The world has suffered for a long time, and this bridge section has almost spread from graphics cards to cars and other more fields.
This difference lies in the 1.25 billion prepaid acquisition, which was transformed into ARM's 20-year authorization, and the oligopoly + oligarchy merger that many people are worried about did not land in the end.
NVIDIA has the best capabilities in the industry in the field of image processing, and the advantages of computing power + processing power are further expanding. Even AMD YES still fell behind. During the same period, NVIDIA will release three Arm IP-based processors in 2021, TB-class data center CPUs, new-generation DPU, and 1000TOPS autonomous driving SoC Atlan.
If Nvidia completes the acquisition of Arm, it will directly grasp 90% of the global mobile processor share, 90% of the Internet of Things processor share, and the ability to participate in Arm's underlying access and next-generation planning. In other words, there is a danger of "more difficult to obtain Arm technology" for other chip manufacturers around the world - Qualcomm, Huawei, Intel, Microsoft, etc. Along with this, there are a series of impacts such as Internet security/national regulations, the most critical of which is the regulation of "monopolies" in China, the United States, the United Kingdom, and the European Union.
As for if this acquisition is ultimately successful, given that Nvidia has a graphics card in its hand in the past few years, watching the mining market market speculation, the night of NVIDIA in 2021, the operation of harvesting profits, and how the automotive autonomous driving/Internet of Things field is being played, it is obviously an unpredictable topic.
China/US/EU/East Asia, chips have more than one war
The struggle for strength in human society has always been a development trend in which the share is roughly fixed and the strong have the right to speak. From arable land resources, to land resources, to solid fuels to gold standards, silver standards, oil standards and currency peggings. The various transnational wars that subdivide the world are almost without exception.
In the era of Internet intelligence, the battle of computing power has begun to be pulled to a new height, who grasps the future power of computing power, first grasp the relevant definition of intelligence, and then roll back to such a previous land/oil/finance general process.
The current advantage of computing power/chips is in the hands of the United States. Intel/AMD, Qualcomm/Nvidia, are all high points of existence. The current situation of money and technology, several giants plus the acquisition of some start-up companies of high-quality blood, the sense of giants under the rollback was once full. However, that is, the chip shortage that is still continuing at present, as well as a series of actions such as forcing Japanese and South Korean companies to hand over the core technology of chips, the pattern of this world can be seen by the clear-eyed people, and the pattern has changed.
The original chip globalization pattern has been challenged by a new round, and if you want to win the future discourse power related to intelligent computing power, you must grasp your upper level discourse. Interestingly, just the day after the ARM and Nvidia acquisitions collapsed, the EUROPEAN Union's €34 billion European Chip Act landed, and a few days before the bill landed, the $51 billion U.S. Chip Act landed. China's chip program has also advanced.
Competing for self-sufficiency under the chip shortage and competing for the right to speak in the future chip computing power has become a new round of main line based on the country. Mixed with many voices such as "500 billion US dollars is enough to be self-sufficient" and "the EU is extremely difficult to compete for 20% share".
The chip battle turns from defense to attack, who will fight? How to play?
There are many classic cases in history of turning defense into attack, the Battle of Shuishui, the Battle of Xuzhou, the Battle of Jinzhou, etc. are familiar stories in the hearts of adult men, resisting pressure, completing the breakthrough, and turning defense into attack.
For the global semiconductor-related industry, how to fight and how to fight its own water war is relatively embarrassing, but there are also opportunities in embarrassment. The main reason for the embarrassment is that the bottom layer of ARM, x86 architecture end can not be touched and locked, it is very difficult to jump out of here, the underlying property rights of x86 are in the hands of Intel and AMD, ARM has independence, but from the acquisition of Nvidia, it can be seen that the semiconductor industry is salivating.
NVIDIA, which did not win ARM, is still at a level with Broadcom and Qualcomm, while NVIDIA, which won ARM, can break the wrist with Intel and AMD.
