Intel chips away at networking

Intel is dipping into petty cash to buy another building block for the Internet Economy. It is taking on the telecoms components and products division of Stanford Telecommunications, a company currently being acquired by Newbridge Networks. Stanford makes silicon components for "head-end equipment used by cable companies and broadband wireless service providers as well as customer premise equipment (CPE) such as cable modems and set-top boxes". The transaction (on undisclosed terms) also allows "Newbridge to retain access to the silicon components required for the Wireless Broadband Products group". In other words, Newbridge is to become an OEM for Chipzilla silicon. To date, Intel has splashed out more than $3 billion on ten acquisitions in the networking sector, and two more(Olicom's R&D team and the Stanford buy) are pending. Acquisitions span R&D teams, networking equipment, silicon and software. Just goes to show how seriously the company is taking networking, John Miner, head of Intel's new Communications Products Group, says. Intel is not exactly in Netzilla league yet. And it's a wee bit difficult to work out just how well the company is doing as a Building Block Supplier For The Internet Economy -- as it declines to break out figures by product group. But when it comes to strategy, Intel is more than willing to talk the talk. Yesterday, Miner ran some UK rags, including The Register through Intel's thinking on networking. First up, Intel prefers to think in terms of communications industry and computer industry, each of which produces $200 billion or so in annual revenues worldwide. Fuelled by the Internet, these markets are merging into one enormous opportunity. And Intel likes what it sees. Miner cites estimates that only four per cent of the servers needed for the Net infrastructure of 2005 have shipped. On previous form, Intel could anticipate more than 80 per cent of this business, and possibly more than 90 per cent when IA-64 boxes kick in (our calculations, not Miner's). But the company's ambitions extend much further than that -- it wants to become the OEM for the networking world. The convergence of voice and data round a single IP telephony network is a strategic inflection point (a concept popularised by Andy Grove in his book Only the paranoid survive) for the comms industry, Miner argues. And this gives Intel its chance to punch its way into the sector. As networks unify around the agreed IP standard, network providers must slug it out on the basis of services. This fits nicely into Intel's bag. Standardisation enables siliconisation enables commoditisation. So Intel will be seeking out a hell of a lot more Newbridge style Internet building block supplier deals. Intel also reckons that IP services -– or some of them, at any rate -– play to the company's strengths. This is where the $500 million-plus investment in server farms, announced in June, comes in -- two Intel farms are already up and running Stateside and a third is slated for Europe (London, we guess) in the next few months. Intel wants to sell this capacity to ISPs and small and medium businesses, because that's where it thinks it can make the most impact. Intel also want to fill the racks of the world's ISPs with its servers, shock, horror. This is where channel conflict could rear its ugly head. One could easily work up a scenario where, say, Intel found itself competing with IBM for a Tier 1 ISP contract to supply a couple of thousand servers. But hey, that's a conflict worth fighting. Outside the ISP environment, there won't be any Intel Outside-branded servers, Miner says. But there will continue to be Intel-branded networking equipment, software, and services. The logic of this is that Intel will move upmarket at some point. You would have thought the big networking equipment suppliers -- Cisco (the real Netzilla), Lucent, Nortel et al -- would be just a little suspicious of their friendly silicon manufacturer. You would have thought. ®