Some consider the many of billions of dollars invested in the semiconductor supply chain to be huge bets on yet to be proven technology and future business. Even if you take a strict view of this as simply business it is possible to learn something from gambling.
The Atlantic tells the fascinating story of how Don Johnson took Atlantic City casinos for $15 M playing blackjack. Last year he won $5 M from Borgata in February, $4 M from Caesars in March, and $6 M from Tropicana in April. This wasn’t luck and he wasn’t card counting. How did he do this and how does this connect to semiconductors and Apple?
Mr. Johnson is not simply a skilled blackjack player. Yes, he always made the right strategic decision based upon statistics and probability as he played. But what he has is very specific knowledge of how the gaming industry works and calculating odds gained over a lifetime of working in this industry. He started in horse racing as a jockey and worked his way up to running betting operations and finally managing a casino in Pennsylvania. In the casinos’ desire to land high rolling customers during these recent lean economic times, they often negotiate special deals in terms of rules of play and financial discounts. Mr. Johnson not only capped his downside potential he negotiated the house’s advantage from over 1% to less than 0.25%. He changed the rules before playing to win. As a result, Borgata and Caesars have banned him and Tropicana will only allow him to play under the regular rules. But he truly is laughing all the way to the bank.
At last week’s Semiconductor Equipment and Materials International (SEMI) Silicon Valley Lunch Forum “Manufacturing For Mobile Solutions: The Impact of Mobile Devices on Supply Chain Dynamics“, Tracy Weed (Sr. Director of Strategic Business Development, Synopsys) provided highlights of this year’s Industry Strategy Symposium. As discussed in my coverage and that of others for ISS and more recently in Two Conferences – Two Industries Challenged By Post PC Era, it is clear that mobile devices are the driving force in electronics today. Dean Freeman (VP Research, Gartner) in “The Shift to Mobility” discussed the economic impact and outlook based upon the changes driven by mobile devices enabled by the cloud. In addition he examined some of the challenges and opportunities ahead for semiconductor equipment manufactures based upon advanced process technology nodes, future shift to 450 mm semiconductor wafers, and thru-silicon vias (TSVs). Tom Quan (Director Open Innovation Platform Marketing, Design Technology Platform, TSMC) was the final speaker who discussed their efforts to develop the 28 nm processand the one beyond (20 nm) align with advanced packaging (such as 2.5D using silicon interposers and 3D to enable these cutting edge devices.
All were good speakers providing interesting updates and insights in to the markets and their company’s offerings. However, the presentation that stood out the most when connected with Mr. Johnson’s story is Colin Baldwin’s (Direct of Marketing, Open-Silicon) “The Evolving Ecosystem for Mobile SoC Development”. He did a good job of describing Open-Silicon’s design and engineering offerings tailored for companies building application-specific integrated circuits (ASICs) for mobile applications. He used the term “Design Lite” to describe how companies are now outsourcing major portions of the ASIC design process to companies like his. This is similar to the way product companies have outsourced both the process development and fabrication processes for semiconductors. The reason there is an increasing need for “Design Lite” services is that the IC development process for mobile devices have become exceedingly complex. And the timelines are too long to support the rapid introduction of new mobile products that often require new ASICs.
To support his analysis of the situation that companies are being overwhelmed due to the introduction of new designs and revisions of old designs, he presented this slide which has a tally of new mobile phones released in 2011:
At the high end, Motorola introduced 53 (yes fifty-three!) new phones in 2011, Samsung 25, and Nokia 20. By comparison, Apple only introduced one new phone: the iPhone 4s. In fact, Apple has only introduce one new phone per year since they entered the mobile phone business. Yes, there is a color option (white or black) which is a simple change of the case and there is a memory size option which is achieved by loading a different set of NAND Flash memory parts which are soldered to the printed circuit board (PCB). These options are simple configuration items which require neither a change to the design nor to the ASICs. Wherever possible, the other phone vendors attempt to reuse the ASICs between their models. But this is not always possible especially if the desired functionality differences between models can only be done in hardware (typically within the ASIC) instead of software.
There is very good reason to attempt to reuse the ASIC due to the development cost and time. It has been estimated that the photolithography masks alone used to generate an ASIC at the 45 nm node (the process technology used in the Apple A5 microprocessor in the iPhone 4s) costs $800,000 to $900,000 to produce. And the masks for leading edge 32/28 nm could be as high as $2 M per ASIC design. This is in addition to the estimated $80 M to design a 45 nm part and the even great $130 M for a 32/28 nm part. And as Mr. Badlwin accurately describes in his presentation that product companies have insufficient resources to complete all the designs in the time required by the market. The higher the number of models of phones or other mobile devices, the higher likelihood of requiring multiple ASIC designs.
By simplifying their product offering to one phone per year and sharing their ASICs among multiple product, Apple has greatly reduced their ASIC development costs and resources required. They have greatly increased their risk by having a simplified product line if customers dislike their one offering. However their product refresh rate of one year on phones and roughly eighteen months to two years on other devices minimizes their down side risk of customer rejection.
So, how was Apple able to change the rules to win by building only one phone per year while the competitors are producing six to fifty three? Being a computer manufacturer – with significant ASIC design expertise – along with having retail channels (Apple stores and website) independent of the wireless carriers provided them a significantly different perspective than most of their competitors. Through their retail channels, Apple was able to sell directly to the end customer whereas the other brands have to sell mostly through the carriers. In many ways the carriers are Motorola’s customer more so than the end customer since the carriers often dictate features and functionality of the phones they carry.
Apple also changed the rules via their initial exclusive contract with AT&T Wireless. They had sufficient leverage to have AT&T pay dearly for exclusivity and accept terms that are not their norm. The exact terms of the original exclusivity contracts are unknown. Here, here, and here are a few articles that discuss the contracts. What is known is that there was revenue sharing on the initial iPhones versus a flat subsidy and Apple was given full control over the phone’s software both still atypical behavior from a wireless carrier.
In the end by approaching the market differently than an existing supplier and using their corporate knowledge along with aggressive negotiation, Apple was able to achieve remarkable success with their iPhones. As long as they continue to be successful by building hot products, they will be a market leader with the influence to change the rules. In the long run, to avoid Mr. Johnson’s fate Apple will need to know when to change strategy since they cannot simply walk away with their money.