Since 1999, venture capitalists had poured more than $300 million into Procket. In 2000, the start-up was one of the most highly valued companies in Silicon Valley. But as of Thursday, when the Cisco deal was announced, Procket had been in financial trouble for some time and had been rumored to be up for sale. Last week, Procket put itself on the block. For Cisco to get such a good price for Procket's technology and be able to keep it out of the hands of competitors is likely well worth the cash spent, say many analysts.
But Procket, which makes routers that directly compete against Cisco's own product line, is an unusual acquisition target for the company. Though Cisco was an early investor in the company, executives there have said publicly on several occasions that Cisco does not buy companies with competing technologies. Cisco also has a long-standing policy of not buying rivals just to keep the technology out of other competitors' hands.
Since 2000, a year in which it bought 23 companies, the networking giant has slowed considerably its rate of acquisitions.
- Metro networks company AuroraNetics
- VPN services company Allegro Networks
- IP networking company Hammerhead Networks
- Navarro Networks, an Ethernet components maker
- AYR Networks, which focused on distributed architectures and integration
- Storage switch maker Andiamo Systems
- Network security specialist Psionic Software
- Security software maker Okena
- SignalWorks, specialist in Internet phone technology
- Linksys, maker of networking gear for consumers
- Voice technology specialist Latitude Communications
- Riverhead Networks, maker of intrusion protection hardware
- Security start-up Twingo Systems
At first glance, it appears Cisco has broken both rules. But executives said the main reason for buying the start-up was the 130 engineers Cisco will hire as part of the deal, many of whom have worked for Cisco in the past.
"The key reason for the purchase is because of the people," said Mike Volpi, senior vice president of Cisco's Routing Technology Group. "We will be getting 130 very talented people. We felt that it was worth us buying the company to keep the team together, rather than try to pick them off one by one."
Volpi said Cisco has no intention of selling Procket's routers, which vie with its newly announced CRS-1 core router. Instead, Cisco plans to integrate pieces of Procket's software design into the CRS-1 product family.
Reading between the lines, Cisco's purchase of Procket's assets indicates that Cisco may not be as happy with its new router as it has publicly stated. At an investor conference Tuesday, Cisco CEO John Chambers said the company was satisfied with its routing strategy and would only buy a competing company to acquire talent. Volpi reiterated Cisco's commitment to the CRS-1.
"This acquisition says nothing about the CRS-1 product," Volpi said. "We are extremely confident with the product as it is. We plan to ship the product in July as we had mentioned before, and there will be no features from the Procket acquisition integrated in the product."
Cisco introduced the CRS-1 only last month, after four years and $500 million spent developing it. The product, which is starting beta tests with customers now, probably won't start generating revenue for the company until next year. Cisco developed brand-new software for the product and also worked with IBM to develop a new chipset.
Since the CRS-1's release, critics have been quick to point out its shortcomings. Tony Li, a co-founder of Procket, told CNET News.com in a recent interview that the CRS-1 was inadequate for customers needing a scalable router.
"Although it's significantly bigger than I had originally thought, it should still be an order of magnitude larger," said Li. "Customers that have vision and understand the growth of the Internet will probably be disappointed with this product."
Li worked at Cisco when it developed its flagship product, the GSR 12000. He also helped develop Juniper Networks' first core router. He left Procket earlier this year.
Before the CRS-1 was introduced, Cisco's core router offering was looking outdated compared with offerings from Juniper, Cisco's main competitor. Cisco introduced the GSR 12800 in December 2003. This latest incarnation of the company's long-lived GSR family of products started shipping to customers in the first quarter of 2004. Though the product quadruples performance on the GSR product line, it still uses all of Cisco's older technology, including its hardware and IOS software.
Some analysts believed Cisco would use the Procket technology to help revamp the GSR product line. For the past few years, Cisco has taken each version of the GSR, including the GSR 12000 and the GSR 12400, and added new software features so it can be used on the edge of the carrier network to aggregate lower-speed traffic coming into the core.
Analysts thought Cisco might be able to leverage its software and sophisticated chip design to revamp the GSR. From the beginning, Procket's strategy was to build an entire portfolio of routers--from smaller edge routers used to aggregate lower-speed traffic to ultra high-end routers used to shunt large volumes of traffic across the Internet--all based on the same chipsets and software technology.
But Volpi said Procket's chipsets are useless to Cisco since they're built on a different architecture. Procket uses a shared-memory architecture, while Cisco has built its GSR and CRS routers on a centralized switch fabric architecture. He said Procket's software is also not well suited for any other routing products other than the CRS-1.
The first version of the CRS-1 will not use any Procket technology, Volpi said. But it could be used in future product releases. Some suspect the Procket technology will be used to build a scaled-down version of the CRS-1. Juniper took this strategy with the T-series by introducing the T-320. Avici Systems, which produces another scalable core router, has also introduced smaller versions of its products.
"The CRS is good for what it is," said Stephen Kamman, an analyst with CIBC World Markets. "But it's not a great platform to scale down to a smaller form factor. This is where Cisco will likely get a volume of sales."
If what Cisco claims is true, that there isn't an immediate need for the Procket technology, Cisco will spend roughly $685,000 per engineer. The success of the acquisition will be judged by how well the company will be able to keep the team together. Though he would not provide details, Volpi said the engineers had noncompete clauses in their contracts for an unspecified amount of time and were being offered retention bonus packages that will include a mix of cash and Cisco stock.
Procket's CEO Roland Acra, who left Cisco for the Procket job just six months ago, said the Cisco acquisition is a good deal for most of the employees.
"The real sad outcome would have been if the company went belly up," he said. "There is no appetite right now in the VC community for funding another core routing start-up. And I don't think there will be one for at least a decade. Cisco has a similar culture to Procket, and it gives people who are passionate about technology a place where they can contribute."