What do you mean by Intel is making the wrong bet by going all-in on mobile? By mobile are you talking about smartphone/tablet or laptop? What would you do instead?
I feel Intel is going to make the mistake of taking their datacenter advantage for granted. They should be doing everything to cement that advantage instead of an unnecessary focus on mobile, by which I mean the present ARM space of smartphone/tablet.
This goes with any business, but Intel should figure out what its "plus" is. They own perf/watt and perf/thread, and their uarch, x86, is the standard for cloud loads. As the cloud matures and becomes more uarch agnostic, everything being open source, they still have to find a way of differentiating themselves.
The point is 20 years ago the majority of coders were coding x86, directly, then 10 years ago say WinIntel, and now our apis are REST calls. That is to say, the majority of programming has been getting increasingly removed from the underlying hardware. Intel it seems finally clued in to this with MS announcing WinRT.
So in short, Intel should start innovating at the platform level. And hey, the platform isn't just a microprocessor anymore.
I can see what you mean to some extent. The top of the line server parts are still Sandy Bridge-E. By the time Ivy-Bridge-E is out Haswell will be available on the traditional PC side.
I would actually make a different argument though. As time goes on the absolute cost of the CPU is not that important. For big data centers lowering costs is about efficiency more than cheaper hardware. So in my mind a lot of what Intel is trying to do on the Mobile side, both laptop and smartphone/tablet, will help them in the data center. If Intel can drive down both peak and idle power consumption while maintaining or even improving both single threaded and multi threaded performance, there's potential to really lower the power needed to run a data center and in turn create a lot of value.
The real question is what's the next big thing going to be? Eventually the current CPU model will give out to something else. Maybe that's something like Psi or Tesla or maybe something completely different.
phaet0n|13 years ago
This goes with any business, but Intel should figure out what its "plus" is. They own perf/watt and perf/thread, and their uarch, x86, is the standard for cloud loads. As the cloud matures and becomes more uarch agnostic, everything being open source, they still have to find a way of differentiating themselves.
The point is 20 years ago the majority of coders were coding x86, directly, then 10 years ago say WinIntel, and now our apis are REST calls. That is to say, the majority of programming has been getting increasingly removed from the underlying hardware. Intel it seems finally clued in to this with MS announcing WinRT.
So in short, Intel should start innovating at the platform level. And hey, the platform isn't just a microprocessor anymore.
ikono|13 years ago
I would actually make a different argument though. As time goes on the absolute cost of the CPU is not that important. For big data centers lowering costs is about efficiency more than cheaper hardware. So in my mind a lot of what Intel is trying to do on the Mobile side, both laptop and smartphone/tablet, will help them in the data center. If Intel can drive down both peak and idle power consumption while maintaining or even improving both single threaded and multi threaded performance, there's potential to really lower the power needed to run a data center and in turn create a lot of value.
The real question is what's the next big thing going to be? Eventually the current CPU model will give out to something else. Maybe that's something like Psi or Tesla or maybe something completely different.