Using the popular technical term "computing power" to divide the semiconductor vendor landscape, the former PC trio can be replaced by the computing power trio, after all, the three are now aiming at the direction of the development of computing power rather than the traditional PC.
As the triumvirate of computing power, NVIDIA, AMD, and Intel (and there's a reason they're all listed that way), the conversation around acquisitions during this Chinese New Year of the Tiger seems to be shaping the tone of the coming year, and will be the cause of a lot of change in the future of computing power competition. Because precisely in the last week, three coincidentally about the acquisition of various news events, we will take this opportunity to a little gossip.
NVIDIA: Missing out on arm isn't necessarily a bad thing
The most shocking news from last week has to be the failure of NVIDIA's acquisition of arm due to obstruction from all sides, which is almost certainly a deal that could have affected a third of the semiconductor industry. From the perspective of the news from all parties, a considerable number of regional market management bodies did not approve the transaction is the main reason for the failure, and NVIDIA with $ 1.2 billion in exchange for the so-called 20-year authorization of many people feel that NVIDIA loss. From a financial point of view, NVIDIA acquisition will not only lose 1.2 billion, after all, the pre-acquisition of a variety of third-party evaluation costs are also fewer hit the water, but for NVIDIA, the overall may not be a bad thing, after all, the longer the price of the graphics card firm, the better the financial report.
First of all, arm is worth 40 billion dollars, softbank bought arm after arm to do some subtraction, part of the service is not involved in the transaction, China is also in the form of a joint venture for part of the cut, such an arm if all can be sold at a premium (softbank bought only 37 billion), taking into account softbank's poor economic situation and continue to cash in a sense of urgency. NVIDIA is suspected of taking over. Remove any anachronistic behind-the-scenes trading speculation, we can only assume that NVIDIA since the stock price soared, it seems to really want to take advantage of the opportunity to unify the arithmetic rivers and lakes. After all, look at their own two rivals, can be CPU + GPU + FPGA full quota standard configuration, their own only rely on GPU obviously can not continue to maintain the semiconductor first stock of the high market value, and to take arm undoubtedly let themselves become a new ecological "king", said Qualcomm is a patent hooligan, if NVIDIA really take The company has a lot of patents that are related to computing, and I'm afraid half of them can't escape from this company.
A question worth thinking about is what happened in the year and a half between NVIDIA's agreement to acquire arm and its eventual rejection, arm's revenue did not grow much, and RISCV is getting hotter by the day. Thanks to many manufacturers of NVIDIA's acquisition of arm after the uncertainty (who knows after the acquisition of arm will not be another MIPS), a lot of arithmetic-related enterprises began to RISCV as their future plan B, which indirectly caused a big explosion of arithmetic devices based on the RISCV core. If the deal continues, this trend will become more and more obvious. Now arm continues to be independent, if no one continues to apply for the acquisition of arm in the short term but choose arm listed, the user's confidence in the future of arm independence may be reshaped, so RISCV may return to the original Plan B role. As for the rumors of Intel or other homes want to acquire arm, it is estimated that in the short term are very difficult to get over the regional regulatory agencies. Softbank to take the money, can only be to promote arm listed, then in the short term arm to do is to RISCV camp of the strong suppression, otherwise the valuation will be greatly discounted.
As for NVIDIA, at least the stock market reaction to see investors cautiously optimistic about the outcome of this future, that they are not too sure that the acquisition of arm on NVIDIA is necessarily a good thing, just NVIDIA in the short term can only continue to move forward with a single-core, while the server market is still very difficult to squeeze in. Of course, it's a pity that the old Huang thought of a better future for those ......
AMD: Successful acquisition of Xilinx, beyond the old rival
AMD, the biggest competitor of NVIDIA in the past two years, had a good life, although not as good as NVIDIA's. The company's main business is to provide the best possible service to its customers. NVIDIA so dazzling, but the market value also followed the dawdling rise, CPU market share climbing, GPU market benefited from the shortage of graphics card profit margins step by step up, coupled with the acquisition of Xilinx (Xilinx) progress smoothly, in this Spring finally completed the transaction as desired. In addition to the processor composition of the old rival Intel in addition, an unexpected joy is that the moment the deal was reached, the two combined total market capitalization, beyond Intel. to know four or five years ago AMD's worst when the market value of less than one-twentieth of Intel may be. Now you have to ask AMD and Intel who is the bigger company, I'm afraid people have to turn on the financial channel to take a look at the market capitalization to answer you.
