Sunday, May 07, 2006

The big bandwidth bandwagon and the commodity cpu

As opposed to oil, presently bandwidth is expensive.

Continuing the trend, bandwidth will get cheaper and cheaper. There is nothing hard for bandwidth companies to do accept compete with each other. The tec behind the internet gets cheaper and better every year and the wires can take much much more.

Companies that are purely in the business of providing bandwidth are going to struggle unless they offer some sort of additional service. The provision of these 'bundled' services will also become highly competitive and hence cheep.

To consumers this means everything that comes over the wire will drop to virtually nothing for the distribution costs. Many of the additional services will also get very cheep very quickly.

This means,
very cheep voice contact to anywhere
very cheep video contact to anywhere
video on demand and high def will get cheep.

Phone companies will be the biggest losers, expect them to desperately try to diversify their services and add 'bundles'. They will always loose to more competitive and focused companies on the net that provide the services. Say Vodaphone try to bundle games to augment their services and give people a reason to keep paying. This will struggle even medium term, specialist game services on the net will provide better options for consumers who want to choose their offerings. The bundled service will always be inferior so will stay cheep, it cant justify keeping up the distribution cost.

The content companies and the service companies will continue to be kings. The people that provide new and unique content or the people that provide useful services. Not the distribution providers.

The only way to make money on distribution is monopolise it and mark up. Something that is no longer easy to do. Game consoles are a good example where this trend still continues but even their is loosing traction against app stores.

Over the medium to longer term the device at the end of the distribution line will be factored into the distribution paradigm. What i mean by this is that the phone, console or tv will also become very cheep (and often become part of the wire service, like a tv or phone is now)

There are still a few years to go here but chip tec is already beginning to go wide rather than deep. Ie it is going more toward parallel processing rather than faster processing. Both Intel and AMD are going this way and so is IBM/Sony with Cell. its the only big way forward in chip tec. But it is much less technically hard to make parallel chips than faster chips, so once again the competition is coming, and this means cheep chips on the end of cheep wires.

It will become just as hard to monopolise your chip IP in the future as it is becoming to monopolise the wires today. Intel will struggle and once again must move to 'bundle' service to keep value for its customers high.

(Intel will loose its high profitability for chip tec. And will need a new buisness model over time. They will have to get into software. Intel's money comes from the fact it has the deepest pockets to invest in the fabs that take the latest chip fabrication tec to market. This means they can fab the fastest chips the cheapest and quickest to market.. CPU's going wide changes all this)

Its going to get harder and harder to justify the hardware upgrade cycle to consumers. I see time frames of around 2-4 years for the wires and 4-6 years for the devices when these transitions really bite. The wires will be the most dramatic as there is nothing physical for the consumer to switch. The display will come last as there is more innovation to be had here. Until we go all the way to 3D projection screens will be moving.

New content that requires upgrade will always be actively sought after by the chip people. -> app stores deliver this.

Its all good for consumers.

No comments: