Intel (INTC), which is the king of semiconductors and microprocessors since 1991, is being dethroned as the world’s leading chipmaker for electronics, ranging from personal computers all the way to technologically advanced automobiles. The microchip behemoth is facing competition from multiple fronts. From one end there is currently a paradigm shift from decreasing demand for general all-purpose computer chips to increasing demand for domain-specific chips. And from the other end, foreign companies, such as Samsung, are trending to outplay Intel in their very own game by outperforming the incumbent leader in manufacturing general purpose microchips and unit sales.
Many tech companies today require specifically tailored chips made for specialized needs, used in mobile devices, servers running tasks in artificial intelligence (AI), and the newest in-home devices, dubbed the Internet of Things. In fact, many current companies that make their own gadgets and products, such as Alphabet (GOOGL) and Amazon (AMZN), are in the process of beginning their own in-house chip factories.
On top of in-house microchip manufacturing, a whole new wave of chip startups could bring fresh competition into this sector. According to Intel co-founder Gordon Moore, the number of transistors on a microchip doubles about every one-and-a-half to two years. This means that the performance of a microprocessor doubles about every two, or so, years. However, a look at the latest generation of microprocessors indicates a performance improvement of only three percent. Additional research and projections state that the next big reduction in the size of transistors will not make general-purpose microprocessors that much faster. The only way to empower future AI devices and machines is to incorporate “reduced-instruction set computer”, or RISC, microprocessors into the manufacturing process. Apple (APPL) is one of the first major tech companies to use RISC-based chips in their iPhones. As of now, there are a few players in this space, with new competitors jockeying for position in this upcoming and lucrative sector.
Google recently developed a microchip, called the Tensor Processing Unit, or TPU. The TPU outperformed general computer microchips in tasks like machine learning. While general all-purpose microprocessors are largely useful for running everything from web browsers to complex spreadsheet construction, TPU’s only focus on specific tasks, such as facial and fingerprint recognition. These units expend power more efficiently than regular computer chips. Dedicated semiconductors for specialized tasks seem to be the way of the future for this industry.
In other news, Samsung Electronics will release guidance on their chip sales, this coming Friday. The South Korean conglomerate is expected to outpace Intel in semiconductor sales in the coming quarter.
Also published on Medium.