The word platform gets used often in this business. It is a heavily loaded term, which gets bandied about a lot, up there with ‘cloud’ and ‘open’ in terms of repeated, heavy usage. Despite this linguistic abuse, there is still a lot of value in having an actual platform. Businesses seek to build platforms to create some form of lock-in. Use one platform and it can become hard to move off it, creating repeat business for its owner. A platform can block out competitors, bind customers in and create valuable partnership opportunities.
In the mobile industry, the most widely known platforms are probably the operating systems for smartphones, like iOS and Android, where we commonly speak of ‘platform wars’. However, we think there is another layer of platforms a bit further down the stack that is equally if not more important, the mobile baseband.
Basebands are to mobile phones what CPUs were to PCs, they are the key piece of silicon that end up driving most of the other hardware choices for a device. Technically speaking, a baseband is a modem and controls the communications between the phone and the carriers’ base stations. However, baseband vendors have pursued an integration strategy for many years. The end result of this is that when phone makers buy a baseband today they are also typically buying that baseband, the radio transceiver and increasingly the applications and graphics processor for their device as well.
The baseband vendors have created a platform from this one product. These vendors now commonly provide tools, testing and help with carrier certification. The transition from a product business to a platform business took place about ten years ago in mobile, largely coincident with the rise of 3G networks. It was this ‘platformization’ of the business that drove Texas Instruments, once the king of basebands, out of the market. Today, the best known baseband vendor is Qualcomm, and they designed and then reaped the benefits of a platform approach to handset silicon.
One of the key elements of building a platform is to include services that would otherwise have to be provided by someone else. This tends to transfer value and margin dollars from one segment to another. Again, the platformization of handsets makes a good example. Prior to Qualcomm’s arrival, much of the software that went into basebands (aka the protocol stack) was written by the handset vendors themselves. This required hundreds if not thousands of engineers, and hundreds of millions of dollars in annual expense. Qualcomm did much of this software work, freeing up its handset customers to deploy capital elsewhere. And went on to become the leading handset baseband vendor by a wide margin for 3G phones.
However, the transiton to platforms continues. And while Qualcomm is the best known baseband vendor they may not be the largest anymore. That title arguably goes to Mediatek of Taiwan. Mediatek took the platform approach one step further and in doing so created one of the most important forces in mobile today.
Mediatek grew up in the one of the most competitive industrial landscapes there is – the consumer electronics supply chain in Taiwan and China. They sold chips for DVD players and optical drives for many years to low-margin assemblers of PCs and other electronics. Through some trial and error, clever thinking and luck they created a new business model for the cell phone industry. They recognized that their traditional customers had very limited engineering talent. These were typically companies with a few assembly lines capable of putting chips on a board and wrapping that board in plastic. None of these companies had the ability to design cellular phone software or build pretty user interfaces. These companies relied on their chip suppliers to provide basic software like device drivers and user interfaces.
So Mediatek took all that software work and bundled it into a complete package which they called a reference design for mobile phones. These reference designs were essentially blueprints for building a basic, 2G feature phone. Anyone with an assembly line could buy a Mediatek chip and this would come with everything they needed to build a phone. Mediatek actually went a step further than Qualcomm. At the time, in the early 2000’s, most of Qualcomm’s customers were still large handset makers who had the ability and desire to build some of their own pieces of software. Companies like Motorola no longer had to build the protocol stack for their basebands, but they still wanted to customize each device in certain ways. Mediatek’s customers did not even want this much. Mediatek added device drivers, suggested specific components parts and laid this all out. Over time, they also added a huge range of software options that gave their growing customer base some ability to customize phones in certain ways like local languages, color screens, Java licenses or Bluetooth. This was far less customizaiton than the big handset vendors could produce but required almost no customer engineering.
Mediatek was just looking for a new product to sell to existing customers, but they opened the door for all these small assemblers to begin selling inexpensive phones. This group came to be called the ‘shanzhai’ or ‘grey market’ or ‘white box’ handset supply chain. Today, we call them branded Chinese handset makers, and they contribute over half of the phones sold each year.
We have greatly simplified the history here but we would refer you to the works of Professor Willy C. Shih of Harvard Business school who has written extensively on the Mediatek phenomenon.
In our experience, most people in the US and Europe still underestimate the size of the Mediatek ecosystem. By our latest estimate, the major third party analytics firms still undercount Mediatek and its competitors by 300 million to 400 million units a year. This is a big number to miss, but it is hard to get an exact count of Mediatek’s customers as the ease-of-use means that the barriers to entry are very low, and there are ample new entrants (and exits) all the time.
Mediatek has not enjoyed this opportunity all alone. In China and Taiwan new competitors have popped up, notably Shanghai-based Spreadtrum and RDA Micro and Taiwan-based M-Star (which is now being acquired by Meditaek). Nor has this lesson been lost on Qualcomm which is actively building its own presence in this market with its own set of Qualcomm Reference Designs (QDR). Let us know if you would like to know more about the competitive dynamics of this market, but that subject is beyond our scope here.
There are three important implications of Mediatek’s business model.
First, they demonstrate that handset vendors do not have to own their own silicon. In the US, the success of Apple and its own A5 line of applications processors has created the impression that successful handset companies need to design their own chips. Apple’s A5 has all kinds of iOS specific functions burned into silicon which gives the iPhone certain performance and probably cost advantages. But Apple is, as always, a special case, few other vendors have the scale, or the margins, to do this. There are still thousands of Mediatek customers making a good business selling phones using off-the-shelf silicon.
Second, the handset business is now, more than ever, a race between time and innovation. The major handset vendors, other than Apple and Samsung, are generally lacking in profits. There is a yawning low-end tier of handsets. Companies that cannot innovate in some way will be relegated to competing with the Mediatek ecosystem, competitors who have very low labor costs and limited R&D budgets. If you think your business is competitive, imagine competing in a market where the average selling price for a feature phone is $20, retail. This is particularly important now because Mediatek and Spreadtrum both began selling Android reference designs last year. We have seen a corresponding drop in handset prices in many markets and a surge in smartphone adoption globally. This is great for consumers but very challenging for handset vendors. And this is by no means a China-only or emerging markets phenomenon. Mediatek-powered phones are now common in Europe and steadily creeping into the US.
Third, Mediatek’s adoption of Android should be seen as a key shift in the industry’s balance of power. For years, one of the key attractions of a Mediatek reference design were its software and UI capabilities. This brought Java into many new phones, and is probably still a decent piece of licensing business for Oracle’s Sun unit. Mediatek and Spreadtrum adoption of Android gives that OS a virtual lock on the China handset ecosystem. We have witnessed first hand how other OS offerings have dropped this particular ball and now have little hope of fighting their way in. We detailed this effect in our blog recently highlighting the growth of low-end Android devices.
In our view, the volume that the China handset ecosystem now brings makes it a hugely important bloc in determining other, complementary segments of the industry. Mediatek and Spreadtrum, via their reference designs, wield the vote for that entire bloc in making software decisions going forward. This gives them considerable weight in the industry, but weight which has yet to be fully felt by industry players in the US and Europe.
Mediatek’s rise makes for an important study in how power and economics are shifting in the mobile industry. A lot can still change here. Qualcomm looks set to give Mediatek a serious run for the money. More handset vendors, notably Huawei and Samsung, are looking to build their own silicon and may yet succeed. Nonetheless, Mediatek and the whole reference design model is still in early days of reshaping the mobile industry.