NVIDIA’s $40 billion acquisition of arm’s China chip?

On September 14, NVIDIA officially announced that it would acquire arm, a semiconductor design company of Softbank group, for us $40 billion. NVIDIA delivered $21.5 billion worth of shares to Softbank and paid another $12 billion in cash, including $2 billion at the time of the signing. The deal is expected to be completed within the next 18 months. < / P > < p > if the deal goes well, it will be the largest acquisition in the history of semiconductor industry. Previously, the highest acquisition price of NVIDIA took place in March 2019. At that time, NVIDIA paid $6.9 billion for mellanox, an Israeli chip manufacturer. < / P > < p > it is reported that apple, Samsung, Qualcomm and NVIDIA have all had “scandals” with arm. It is reported that Softbank has made several contacts with the above-mentioned enterprises. But in this competition, in the end, NVIDIA insisted to the end. NVIDIA is a global giant in the field of GPU, and has a deep layout in the fields of artificial intelligence, automatic driving and cloud data center. The market value of the company is as high as 300 billion US dollars. < / P > < p > as a global leader in GPU and AI, why does NVIDIA pay so much for the acquisition of arm? At the same time, what kind of far-reaching impact will this largest transaction in the history of semiconductor industry have on domestic enterprises? < / P > < p > when we review the acquisition of mellanox last year, we can find that NVIDIA is not satisfied with continuing to “win” in the GPU field. After the acquisition of mellanox, NVIDIA’s operation capability in the data center has been further strengthened. < / P > < p > according to the latest second quarter financial report of this year, NVIDIA’s data center business revenue reached 1.752 billion US dollars, surpassing the game business of US $1.654 billion, becoming the largest revenue source of NVIDIA for the first time, including the acquisition of mellanox. For NVIDIA’s growth in data center business, CEO Huang Renxun said that NVIDIA’s future success does not lie in its ability to optimize performance at the chip level, but to promote the development of data center scale computing. < / P > < p > but the data center is not the end of NVIDIA. According to NVIDIA’s vision, it hopes to push its AI strategy to a new height through GPU and data center business, and extend to the cloud computing field with broader prospects. < / P > < p > it is worth mentioning that in recent years, many leading cloud computing manufacturers are actively developing and deploying products using ARM architecture. The advantages of arm architecture in the field of cloud computing are also valued by NVIDIA. < / P > < p > What’s more, arm, as the world’s leading semiconductor IP provider, plays an important role in the chip industry. There are even reports that its chip design has been used in more than 90% of the world’s smart phones, tablets and other mobile devices. According to Teng ran, deputy general manager of CCID Consulting IC Industry Research Center, “at present, NVIDIA has the world’s leading GPU and AI platforms. Arm not only has a broad market and prospect in consumer grade mobile devices, but also has overcome difficulties in traditional markets such as HPC and PC in the x86 camp. If NVIDIA can better combine its advantages in GPU and AI fields with ARM technology, it will surely play its role in HPC and PC fields. ” < / P > < p > in fact, in the mobile field, NVIDIA has also launched Tegra series processor based on ARM architecture before, but due to the high fever problem, it is not suitable for mobile platforms such as mobile phones and tablet computers. In this regard, Shen Meng, director of Xiangsong capital, believes that “in the past, NVIDIA’s advantage has always been the development of technology in the field of GPU, and its products are more positioned in advanced applications such as AI and GPU. However, NVIDIA’s main competitor, AMD, is expanding from the Perspective of the integration of CPU and GPU. So for NVIDIA, if arm can be acquired, the most important point is to extend from GPU to CPU, and even enter the Internet of things system with a larger market in the future. ” < / P > < p > as mentioned earlier, once the acquisition of arm by NVIDIA is completed, it means that NVIDIA will probably enter the CPU field in the future, and even break the existing chip market pattern of mobile terminals. And in this, the most reluctant to see NVIDIA enter the arena is the old rival Intel. < p > < p > in July this year, NVIDIA surpassed Intel in market value and became the third largest semiconductor manufacturer in the world. At present, NVIDIA’s market value is significantly ahead of Intel by more than $100 billion. < / P > < p > in the field of GPU, Intel, AMD and NVIDIA form a “tripartite confrontation”. According to the report released by JPR, a market research organization, in the fourth quarter of 2019, Intel ranked first with a global market share of 63%, while AMD and NVIDIA ranked second and third with 19% and 18% market share. < / P > < p > it should be noted that this market report includes CPU built-in core display calculations. If we put aside the low-performance core display, NVIDIA’s global unique share has reached 73%, and the remaining 27% of the market share is occupied by AMD, while Intel has no revenue. < p > < p > but Intel still intends to “take a share” in the unique market. At present, Intel has officially announced that it will launch a 10 nm technology independent graphics card, using its own GPU architecture, which is expected to be launched in 2021. This is also Intel after the launch of Duxian in 1998, after many years to enter the field of independent GPU. < / P > < p > regarding the acquisition of arm by NVIDIA and Intel’s redevelopment of Duxian layout and other topics, the reporter asked Intel’s relevant contacts for details. The other party said that it would not comment on the acquisition and did not respond to the topic of Duxian layout. < p > < p > in Shen Meng’s opinion, it is not difficult to understand Intel’s