Artificial Intelligence Chips Set To Redefine the Processing Limits Of Computers
Artificial intelligence (AI) technology is on its way to enter into the processors of computers and smartphones. From robotics to transportation, AI is prevailing in every sector in the world. Today’s computer programs generate enormous amount of user data, which becomes a laborious task to maanage using traditional chips. However, incorporation of AI chips offered novel methods such as pattern recognition to analyze the data. Using AI chips, a simple tabletop speaker can be transformed into a smart home assistant and mobile into a smart camera, which in turn fuels the demand for AI chips. According to a research firm, Allied Market Research, rise in demand for AI chips and emergence of autonomous robotic technology are expected to boost the growth of the AI chip market. Moreover, rivalry between tech giants such as Google, Intel, and Tesla to deliver the fastest computing chip has been instrumental in supplementing the growth of the market.
Tesla To Launch Self-Driving AI Chip
Elon Musk, Co-founder and CEO of Tesla, recently announced that Tesla’s super-advanced AI chip will be unveiled in the next 4-6 months. These chips will take over the existing Nvidia Drive PX 2, which are currently installed in self-driving cars. Musk confirmed that the users of pre-ordered “full self-driving” feature will get the upgraded chip for free.
According to the company, Tesla’s AI chip can process up to 2,000 frames per second, unlike the PX processors that compute 20 frames per second. Pete Bannon, Project Leader, stated that the AI chip supports existing machine learning (ML) networks “with a lot of idle cycles to spare”.
Tesla has so far relied on Nvidia’s Drive Platform. However, to focus on the particular operations and improve efficiency, Musk has decided to develop in-house AI chips that can boost the performance of self-driving cars. He recently declared, “We have been in a semi-stealth mode in developing AI chips since last 2-3 years. However, it is time to let the cats out of the bag.” According to Musk, in the next four to six months, the world will witness the fastest AI chip in the market, which will cater to the specific requirement of Tesla’s autonomous cars.
Nvidia Unstoppable In AI Chips Business
Nvidia, the pioneer of graphics processing unit (GPU), uses AI chips to offer a more realistic gaming experience. Moreover, AI can help tackle the challenge of creating life-like light and create shadow effects in the game. Recently, Nvidia unveiled a new graphics chip, Nvidia Turing, which fuses together real-time ray tracing, artificial intelligence, and programmable shading. Making illusions such as lifelike reflection, refractions, and other effects require high computing power, which traditional chips fail to deliver. On the other hand, AI chips can speed up the process. Nvidia’s AI-based approach involves creating an image by overlaying multiple elements to achieve illusions of tackling lights and shadows. The traditional approach increases efficiency by rendering part of an image, while AI predicts and fills in the remaining light effects to give a lifelike experience to gamers. Furthermore, AI chips have provided an impetus to the revenue of Tesla. Since the launch of Turing, the company shares have skyrocketed, making it second to none in the AI chip market.
Surge In AI Chip Start-Ups
For decades, leading tech companies have overlooked start-ups that manufacture AI chips, questioning their ability to compete with giants such as Intel and Google. However, innovations in AI and ML have reopened the question of how to build computers, which can be answered by start-ups who offer ground-breaking ideas and technologies to build faster computers. Today, there are at least 45 start-ups that are on the verge of redefining the AI chip industry. Furthermore, there are about five companies that successfully raised more than $100 million from investors. For instance, venture capitalists invested more than $1.5 billion in AI chips start-ups in 2017, nearly doubling the investments as that in 2016. It is a challenging task to go toe-to-toe with Intel and Nvidia in the AI chip business, which would require billions of dollars of investment. However, these start-ups help to develop dedicated chips that provide enough computing power to support ML, which are acquired by existing tech giants. Owing to the dearth of new and fast processing power, start-ups have gained a rare chance against the entrenched companies to compete toward building faster AI chips.