For the chip industry in various countries around the world, further attacking ARM is obviously a good way to turn defense into attack, although it is extremely difficult but not without opportunities. Because on the day of ARM's acquisition with Nvidia, its CEO was immediately replaced, Rene Haas will gradually complete the handover of work after the debut, and at the same time begin to promote ARM's IPO (currently mainly based on NASDAQ), the deadline is currently at the end of ARM's fiscal year, March 2023. It can be understood that after ARM won new power in the capital market, it began to promote the research and development of next-generation technologies, and also opened up for more semiconductor companies to cooperate with them in depth.
However, the wealth fundamentally relies on ARM, and there are many variables for semiconductor companies. After all, arm independence/importance is important and regulated globally.
However, in the era of computing 3.0, heterogeneity has become a new trend. The importance of a single architecture is declining in the 2.0 era of general architecture service computing, and under the combination of x86+ARM+ open source risc-v, various CPUs, DSPs, GPUs, AI chips, and FPGAs exist at the same time to achieve the composition of intelligent computing power.
This is also one of the core reasons why Intel, the giant who controls x86, is not in a hurry to spend a lot of money on ARM transformation (analogous to Volkswagen's broken wrist fuel transformation electric vehicle), because there are infinitely many solutions to counter ARM's embedded layout, and there is no need to simply benchmark.
In the future industrial field, there will undoubtedly be a large number of intelligent computing power related keywords, from the perspective of the existing automotive industry, there are not many companies that can have core interests in the computing power side, because of the reason that technology is first-mover, the supplier level cannot match the needs of car companies to make rapid efforts. At present, in addition to Tesla's self-developed chips, the rest of the car companies have very few similar actions, on the one hand, because of technical strength / investment funds, etc., on the other hand, because car companies have different judgments on the future development of the automotive industry, such as "can rely on core partners to achieve similar effects, do not have to develop themselves."
However, this also directly reflects that "most car companies are at least 3-5 years away from Tesla's intelligent underlying technology", including the soa concept of the electronic and electrical end (the industry judges that a small number of car companies can reach Tesla's existing capabilities in 2023, that is, centralized control), and then there is a self-developed core chip part tesla has rushed to run in 2019, and the news of other car companies in the automotive industry that entered the self-developed chip around 2021 has been released one after another.
However, despite the technical first-mover/architecture of the bottom layer, etc., looking back at the relevant developments in the Chinese market, we can still see many opportunities. For example, Huawei's A+D layout, the general computing terminal is ARM's Kunpeng computing platform, and the AI computing terminal is the ascending computing platform developed by relying on open source architecture. Another example is the one-cloud multi-core proposed by Ali.
ARM has a foreseeable advantage in the coming decades, and the moat remains strong. As of November 2021, ARM has about 1,000 authorized cooperation and 320 partners around the world, but no more than 20 manufacturers have purchased architecture authorizations, and China has Huawei, Feiteng and Huaxintong. In the automotive industry, not long ago, Geely Automobile's technology company ECARX and ARM's Core Engine Technology are IP licensed and have not yet touched the bottom.
Write at the end:
It can be seen that in the era of "computing power main line" in which you and I are located, in the future development trend of automobiles, whoever can grasp the computing advantage can grasp the relative core advantage. However, for most car companies, it is quite difficult to rely on their own technical reserves to form intelligent advantages, because of the complexity/advanced nature of the chip industry itself, the traditional manufacturing industry is very different from its position.
In view of the fact that the development direction of the chip industry is very multi-source, and the heterogeneous trend + advantages in the computing 3.0 era are constantly emerging, we can see many possible breakthrough directions in the future. For example, under different combinations of computing power, the realization of automatic driving can focus on different fields such as vision, sensing, radar, and communications.
The pressure is huge, the opportunities are abundant, and this is the conclusion that the recent series of hot events can leave to you and me. And it is worth looking forward to that with the landing of relevant laws in Europe and the United States, the time for you and me to witness the landing of China's relevant strategies in 2022 is also getting closer and closer, and at that time, we are likely to see what kind of different thinking can be brought about at the national innovation level. For example, the power of the open instruction set RISC-V?