rework. “Vector computing has become the development trend of the next generation of computer technology, but vector computing needs the powerful computing power of high-performance GPU. If Intel wants to improve the ability of deep learning and vector computing, it must develop its own high-performance GPU.” However, with the launch of NVIDIA’s new generation of < 7p > technology, the new generation of NVIDIA < 7p > was launched again. As a latecomer, Intel will not be easy to challenge NVIDIA’s current monopoly in the independent market. < / P > < p > looking at the CPU field, although Intel’s x86 processor once dominated for many years, now with the decline of PC market, ARM processor has a broad prospect in the mobile field, which has a great development momentum to replace x86. To some extent, NVIDIA’s acquisition of arm may also have the intention to speed up the process of arm replacing x86. Once its transaction is successful, Intel will likely fall into a disadvantageous situation in the competition in related fields. < p > < p > in Shen Meng’s view, “Intel’s advantages in CPU field are mainly applied to PC, notebook and even server. Combined with its own strength in production and manufacturing, Intel’s disadvantage lies in its high power consumption and poor performance in the mobile market in the past. On the contrary, NVIDIA and arm have very obvious technical advantages in their respective GPU and mobile architecture. If the two companies are combined, NVIDIA will rely on the technology accumulation of arm for many years and erode Intel’s market share in the CPU field. ” < p > < p > arm has been so successful since its establishment for 30 years. On the one hand, it is based on its strong chip architecture design and technical capabilities, and on the other hand, it has always maintained its neutrality to the markets of various countries. However, this neutrality is likely to be broken with the acquisition of NVIDIA. In announcing the acquisition of arm, NVIDIA said that the deal would need to be approved by regulatory authorities in the UK, China, the European Union and the United States, and the deal is expected to be completed within 18 months. Among them, Britain is the first “roadblock” in the deal. < / P > < p > previously, Softbank group had promised the UK side that it would keep the UK high-tech company founded in the UK when it acquired arm in 2016. However, after Softbank transferred arm to NVIDIA, an American company, it immediately aroused strong opposition from the British side, believing that this was not in the public interest of the UK. < / P > < p > although NVIDIA also said after the acquisition that it would continue to maintain arm’s business model of opening license and authorization before the acquisition, people in the British industry did not appreciate it. In particular, Hermann Hauser, the co-founder of arm, made a public voice against NVIDIA’s acquisition of arm and called on relevant departments in the UK to “rescue” arm. It is also difficult for NVIDIA to obtain regulatory approval from China for its acquisition of arm. In October 2016, the US communications giant Qualcomm intended to spend US $38 billion to acquire the Dutch semiconductor company NXP. However, after 20 months of negotiation, only the Chinese market regulatory authorities failed to approve the deal, which led to the collapse of the acquisition. < / P > < p > how to maintain the neutrality of technology and effectively protect the interests of the Chinese market has become a topic of concern to domestic enterprises and industry insiders. In this regard, the reporter wrote to NVIDIA and called arm China for an interview. However, as of the time of publication, neither company made a response. According to Teng ran, “the main customers of arm include apple, Qualcomm, Samsung, Huawei and other high-tech enterprises in different countries. These customers either directly use the architecture of arm as the basis of their own chips, or they can develop their own chips through arm authorization. Now, although NVIDIA does not directly compete with arm’s customers, NVIDIA has the ability and motivation to compete with arm’s existing customers in the future. Therefore, in addition to obtaining the approval of regulatory authorities in various countries, whether NVIDIA can obtain the trust of arm’s existing customers is also a key issue. ” < / P > < p > compared with NVIDIA, Softbank of Japan has not been so controversial in its acquisition of arm. The reason may be that Softbank, as a venture capital company, does not have a so-called commercial competitive relationship with arm’s customers, so arm’s market neutrality is maintained. < / P > < p > however, Softbank’s development has not been as smooth as expected after taking arm at a cost of $32 billion in 2016. In the past fiscal year 2019, Softbank had a record operating loss of $13 billion. At the same time, the acquisition of Softbank arm, profitability is not increased, but decreased. In 2015, arm’s profit was still $843 million, but by 2019, arm’s net profit had dropped to $276 million. < / P > < p > “the purpose of Softbank’s acquisition of arm at that time was to expand the company’s business to the Internet of things. However, due to the failure of Softbank in some subsequent large-scale investments, including the loss of vision fund, which Softbank values most, the overall performance fluctuates greatly. Although Softbank bought at $32 billion and sold it at $40 billion now, it seems that there is a good return on investment in four years, but in fact, Softbank is trying to survive by breaking its arm, and the profit it has made is insignificant compared with the loss of the entire Internet of things market. ” Shen Meng said. < / P > < p > previously, the issue of arm’s “cutting off supply” to Huawei has become a heated topic. Once arm is acquired by NVIDIA, it means that arm will become an American company. No matter how the acquisition results, it is imperative for domestic enterprises to reduce their dependence on ARM architecture chips and improve their independent R & D capabilities.