed, HPC will be a cornerstone of this.
Governments are also taking notice of HPC’s possibilities, with the U.S. Department of Energy funding 13 HPC projects this summer alone to fuel energy innovation. Healthcare is another key area where HPC will have a positive impact. Pharmaceutical company AstraZeneca worked with A.I. chip designer Cerebras to train HPC-powered A.I. to quickly read hundreds of thousands of biomedical research papers, streamlining research and cutting down the time needed to develop new drugs.
Closer to home, semiconductor production facilities like our fabs at TSMC produce hundreds of thousands of wafers per month, and in the process, they also produce a mountain of data. Our tools precisely track thousands of parameters that are needed to mass produce circuitry reliably at a near-atomic size. Our sensors wring out precious data that can be processed using HPC applications to monitor and optimize for quality, efficiency, and flexibility. In other areas of manufacturing, Procter & Gamble will partner with Sandia National Laboratories to create an ecosystem of HPC-enabled fiber manufacturing models to make solvent-free detergents.
Digitalization like never before
Personal mobile devices and enterprise data centers process increasingly huge volumes of information every single minute, accelerated by the remote nature of collaboration in a pandemic world. Online remote learning is now a way of life at schools; in-person stores have become digital storefronts to survive (or in the case of e-commerce businesses, began doing even more business); work-from-home models and hybrid work models are being embraced; and entertainment at home helps people relax.
From edge to cloud, game consoles to data center servers, HPC solutions are in high demand to provide more computing power and more efficient networking infrastructure. As a result, HPC applications have become one of the key drivers of semiconductor technology and require leading-edge technologies to deliver competitive performance with higher computing power and lower energy consumption.
Expanding computing borders with 5G
By 2025, IDC predicts there will be 55.7 billion connected devices worldwide, with 75% of these devices on an IOT network. Adding to this huge amount of increasing community between devices is the fact that 5G network speed will be 10 times faster than the average mobile connection. Not only is more data being generated, there’s a greater need to process it all at speeds that make it possible to react to this data in real time.
Consider an autonomous vehicle: The system needs to collect and understand data from sensors and GPS and recognize the images captured by its cameras, which requires sophisticated machine learning capability, particular because the vehicle must make the right split-second decision when it faces an unexpected situation. A high-speed, low-latency HPC-powered 5G signal makes it possible to offload that complex decision-making to a powerful computer connected to the vehicle.
The trends of digitalization and 5G are creating more data than ever, and high-performance computing is needed to power the big data analytics, machine learning, and A.I. needed to transform it into insight and knowledge. Although the coming wave of digital data may be intimidating in its sheer size, leveraging the ever-more powerful and efficient computing technology enabled by more advanced semiconductors, HPC can help businesses harness its power and ride it to new heights.
Kevin Zhang, Ph.D., is SVP of business development at Taiwan Semiconductor Manufacturing Co. (TSMC).
Never miss a story: Follow your favorite topics and authors to get a personalized email with the journalism that matters most to you.
