Recently I had a chance to make a presentation at the fourth annual Semiconductor Venture Fair in Redwood City, California. I've done this for two years in a row, and it is interesting to revisit a forecast annually to see how things have changed. In this presentation, I went over my views of what is hot in the industry right now, and what is cooling off. Here is an abbreviated version of the key points:

What's Hot:

  • Cell Phone as a platform
  • Cameras everywhere
  • Wireless (anything)
  • MEMS
  • Flash Memory & alternatives
  • Security (many forms)
  • Nano [mostly anything]
  • Virtualization
  • China & India
  • M&A Exits (where are the good old days of IPOs?)
What's Warm:
  • Optical Infrastructure components (making its way back)

What's Not:

  • MHz alone as a performance benchmark
  • Stand-alone PDAs
  • Optical Infrastructure components
  • ASICs and Custom Silicon
  • Digital-only SOCs
  • CCD Imager Chips
  • Network Processors
  • Infiniband
  • IPO Exits

Some Specific Examples

  • Cell Phone as a platform: As the mobile infrastructure offers faster data rates, the cell phone will emerge as the primary way that people will get to the internet. From discussions with coworkers about the way people use phones in China, this is already the primary way people think of for getting to the internet (instead of using a PC). More and more, the phone platform will offer integrated tunes, video capture, television playback and even financial services. It is truly the most personal of computer and communications devices
  • Cameras everywhere: Post 9/11, everyone is worried about security. Maybe they can't monitor everything in real time, but they can record videos and look for clues to what happened after the fact. If you look around and take notice of the number of cameras that have popped up, you will be surprised. They are everywhere you go. Not only for security monitoring, but also in places that are useful - like in your car for backing up and blind-spot elimination. I think we've only seen the tip of the iceberg in the camera market.
  • FPGAs & MEMS: This is a collection of technologies that don't necessarily go together, but they are both ones that are going to be getting a lot more attention in the near future. First, the FPGA as we know it will continue to take business away from the bottom end of the ASIC market. The next thing I'm expecting is for a class of more coarse-grained FPGAs to be created that have less flexibility to fit any application, but are more efficient than the current 5% (transistors actually used for logic computation vs. overhead).
  • Wireless (anything): This seems to be the year embedded wireless devices. Stand-alone PDAs are pretty much a thing of the past - the only issue now is how many bands do they support (Bluetooth 802.11 or 802.11 and GSM/GPRS). Also in the low-end of devices, smart sensors are starting to be wirelessly connected. The ratification of the IEEE 802.15.4 standard and the higher layer networking standard called Zigbee (tm) has created a number of suppliers that are all offering low cost medium performance embedded wireless devices.

  • Flash Memory & alternatives: The flash memory market is almost as big as the DRAM market, not by units, but by revenue. It is a pretty exciting time with all sorts of new applications for embedded storage. A new class of video cameras is showing up that does away with the tape mechanism and simply records to a flash card (or in some cases, a CF-format HDD). Unfortunately, as technology scales to smaller geometries, it gets harder to make flash memory cells. This will ultimately create an opportunity for new non-volatile storage technologies. There are lots of efforts going on, everything from FERAMs to Carbon NanoTubes. This field is due for a major technology shift in the next couple of years.

  • Security (many forms): Suddenly the unfiltered internet has become unusable. From spam, viruses to spyware and pfishing scams there has never been more of an assault on the small business and home users of computers (larger organizations at least have IT departments to deal with all of this nonsense). Look for biometrics to become a big thing in the near future. The finger-print imaging companies suddenly look like they are doing very well. The software layers above that will be the next great opportunity.

  • Nano [anything]: Maybe it's just a naming convention, but suddenly everything has a nano label put on it. Maybe because some of the US Government types said that anything measured in nano-meters should qualify (like the entire semiconductor industry). With that said, there are some truly interesting things going on in that space. Sorting the commercial opportunities from the science-fair projects is going to be a chore.

  • RFID Wireless Sensors: The big buzz around at the moment is RFID and the more general category of active sensors. The RFID experiments have been driven by a couple of high profile users - Wal-Mart and the Department of Defense. It will take a couple of years for them to work out the bugs, but ultimately it will become an interesting opportunity. It's a shame that a number of start-up companies are all working hard on something that will ultimately sell for $0.05 each. The wireless (anything) comments above cover the wireless sensor space.

  • China & India: Outsourcing has finally moved to beyond just menial tasks and now is clearly on the path to taking a bite out of the US brain-trust. Many smart and motivated ex-patriots are moving back to their home countries. This is going to ultimately catch up with us. Hopefully the US will be able to avoid what happened to England in the early 1900's.

  • M&A Exits: IPO's just aren't what they used to be, not that they are bad, there just aren't enough opportunities for them right now. As a result, start-up companies need to be thinking the most likely way to liquidity is a Merger or an Acquisition, rather than a great IPO. The financial returns traditionally are about 3x less for a M&A rather than an IPO, but that's not always bad.
  • MHz: Processor speed alone is no longer the figure of merit fo