Three reasons why Intel might lose server market even faster than consumer

One of the biggest challenges to Intel’s x86 hegemony in consumer market has been the emergence of smart phones where ARM based processors have 90% market share. The PC market where Intel has a strong hold is declining at (estimated) 10% annual rate.

While a lot of coverage has been to given to Intel’s decline in consumer market, I believe Intel has tough road ahead in server market too. One major advantage Intel has is the amount of (legacy) code which has been written for x86 (the same thing which keeps Cobol on life support), this advantage is diminishing rapidly due to following trends.

  1. Demise of gcc and rise of LLVM
    For years Gcc was used and since x86 was dominant ISA (instruction set architecture), several optimisations were done for optimised x86 code generation.
    Now, gcc is being replaced with LLVM and clang by Google and Apple. Given that both companies are heavily focused on their ARM based devices, it should not come as a surprise that LLVM will have better support for ARM code generation in long run. Rumours of Google (Intel’s 5th largest customer) working on ARM based servers are already out.
  2. VM (virtual machine) based languages
    A lot of server side software development (web or otherwise) has moved to VM based languages (Java, PHP being the old leaders and Ruby, Python, Node.js the new ones).
    VMs makes these platform agnostic and hence making a move from x86 to ARM even simpler (since only the VM has to support underlying platform).
  3. PaaS (platform as a service)
    As more and more services are being moved into cloud whose underlying infrastructure is being dominated by only a few players (Amazon, Google etc.), it makes it even more easier and economical for these players to design their own custom ARM processors and offer it to their customers.