Over the weekend it seems that Softbank have closed a deal to acquire ARM Holdings (what we in the industry simply think of as ARM) for an all-cash deal for around $32B. The deal is expected to be formally announced on Monday but since the Wall Street Journal, the Financial Times and the New York Times are all carrying the story, it seems that it is not just an unconfirmed rumor. ARM's shares are up 18% this year although a lot of that is simply the weakness of the British pound after the Brexit vote to leave the European Union. The current market cap is around $22B so the acquisition price is a premium of almost 50%. Softbank is a Japanese company headquartered in Tokyo. In the US they are probably most well-know as owning a controlling interest in Sprint. Softbank is currently the 3rd largest public company in Japan (behind Toyota and Mitsubishi) even though you have barely heard of it. Softbank acquired nearly 80% of Sprint for $22B in 2013. But it is not exactly the jewel in the crown. It is the #4 player in a 4 player market, never a great position to be in. It has lost money for 7 out of the last 7 years. It has a huge amount of debt in an industry where you need a lot of capital to build out the next generation of technology. Industries built on the base of semiconductors rarely have cash-cows since newer cheaper technology are even more attractive. So softbank have been mortgaging assets like basestations, and cutting costs. But that is rarely the recipe for success in a technology based business. Softbank's founder, Makayoshi Son had announced his intention to leave the company, but a month ago he announced that he would stay. The big challenge is to turn Sprint around and make it cash-flow positive. I am not sure why they are buying ARM, I'm not privy to any internal discussions, obviously. Business analysts who don't know anything about the semiconductor industry (calling ARM a "major chip firm" is a good symptom) just see Sprint is in mobile, ARM processors are in every mobile phone: synergy. But I don't see any synergy between a mobile operator and an IP supplier to mobile. ARM is a good business, with potential opportunities in other major markets than mobile (like servers), but whether it justifies a 40% premium on its current valuation is less obvious. There are also threats. I don't think any other proprietary computer architecture such as MIPS is a true threat, but the almost overnight success of RISC-V could turn out to be a threat in the same way as Linux was a threat to Windows (not in the sense that Windows went away, but all those datacenters run Linux not Windows). Here is the Wall Street Journal's take: The deal, which is expected to be announced Monday, will mark a deeper push for SoftBank into the mobile internet where ARM has a strong market share in providing chips connecting mobile devices, according to the person. Err...sorry. ARM does not make "chips connecting mobile devices." That would be Qualcomm or Mediatek mainly. Or the smartphone makers themselves: Samsung, Apple and Huawei/Hisilicon all make their own chips. To be fair, they are all ARM-based, but there is a big difference between supplying microprocessor IP and supplying chips. Nothing directly to do with ARM, but an interesting factlet about Softbank and Japan. You probably know that Japan is dying out since their fertility rate is about 1.4 children per woman (actually up from 1.26 in 2005) and basically no immigration. Softbank offers bonuses to employees just to have children. Around $400 for a first child but $40,000 for a fifth. That's a big number. Germany's fertility rate has dropped even lower than Japan's, which has led to seeing immigration as the way to fill the gap. This is not a political blog so I won't attempt to argue whether I think this is a good thing or a bad thing, but it is certainly one of the things that put Brexit over the top, drove down the pound, and made British companies like ARM more attractive for acquisition. Anyone who immigrates to Germany has a right to live in Britain (which is the size of New Mexico, roughly, with a population of 65M, about 10 times the actual population of NM, or roughly twice the population of California or Texas). Even lower fertility rates are South Korea (1.19), Singapore (1.19) and Hong Kong (1.12). Previous: The Future of Neural Networks...and Our Robot Overlords![]()
![]()
↧