Someone asked me the other day, "How much venture capital has been invested in network processors?" My research turned up more than $500 million in disclosed investments into network processor startups since 1999. Toss in switch fabrics and coprocessors, and the total exceeds $1 billion.
That's bad news for investors, who are unlikely to get a return on this investment. But it's good news for OEMs, which can take advantage of this R&D at no cost.
The network processor market is small today but is expected to grow rapidly, reaching $1 billion or so by 2005. That might justify an investment of half a billion, if all of that revenue goes to the startups.
The problem is that most of the revenue is likely to be earned by Intel, AMCC and the other big companies that dominate the NPU market today. These companies are devoting hundreds of engineers to the sector to maintain their leadership as the market grows.
With only table scraps left for the startups, venture capitalists will mainly be disappointed. A few may get lucky if one or two startups do well or get acquired for a reasonable price, but that will be the exception.
The switch-fabric space is also hot, attracting more than $200 million in funding. Again, big companies such as IBM and Vitesse dominate today, making it difficult for startups to force their way into the market.
Storage processors are the next big thing. The companies building them have been more secretive than their NPU brethren, but I believe more than $100 million has been invested in the segment. None of the big networking vendors has a product in this market yet, leaving it wide open for startups.
If you are a networking chip maker, all of this means that far too many companies have been funded in these markets. The competition is choking out the weaker companies-such as Clearwater, Power X and Terago-which continue to run out of cash and then collapse.
But with so much investment, it will take a while for enough companies to exhaust their resources. In the meantime, OEMs will reap the benefits of overgrowth in product development.
With so many vendors, there are many products to choose from. There is one to suit to almost any OEM need. And competition is driving down chip prices. OEMs should buy now, before the market consolidates.
Linley Gwennap is founder and principal analyst of the Linley Group (www.linleygroup.com/npu).