Photo Credit: Unsplash.
As programming students and data science enthusiasts, it's interesting to know what is going on in the world of chips and the changes we are witnessing. In this article I'm going to summarize what I've noticed in the sector and how major technology companies are pivoting with the times.
With a global chip shortage and a new emphasis on AI chips many BigTech companies are taking chip production into their own hands. In November 2020 Apple broke its 15-year partnership with Intel. As Intel fell behind in manufacturing, Taiwan's Taiwan Semiconductor Manufacturing Company, or TSMC, has gained in strategic importance.
In 2021 this trend is accelerating. Apple, Amazon, Facebook, Tesla and Baidu are all shunning established chip firms and bringing certain aspects of chip development in-house. China is also racing to develop its own chip supply so as to not be dependent on third parties or the U.S.
Apple’s chips are based on ARM technology as opposed to the x86 architecture that Intel’s chips use. ARM was originally designed for mobile devices and chips built with ARM designs are consistently more efficient, leading to longer battery life.
Meanwhile Nvidia's new Ampere AI chip sets the bar that other companies will try to follow. Since 2020 Nvidia has also benefited from this digital transformation. And since Deep learning relies on GPU acceleration, both for training and inference, NVIDIA delivers GPU acceleration across distributed computing now — from data centers to desktops, laptops and the world’s fastest supercomputers.
For companies like TSMC and Nvidia, their market cap value has increased tremendously since these changes and the relative collapse in the manufacturing prowess of Intel and its ability to keep up. Then there is also AMD and others. According to CNBC, at this stage, none of the tech giants are looking to do all the chip development themselves. Setting up an advanced chip factory or foundry like TSMC’s in Taiwan costs around $10 billion and takes several years.
Tesla is boasting to be an AI-company in its own vein working with so much data. Each major tech company is building its own chip for its specific use cases. These specialized and customized chips can help to reduce energy consumption for devices and products from the specific tech company. We are witnessing a chip explosion in the early 2020s, with a host of established companies and fledgling startups racing to build special-purpose chips to push the capabilities of artificial intelligence technology to a new level.
Increasingly, these companies want custom-made chips fitting their applications so it's a BigTech DIY moment in the history of chips and computing. As major technology companies are battling in the Cloud also with AI and machine learning software for companies, specialized chips are starting to matter more.
China, which monetizes facial recognition at scale, even has specialized chip companies for that sector. Artosyn Microelectronics for instance, a Shanghai-based chipmaker, has released a new generation of AI camera chips, the AR9341. Earlier this year, Chinese chip startup Enflame released its second generation of Deep Thinking Unit (DTU), designed to process huge amounts of data to train AI systems.
BigTech and Chinese innovation is going to push AI chips to new levels in the 2020s. The Cloud, facial recognition, new NLP models and more demand for AI makes this somewhat inevitable. Vastai Technologies released its first cloud AI inference chip, which has a peak performance of 200 TOPS (INT8). To get an idea how dominant Nvidia has become even in China, in terms of AI chips, NVIDIA still dominates the Chinese market with a share of over 80 percent.
The ongoing chip shortage appears rather serious and is likely another major reason why BigTech firms are thinking twice about where they get their chips from. Also with a renewed emphasis on cybersecurity, BigTech firms want to be more careful than ever. Tesla also added hype to the AI-train bandwagon when it announced it's building a “Dojo” chip to train artificial intelligence networks in data centers. In 1919 the automaker started producing cars with its custom AI chips that help onboard software to make decisions in response to what’s happening on the road.
As you can see, this makes the chip sector particularly dynamic in the new normal. Behind the spate of designs is the expectation that AI is the next technological gold rush, and now AI is helping to design its own chips. Google is using machine learning to help design its next generation of machine learning chips and other firms will follow. It's somewhat ironic that Google’s own TPU (tensor processing unit) chips are now optimized for AI computation by machine learning designs.
Nor should Chinese BigTech be discounted. Baidu last month launched an AI chip that’s designed to help devices process huge amounts of data and boost computing power. Baidu said the “Kunlun 2” chip can be used in areas such as autonomous driving and that it has entered mass production. Autonomous driving companies need their own specialized chips.
The new generation of Kunlun AI chips, using 7 nm process technology, achieved a top computational capability two-to-three times that of the previous generation. China is speeding ahead to reduce its dependence on Qualcomm and Nvidia.
Apple has invested heavily in its silicon department, including major purchases, starting with a $278 million purchase of P. A. Semi in 2008, which started the department, and most recently, $1 billion for part of Intel’s modem business in 2019. Taiwan Semiconductor Manufacturing Co. has succeeded by just focusing on production and leaving the design to other companies. Its factories have passed Intel in capabilities and it's shaken up the entire industry.
So with the future of chips you can see many moving parts. To give you an idea on how vital Taiwan's TSMC has become, the world’s largest chipmaker, Taiwan Semiconductor Manufacturing Company, has overtaken Chinese tech behemoth Tencent to become Asia’s most valuable firm. That's rather surprising to many analysts and technology folk. It places Taiwan's strategic importance in geopolitics as more critical than it appears in China's push for territory, technology and talent.
As far back as 2018 BigTech's entry into AR and VR also has needed specialized chip sets. For instance Microsoft created a co-processor for its HoloLens headset (known as a Holographic Processing Unit, or HPU) that handles the information provided by HoloLens sensors. Apple's foray into AR glasses likely has needed much the same. With AI becoming even more ubiquitous in this decade, chip supply chains are having their own revolution and everyone wants in.
As for supply chains in chip shortages, one of the biggest industries that has been hard hit is the automobile sector that doubts whether it will easily end in 2021. This may also complicate the transition plan of major car markers to EVs. Industry leaders are saying that the shortage is thought to have been exacerbated by the move to electric vehicles. Interestingly Bosch, which is the world’s largest car-parts supplier, made a bold statement. It basically said it believes semiconductor supply chains in the automotive industry are no longer fit for purpose.
The chip shortage is so severe for automakers that it's shifting the entire industry in the race to EVs. Germany’s Volkswagen, Europe’s largest carmaker, has lost market share in China in 2021 as a result of the chip shortage. It's not overly clear how much longer semiconductor shortages will last. Daimler CEO predicts the auto industry could struggle to source enough of them throughout next year and into 2023.
In 2021 with Advanced Micro Devices Inc. and Apple Inc. forging ahead with their own capable designs and TSMC’s more advanced production technology and Intel's stunning lack of manufacturing execution, it's a brave new world for the future of chips. BigTech has been making their own chips for at least the last 3 to 4 years but this year is like no other.