The acquisition of Xilinx was hardly a disadvantage for AMD. Financially, the choice of equity swap does not have much financial expenditure; technically, after the inclusion of FPPGA, AMD has a more comprehensive pool of arithmetic technology at the same time also significantly improve their own AI technology reserve capacity (if we take NVIDIA as the AI strength of 100 points, Intel can probably reach 85 points, while the acquisition of the pre-AMD is probably only less than 50 points, and now can reach the level of 70 +); from the perspective of the market, after the addition of Xilinx's FPGA, AMD instantly expanded its competitiveness in the server market, while also penetrating a number of market segments such as industrial, medical, defense and aerospace. The two previously in the server and core computing equipment, to face Intel's CPU + FPGA combination of competition, and now if you reevaluate the two sides of the server side of the strength of the two sides, it has been difficult to instantly distinguish between the obvious strengths and weaknesses. The only uncertainty is whether AMD is able to better integrate FPGA technology, and how to face Intel's best ecosystem building with the server ecology capability that both sides lacked slightly before.
Intel: IDM2.0 is very interesting
Intel quietly announced the completion of the acquisition of the semiconductor foundry Tower, the acquisition of $ 7 billion is not expensive, even taking into account the shortage of fab profits from the surge, which is not high annual revenue for Tower has been very high. premium. However, we can also see Intel's strategic direction to go very firmly, starting from IDM2.0 to the later foundry services strategy, and then the ensuing open X86 architecture, as well as the current acquisition of Tower Jazz, Intel continues to enrich their own technical strength in semiconductor manufacturing. If Intel seems to lack an edge over its two main computing rivals in terms of products, manufacturing is probably the key piece of the puzzle that Intel is still able to suppress right now.
For Intel, catching up with Samsung and even TSMC process levels is now an important means of reclaiming its CPU performance advantage, but to catch up with rivals requires sufficient investment in R&D, which not only obtains profits from its own products, but also from the foundry side as much as possible to share the cost of R&D. Otherwise, the sky-high cost of process R&D can only be dragged down by itself. The acquisition of Tower is a strategic move for Intel to continue to implement its IDM2.0 and foundry business strategy.
First of all, although Tower is not big, but it is a standard foundry, with a perfect foundry service system, and perennial as IDM Intel is the most lack of foundry service system, the acquisition of more than build a team of their own instead of a lot easier. Moreover, with Tower, Intel will logically enter the foundry circle, can carry out more foundry services, not just limited to the production line before Tower. We analyzed the current foundry market pattern, in addition to the GF from AMD, Tower is Intel can now buy the largest foundry, and GF is now in the hands of the Middle East, and Tower is controlled by Israeli capital, for Intel, Israel is basically the United States outside of the second largest R & D and generation base is located from this point on the Jewish gene! ...... You guys get it.
Secondly, Intel has been fighting for the U.S. mainland's various fab funding support in the past two years, and has also gotten Israel's semiconductor factory fund, and hope to get more U.S. government support from the hands of TSMC. A feature of government support is the hope that the fab can meet U.S. defense standards and services, which Intel naturally will not let go easily. tower has two fabs in the U.S., Intel can apply for defense certification for the two plants after the acquisition. More importantly, Tower, although small, but is also the largest analog wafer foundry, and Intel's lack of experience in analog there is no analog fab, most of the defense applications happen to be analog and mixed-signal-based, take Tower allows Intel to quickly have a more complete analog chip production capacity, which is very meaningful to the future to get more U.S. government funding support.
This is the first time that a company has been able to get the best out of a company.
From these two points, it is not difficult to see that Intel's silent acquisition of Tower, its strategic significance is very important, on the one hand, Intel directly into the role of foundry and can use the mature foundry service system to quickly transform their original fab, on the other hand, Intel has become the most powerful U.S. domestic analog foundry, and may even be in the United States in the future to build a 12-inch analog foundry. Preparing to build a 12-inch analog foundry, which will be the most ideal chips to get the government's financial support. Of course, when Intel can rely on the foundry business to feed will CPU, and re-establish the lead in arithmetic technology, we will be difficult to predict.