Last week’s announcement that the IP behind Meshnetics’ ZigBee stack is being acquired by Atmel underlines the continuing consolidation of the short range wireless industry. Since the boom in short range wireless that was started by Bluetooth and Wi-Fi there has been a growing number of VC funded silicon and stack companies entering this market space. It has been obvious for some time that the number of companies is not sustainable and that at some point the bubble would burst. The sale of Zensys to Sigma heralded the start of the process. 2009 will be the year when momentum builds and a lot more wireless dreams hit the buffers.
Funding semiconductor start-ups is always a gamble, especially when the companies are aiming to support industry standards. Because multiple companies will target the same standard, the basic fact of life is that most will fail to gain market share. That’s an inevitable consequence of the cost structure for wireless chip designers; the expense of designing a chip means that they need to sell millions to survive. The general view is that the cost of designing a complex wireless chipset ranges from around $3 million to $10 million. The higher end of that typically includes application processors and embedded protocol stacks. That cost can rise even higher if there are multiple radios in the chip. At the same time, the manufacturers using these devices are pushing to get prices well below the $5 price point. In the case of Bluetooth, where over 1 billion chips are sold each year, the cost is now under $2 per chip.
With these price points, companies need to sell a minimum of ten to twenty million chips to cover their development costs. Those that do will make money and gain a significant market share. Those who don’t will run out of cash.
There’s a second problem for start-ups if there’s already a dominant player or technology. If the market price perception is $5 for a chipset, then customers will expect new entrants to match or improve on that price point. As all of the new start-ups are fabless, to achieve that they need to be making tens of millions of chips in order to negotiate a competitive price with the wafer fab. But unless they have a spectacular product win with a high volume customer, they’re unlikely to manage that. That takes them into the Catch 22 position of having to subsidise their chip price to be competitive, which means they’re losing money on each sale. It’s a gamble that they have to take to gain market share, but it burns cash. In times when money is in ready supply, that may not be a problem. Today it’s a Catch 22 that will leave many with only two options – bankruptcy or a fire sale.
Most semiconductor start-ups expect to be sold. That’s why the VCs invest in them. Typically they’re nimbler than larger companies, who buy them to acquire their technology. That happened with Bluetooth, where billions were spent by the big boys in acquiring smaller start-ups. Who remembers Brightcom, Mobilian, Philsar, Silicon Wave or Zeevo? They’ve been subsumed, but some of their technology can still be discerned in today’s offerings from the wining companies. Today, when the next fund raising round for today’s start-ups becomes difficult, the larger chip companies will no longer look like affluent suitors, but will be waiting in the wings like hungry wolves, ready for a free meal when the money runs out.
Bluetooth had its shakeout when the market was high. Wi-Fi did similarly, although the slower growth and constant evolution of that specification has attracted a new generation of low power niche developers, including Gainspan, G2, Ozmo Devices, Nanoradios, Redpine Signals and ZeroG Wireless. Some of whom will inevitably will feel the cold breeze of recession blowing around them this year. But the most vulnerable sector in the short range marketplace is probably the 802.15.4 and ZigBee manufacturers.
Today I suspect that no ZigBee chip company is making money. Going back to the basic principle of development cost, there just isn’t the market size for any of the players to recoup costs. One or two of the leaders may just about get there this year if the market expands. Their case is not helped by the fragmentation of the stacks. Whereas ZigBee was the only game in town for the low power 802.15.4 radio a few years ago, today the industry has fractured, with competing protocol stacks dividing up the market. Foremost amongst these are 6lowPAN, RF4CE, WirelessHART and ISA 100.11a.
This diversification gives chip manufacturers a real headache. Each of the protocol stacks by itself is complex and generally requires significant support to enable customers to produce working interoperable products. Now, instead of having to support a single stack to sell their chips they may need to support five or six. That multiplies the cost. For smaller start-up companies, they have the dilemma of supporting multiple stacks or concentrating on a single wireless protocol, which becomes a smaller portion of the market. This diversification makes it more important for companies and standards to grab a significant application area and dominate it. Unfortunately there don’t appear to be enough of them to go around. Moreover other standards and initiatives, such as Bluetooth low energy and ANT+ are chipping away at some of the more attractive, high volume consumer applications, further compounding the problem. As a result, Atmel’s recent move to try and consolidate to gain share in the ZigBee space will probably not be the last.
Ironically, the one area which is possibly underserved by vendors is the new Bluetooth low energy specification. Perhaps because many thought that it was just a variant of the existing Bluetooth standard instead of realising that it is a completely new radio specification, the only new entrants to this market have been Nordic Semiconductor and EM Microelectronics. Given the potential size of this market, some of the current ZigBee players might do worse than take this opportunity to jump ship. The frying pan may be a safer place than the fire.