With several product misfires and a shrinking market share to contend with, it came as a disappointment but no horrific surprise when one of my industry contacts let me know that Conexant had abruptly decided to take its wireless products group off life support -- just in time for its quarterly financial report. I’m not sure how many of the original Harris team that pioneered the development of inexpensive, reliable wireless data still remain, but it’s sad to see such a collection of talent, decades-long in its assembly, dismantled and tossed on the trash heap because of the short attention span of upper-level management.
Once a flagship group of Harris’ Semiconductor Division, the wireless team managed to import a good chunk of spread-spectrum knowhow from the dark corners of the company’s military radio group and apply it to produce some of the best early 802.11 silicon on the market. Back in the mid-1990s, I spent some time with those folks and am still impressed at the many tricks they used to make the relatively humble silicon processes of the time deliver such good range and the ability to maintain good throughput under less-than-ideal conditions.
The wireless group followed the semiconductor division when it was spun-off and re-branded as Intersil, where a series of problems seemed to keep their radios a bit less integrated and many months behind the highly-aggressive offerings from the likes of Atheros, Broadcom, and Marvell. The difficulties seemed to be a combination of inconsistent direction and consistent under-funding, compounded by yield problems that, according to reliable sources, plagued most of the products coming out of the Intersil radio group. I’m not enough of an RF silicon expert to make any judgments here, but one could speculate that some of the design techniques they used to get such high-performing chips also limited their ability to tolerate even moderately mild process variations.
Nevertheless, the basic radio technology in many of their products was very sound and did not rely on the digital signal processing sleight-of-hand that several other companies use to mask the fair-to-middling RF performance in their radio subsystems. I’d like to imagine that with proper leadership and sufficient investment they’d have been able to fix the yield problems and modernize the digital end of the chip sets. Whatever the cause, I’m sorry it didn’t happen, because it might have allowed them to lead the market in performance at a time when other companies were so focused on slashing BOM costs that many of their products were having trouble reaching the far end of a small home.
Unfortunately, this was not meant to be, and the wireless team was eventually sold off to Conexant.
In many ways this was a very good match-up which complemented Conexant’s broadband gateways and other networking products...but the company’s other business challenges, its heavy debt load, and Byzantine internal machinations (enough for another Editorial, or maybe even a book) seemed to keep them from supporting the wireless products team in the aggressive manner it would have needed to compete effectively with the now-firmly-entrenched Brands A, B, and M. Nevertheless, the group’s sudden demise was a bit of a surprise, because they had just rolled out their first 802.11n product in mid-September, a device they claimed to be the lowest-power product on the market.
The little bits of information I’ve been able to pick up indicate that the employees were aware of the precarious situation and had been trying to organize an effort to buy out the group and create a new company, but some on-again-off-again negotiations with Qualcomm to buy the group kept management from taking the option seriously. My sources tell me that Qualcomm offered Conexant $180M for the wireless group a year ago, but their former CEO (and major shareholder) Dwight Decker may have pushed for too much (some speculate he wanted a seat on the Qualcomm board) and the deal cooled. This year, Qualcomm offered $30M but Conexant couldn't make the deal stick -- there is conjecture that it was scotched because they wanted to include some of their India operations in the mix. From what I can tell, the final decision was made suddenly, in part perhaps because the current CEO, Dan Artusi, wanted things wrapped up before their November 1, 2007 quarterly conference call.
Regardless of the reasons, another important chunk of this country’s top RF engineering talent, many years in the making, has been cut adrift and will likely be scattered to the four winds in a matter of a few days. Hopefully, there is some truth in the rumors that a Conexant exec is quietly working to assemble the funding it will take to keep the group together as a design and service company. If they do, I suspect they will have no shortage of customers. Meanwhile, I’ll keep my fingers crossed for the many talented folks who found themselves in a tough situation that never should have happened.
Comments? Questions? Job leads for the Conexant wireless team? Write me at LHG at EN-Genius dot net or post your comments on our blog.
