Sell ARM based systems
Intel based systems are power hungry and with rising energy costs, many people are realizing that they do not need power hungry Intel based systems and they would be happy to move to systems using alternative ISAs. In the past, computer companies such as Dell had to sell Intel-based systems because Windows required it, but the success Dell has had with Linux (and the fact that I am typing this on a Dell laptop running Gentoo Linux) shows that people do not need Windows anymore. Power is becoming an ever growing concern in many places.
Laptops, netbooks, datacenters and embedded systems are all places where power consumption is key. I had a Dell Inspiron 4000 in 2000 and I loved it because it had a dual-battery feature that gave me an 8 hour battery life. When I purchased a Dell Inspiron E1705 in 2006 because of the experience I had with my previous Dell laptop, I noticed the difference in battery life immediately and I have been very disappointed. Improvements in operating systems has improved the situation, where I can manage to get a 3 hour battery life, but still, that is a far cry form the 8 hour battery life where I could go someone, work all day on something and not need to plug my system into a power outlet once. Netbooks, which tend to have lower power requirements and corresponding good battery life have become very popular because of this. While a market for high-end systems that use enormous amounts of energy certainly exists, it is quite small compared to the mainstream, which Dell's offerings have not targeted well since Dell abandoned the energy efficiency of the Inspiron 4000. Another example of this is the sheevaplug, which has become quite popular among hobbyists:
The sheevaplug is an embedded system, but it is being targetted for datacenters, and as virtualization becomes more persuasive, systems like it that have enormous numbers of computational cores will begin to phase out existing Intel-based systems.
My proposal for Dell is simple. In each product segment where power consumption is a concern (netbook, laptop, desktop, datacenter) offer a system that is based on the ARM architecture. Nvidia has a chip called the Tegra 2 which would be excellent for netbooks and laptops:
The laptops could run either some flavor of Linux (likely Ubuntu as Dell already offers it) or Google Chrome OS. Budget desktops could be produced with the Tegra 2 as well. Google is said to be designing a Google Chrome Netbook based on the Tegra 2, which would likely harm Dell's business should Dell not produce a similar product:
For datacenters, Dell could license the ARM architecture, have its engineers design a multicore chip (with an accompanying chipset) suitable for server virtualization and contract TSMC to produce the chips. Dell could then cheaply produce servers for tomorrow's datacenter, running Linux with network cards installed that have hardware acceleration support for virtualization. Anandtech has an article on one such network card that Dell could use, which would enable Dell to produce systems that will replace dozens of others:
I realize that the idea of having a single server that replaces many others would likely be detrimental to other areas of Dell's business, but this is the future and if Dell does not do it, some other company will and Dell will lose far more business in the long run than they would if they produced such a product themselves.
In summary, I believe Dell should offer ARM-based laptops, netbooks, desktops and servers (intended for datacenters). A single good quality product in each segment designed and marketed for low-power and low-cost to test the waters would be sufficient. I imagine that if Dell were to bring ARM-based systems to market, even as a trial to test the market demand, your company would have such success that you would want to replace some of your existing product lines outright with ARM-based systems.
Status Update: This idea has been partially implemented with the announcement of the "Copper" ARM server. Thank you for posting.