After more than a year of entanglement, under regulatory pressure, NVIDIA had no choice but to give up the acquisition of ARM, which was a relief for semiconductor manufacturers who were using ARM authorized IP, but for The Japanese SoftBank Group, which had poor cash flow, $66 billion was a big single flight, and ARM's hot potato had to continue to be held in its hands.
The reason why ARM is a hot potato of softbank group is because Sun Zhengyi was hot-headed and expensive.
Industry analysts gave a valuation of $25 billion to $35 billion for ARM, which was originally sold for $32 billion. In addition, the US dollar into the interest rate hike cycle, technology stocks discount into fate, ARM wants to be stuck in the valuation of 35 billion US dollars ceiling listing is not small, which means that SoftBank intends to cash out ARM through the way of IPO, even whether the capital can be recovered needs to be questioned.
Can ARM rely on performance to drive valuations up over time? After all, in the field of mobile processors, people are the boss. At present, this road is also full of thorns, because ARM's business model has encountered the resistance of competitors, and the strength is not small.
ARM's business model is a closed instruction set, but provides CPU/GPU soft core and hard core design licensing, commonly known as selling chip design (IP license). It should be noted that the soft core (soft core) design authorization can modify part of the design code, customers can customize according to their own needs, to meet personalized needs, suitable for Apple, Qualcomm and other large companies with chip integration design capabilities, hard core (hard core) design authorization can not modify the design code, can be seen as a complete kernel design product.
However, whether it is a soft-core design license or a hard-core design authorization, customers do not have independent chip core design capabilities, which are firmly in the hands of ARM, and ARM has monopolized the mobile processor market.
Intel, AMD monopoly X86 processor market, the two days of life that is called a moisturizing, push this and that, monopoly of the mobile processor market ARM days should also be very good. The fact is that if Intel is rich, ARM can only be regarded as a middle-class home, and the gap is related to ARM's business model of obtaining income by licensing.
ARM's licensing revenue is mainly divided into two parts. First, the IP authorization mentioned above, that is, the income from directly selling the kernel design, the price is between a few million yuan and hundreds of millions of yuan, which depends on the strength of the customer's wallet, but there are actually very few customers who can buy ARM kernel authorization, including Apple, Qualcomm, MediaTek, Huawei, Samsung, one hand is counted, Huawei is pulled on the entity list, ARM's large customers are one less. In short, IP licensing revenue does not account for the majority.
The real head income is royalty, that is, according to the price of the chip, generally in a few cents / piece, high-end including mobile phone chips, ARM extracted royalties will exceed 10 yuan / piece, but low-end such as power management, TV image processing chips, the royalties extracted are mostly about 1 yuan per piece, and some will even be as low as a few cents yuan / piece.
The iron law of the market is that high-end products are at the tip of the tower, the proportion is very small, the largest proportion is the base part of the tower, ARM is no exception, which makes its high-end chip proportion small, the proportion of low-end chips is large, affecting the royalty income, and the scarcity of large customers, but also makes the total amount of IP licensing income with the highest single income limited, the result is that the number of chips using ARM IP is extremely large (25 billion chips shipped in 2020), but the overall revenue is not high, and the growth is not fast.
According to public data, ARM's operating income to 2015 as a watershed, before 2015 belonged to the rapid growth stage, from 2008 546 million US dollars to 2015 1.559 billion US dollars, 7 years of growth of about 1.86 times, after 2015, ARM entered a growth bottleneck period, 2020 revenue of 1.98 billion US dollars, 4-year annualized growth rate of 6.75%, and 2017, 2018, 2019 three years of revenue are more than 1.8 billion US dollars, Growth stalled, with IP licensing revenue falling even 11.5% in 2018, which was almost impossible before, suggesting that the potential of ARM's large customer market has been almost exhausted.
It can be seen that ARM's business model is to look fragrant (low cost, no tapeout, but also save the huge cost of building a chip manufacturing plant), but it is not delicious to eat, don't look at it is filled with a large bowl of gold, but it is completely incomparable with Intel's vat of gold, even if X86's second oldest, AMD, which has been beaten by Intel for many years, the net profit in 2020 will reach 2.49 billion US dollars, which is 1.26 times THE operating income of ARM in the same period. The operating income has not dried up to the net profit of its peers, which shows the limitations of ARM's business model.
ARM's business model is inherently inadequate, and if it encounters spoilers again, the rice in the bowl will be difficult to eat. Now, the spoiler has emerged, and it's the RISC-V architecture.
X86 chip represented by Intel is the instruction set, microarchitecture and product fully enclosed, belongs to what chip to sell, the market can only passively accept, ARM is instruction set closed, microarchitecture part open, product fully enclosed, that is, customers can partially customize, can also buy ready-made products, RISC-V is further, instruction set open source, microarchitecture open, can also let customers buy ready-made products, more flexibility. The difference between the three is shown in the figure below.
The difference in business models determines the cost of customers using the three architectures, the highest cost is the X86 chip, followed by the ARM architecture, and the lowest cost is RISC-V. Because of this, RISC-V is popular in the field of Internet of Things. The characteristics of the Internet of Things chip are that the personalized requirements are high, the absolute number is not large, but it is sensitive to the cost, and the ARM architecture IP authorization plus royalty extraction, which raises the cost of the Internet of Things chip, the customization elasticity is small and the customer demand is depressed, and the flexibility of the RISC-V open source can just meet the demand for the Internet of Things chip, so in just a few years, it has conquered a lot of territory, and the Internet of Things field has the meaning of being equal to ARM, and also encroaching on ARM's inherent territory in the fields of automotive electronics and 5G equipment.
A clear sign is that in 2021, risc-V's membership has skyrocketed by 130%, showing a strong momentum.
It should be noted that RISC-V is currently hot, not only the credit of the business model, but also inseparable from the god assist of the United States. RISC-V is an open source architecture, and the foundation's headquarters have moved to Switzerland, so it is not subject to U.S. export control regulations, which is why RISC-V is popular in many countries, especially in China. China is the largest chip market, but also the largest Internet of Things market, its enthusiasm for RISC-V, undoubtedly made ARM's pace of expanding the kingdom territory have to slow down.
In the current international political landscape, X86, ARM and RISC-V in the chip market is already a high probability event, which is not good news for SoftBank, whether the acquisition of ARM can make money to cash out is more confusing, Nvidia gave up the acquisition of ARM There is no loss, but it is inevitable that SoftBank under the helm of Son Zhengyi will sigh.
——END——