AMD’s acquisition of Xilinx, which was overtaken by leading US newspapers more than two weeks ago, was finally announced with AMD’s third-quarter earnings announcement. AMD’s alumnus was a little nervous because there were cases where large-scale acquisitions were stalled due to the lack of news media, but when I opened the lid, CEO Lisa Su made this large-scale acquisition relatively easily. Announced. Mr. Ohara has written a very accurate article about the content of the announcement, so please refer to that. Here, I would like to describe my miscellaneous feelings as an AMD alumnus.
AMD has grown through several major acquisitions over its 50-year history
Like most Silicon Valley independent semiconductor companies, AMD has grown with several major acquisitions.
One of the biggest things I’ve experienced is the acquisition of MMI (Monolithic Memories), NexGen (1996) and ATI (2006) in 1987, the year after I joined AMD. Fortunately, I myself have never been acquired in my 30 years of work in the semiconductor industry (which is a rather rare case). Some of my colleagues at the time “returned” due to business separations and acquisitions. With AMD’s acquisition of Xilinx, there will be some of these “going backs” again. However, in Silicon Valley companies, such a situation is a daily occurrence, and heads flocking before and after the announcement whether it will immediately adapt to the new situation, such as “I have come back again …” All you have to do is find another place to work by relying on the introduction from the hunter.
Interestingly, the main product of MMI, the first major acquisition, was PAL (Programmable Array Logic), which was the prototype of FPGA (Field Programmable Gate Array) by Xilinx, which was acquired by AMD this time. However, PAL is a hard wire type (the desired logic is constructed by physically disconnecting the connection), and it was so primitive that it could not be compared with the current highly integrated FPGA based on SRAM. It was very useful from Japanese personal computer makers who designed and manufactured the board in-house. The reason is that there are always bugs in PC boards that are often designed with hardware, and when the bugs cannot be solved by the firmware, they are often solved by biting a small logic chip. The body.
Even if the production start time is approaching, the bug is not solved, and I remember that sudden large orders were received many times, such as when I managed to make it in time by pushing PAL, which is a small logic chip, into the board. There is. However, AMD was unable to incorporate MMI’s technology into its mainstream business and eventually spun off, creating Vantis, which later disappeared and Xilinx has many people migrating in this way. .. At that time, I think few people foresaw that FPGAs would be of such value in the future.
NexGen, the second major acquisition, was the driving force behind the brilliant K6 revival of AMD, which was dying due to the failure of the K5 processor. AMD’s current good fight against its nemesis Intel in the x86 market wouldn’t be possible without the acquisition of NexGen and the introduction of the K6. However, when the acquisition of NexGen was announced, the reaction was disappointing, saying, “AMD CEO Sanders got crazy under the pressure of worsening business performance.” And the last major acquisition is the acquisition of Canadian GPU male ATI. As you know, ATI’s graphics technology has become a great weapon for AMD today to compete with NVIDIA. I think this was also a significant acquisition that determined AMD’s success today. These historic events often prove their meaning well afterwards.
The key to a successful acquisition of a semiconductor company is the fusion of timing and corporate culture
I’ve seen many examples of acquisitions in the semiconductor industry, but the key to their success is how well they capture the purpose of the acquisition and how well the cultures of the two companies merge after the acquisition. I think it’s time to let it go. AMD’s acquisition of ATI was so expensive that the industry’s reputation at the time of the acquisition was by no means good. However, looking back, it seems that the decision was very reasonable and timely.
Although it was effectively a corporate acquisition, AMD’s internal organizational integration team never used the word “acquisition” to call it “Merger” and “OGT: One Great Team”. Under the slogan, he was very careful to give employees of both companies a common motivation.
In the comments after AMD’s acquisition of Xilinx, it was impressive that the CEOs of both companies, Lisa Su and Victor Peng, constantly emphasized “the affinity between the two companies’ corporate cultures.”
There are various opinions about the purpose and timing, but with the concentration of intelligence on the edge node side and the data center side, the strategy that focuses on the data center is quite a lot in the semiconductor market where the market is expanding through IoT. I think it’s a hit. Management leadership, including the CEO, is responsible for future success. The highlight of the future is whether the integration of the two companies will create great added value.
Pros and cons of Xilinx acquisition and future whereabouts
Market reactions to large acquisitions that have reached $ 35 billion vary, including:
- There is a limit to how general-purpose CPUs and GPUs can meet the demands of data centers. FPGA-based accelerators are extremely effective in meeting the demands of diversifying data center efficiency, and are essential for quickly attracting large customers. In that respect, this combination of AMD and Xilinx is ideal.
- AMD’s financial position will be greatly improved by the financially sound acquisition of Xilinx, so the acquisition effect can be expected in the short term. The merger of the two companies will dramatically increase total sales and provide the economic scale benefits needed for further growth in the future.
- The customer / application bases of both companies are non-overlapping and very complementary (as Mr. Ohara mentioned in the article above as a “reasonable acquisition”) and have been for AMD so far. Get tickets to future growth markets such as telecom / network, autonomous driving, and healthcare that you couldn’t get into.
- I understand the importance of data centers, but the acquisition cost is too high. It is very difficult to create added value that is commensurate with the high acquisition cost in the future.
- AMD is currently very successful in the x86 market against Intel. However, Intel will surely come back, so the 18-month integration period until the acquisition is completed is a dangerous time because AMD’s selection and concentration strategy is blurred. Intel, which previously incorporated FPGAs in Altera’s acquisition, hasn’t been able to create synergies (some disagree that it’s “superbly synergistic”), and AMD succeeded in doing the same. I can’t think of that.
- The outlook for the global semiconductor market is quite uncertain due to fierce competition for technological hegemony between the United States and China. Large-scale acquisitions in this situation carry great risks.
Pros and cons It seems to be a valid point, but semiconductors are, in a bad word, “the world of gambling.” In any case, the acquisition has already been announced and the two companies will have a lot of work to do, including approval of antitrust authorities, consolidation of organizations, and the creation of a new roadmap after the acquisition. AMD, which is teaming up with the Xilinx team under the leadership of Lisa Su, will see what happens in the future, not only for me, who is an alumnus of AMD, but for the entire industry as well.
Yoshikawa tomorrow theory
Click here for a list of articles by this author