Blog Post

Chips Work Hard for the Money, Just Like Everyone Else

Many venture firms shy away from investing in semiconductor startups because of the inordinately high cost and risk involved relative to other sectors, but some firms are still determinedly pursuing silicon by backing wireless chips containing radios that can work on multiple types of networks. The trend is being fueled by a robust market for wireless devices, as well as the increasing number of wireless networks available for specific purposes.

The iPhone is a good example of a wireless device that utilizes several special purpose chips, with more than 19 in total and at least five radios, which send and receive the wireless signals that allow a phone to communicate with a variety of networks, from Wi-Fi for data to GPS for location information. Even connecting to the cellular network requires more than one radio, be it to handle older 2G networks or the latest 3G or 3.5G options. For companies designing handsets to run across several networks, the sheer number of them and their frequency bands means adding more chips can get complicated — and expensive.

So if a company can use software to combine multiple radios that talk to a variety of wireless networks on a chip, a handset maker could, in turn, put one chip in a phone and rather than have the phone optimized for certain networks, ship it anywhere in the world, optimizing simplicity and minimizing cost. After all, consumers buying wireless Internet devices (such as, say, the Kindle) aren’t interested in which network it runs on, only that their device is working.

That’s one of the reasons BitWave Semiconductor, which received $10 million in second-round funding this month from Apex Venture Partners, TVM Capital and ECentury Capital Partners, was so attractive. The company expects to ship a programmable multimode transceiver (which is one small part of the overall radio) this month. Another programmable multimode transceiver startup, Sirific, was purchased this spring by venture-backed Icera, which is building a software-based baseband chip for mobile devices.

Others are not building programmable chips, but are focused on multimode chips that use additional hardware to work on different networks. Wavesat, which has raised an undisclosed amount from Skypoint Capital, BDC Venture Capital, Innovatech Montreal, Monet Capital and Sunsino Ventures Group, is making a WiMAX and LTE chip that could fit into a wide variety of devices. Other startups raising capital include Altair Semiconductor, which raised $18 million last December from Bessemer Venture Partners, BRM Capital, Giza Venture Capital and Jerusalem Venture Partners for a total of $26 million. Altair offers a low-power WiMAX chip, but plans to add LTE capabilities for a dual-mode chip within the next year or two.

“I believe there’s a long-term secular trend toward handsets doing more — not just calling, but music, video and data with more and more functionality thrown on them,” Wayne Boulais, a general partner with Apex Ventures and an investor in BitWave, says. At the same time, these are pricey components for a handset maker to buy, so integrated radios make it cheaper. For BitWave and other firms building such multitasking chips, using a programmable approach makes the high cost of putting money in silicon slightly less risky for the investor.

“It can take a long time to create a design that works, and in an investment area where the costs are at the front end to make them, being able to design into multiple different applications mitigates the risk that the market may end up moving on you once you get the hardware out,” Boulais says.

Major vendors are doing this as well, with Broadcom and NXP putting as many radios on a chip as possible and other vendors trying to at least create a smaller package of multiple chips. It’s also one of the motivations behind the proposed joint venture between STMicroelectronics and Ericsson’s mobile platform group announced this month.

Not all multimode software radios will land in devices. Vanu, a Boston-based company, is building a software-defined radio that could reside in a base station on a cell tower, making it possible for one base station to serve several types of cellular networks. Vanu raised $32 million this week from Norwest Venture Partners, Tata Capital and Charles River Ventures. The round follows a $9 million first round last summer.

As consumers, we may not care how many chips our cell phones pack unless we lose out ability to access the web or get our location, but these trends playing out in silicon should eventually get us to cheaper phones with more features. Multitasking can be a good thing.

photo of the iPhone board courtesy of Portelligent

This article also appeared in

4 Responses to “Chips Work Hard for the Money, Just Like Everyone Else”

  1. I would like to add that multi-mode radio ICs is an area that Kaben Wireless Silicon has pursued in close cooperation with customers for years without burning tons of VC dollars.

    Kaben’s unique radio architecture enables reconfigurable transceiver front ends that can support multiple standards. This innovative technology occurs in the Intermediate Frequency (IF) range between the RF and the digital.

    The Reconfigurable IF is built around three novel circuit cores; a Sampling IF Filter which down-converts an IF directly to digital, a Digital-to-IF up-converter which converts a digital baseband directly to IF, and an ultra-low phase noise and spur level synthesizer that is used in the receive and transmit paths.

    The field programmable flexibility of the Reconfigurable IF provides a multi-standard transceiver capability while relaxing the associated component specifications in the up- and down-converters, LNA, and external PA. The Reconfigurable IF also removes the need for external SAW filters in the receive path, while providing all the strengths of a classic super heterodyne architecture. When using a super heterodyne approach, the Reconfigurable IF eliminates the analog baseband normally required by converting directly between IF and digital. For the receiver, this avoids DC offset and IP2 problems, and removes the need for compensation loops in the digital back-end. The elimination of the analog baseband also removes issues associated with 1/f noise.

    The combination of field programmable flexibility and lower component count, results in a quick time-to-market, smaller die size and a reduced BOM.