ARM vs Intel: a Core vs a Chip

ARM versus Intel has probably been one of the most important technology battles in the recent years. Well, may be not that recent in fact. Intel has been around since 1968 (45 years) and ARM officially since 1990, with Acorn dating back to 1978 (35 years). For years both companies have been serving different markets. ARM wan inside small devices like cellphones and Intel was powering computers.

Now when a phone is a computer, ARM and Intel are direct competitors. Industry analysts argue the pros and cons of each company and the underlying architectures. ARM has historically been less power hungry and Intel more powerful. With the advance of smartphones in the post-PC era, and especially with the rise of tablets, ARM and Intel fight for the same spots. ARM has been on a continuous rise and Intel has been steady at best. There is a bet that the Haswell, the latest CPU from Intel is about to change that. And not waiting for the Haswell, Samsung is reportedly already in with the latest Galaxy tablet being designed around the currently available Clover Trail+.

But despite the smartphones and tablets being increasingly popular (and Intel winning some of them), I bet ARM will keep on extending the lead. There will be far more ARM - powered devices then Intel powered. The macro reason is simple. Intel - powered devices are for humans. ARM - powered devices are for machines. The era of the Internet Of Things is coming. Every thing will be connected. Every thing will have a CPU inside. And it won't be Intel. The era of Intel Inside is over.

Intel may win a tablet or a smartphone platform. But at the same time almost any thing, any device will have an ARM core inside. ARM has extremely broad range of CPU cores. They have the Cortex-A that rival the Intel chips performance - wise. But they also have the extremely popular Cortex-M series. A use case: the Cortex-M0 is inside every nRF51-series, the award-winning Bluetooth Low Energy SoC made by Nordic Semiconductor. You can find nRF51's in shoes, door locks, key fobs, and many "smart" things, a market that is just emerging and will be huge. Go and read the latest prediction unveiled by Mary Meeker: wearable computing. ARM in your smart glasses (Google Glass). ARM in your (Bluetooth) headphones. ARM on your arm (pun intended). ARM on your wrist (a watch). ARMs in your pockets (phones, keyfobs). ARMs in your shoes (exercise profile sensors). There will be ARM in a light bulb (Philips Hue or similar). ARM in your wall switch (Z-Wave or ZigBee). Plus there already is the invisible ARMy of ARMs, you use every day. They control your car, your kitchen gear, your thermostats, literally every thing.

To be completely fair, I should note Intel too had cores. The 8051. But it is now (almost) as old as the company itself and it is being phased out by ARM Cortex M cores.

There is no way for Intel to reach the market of the Things. There are two fundamental reasons for that.

Number one: Intel is designed as a generic processor that powers computers used by humans. It has typical human interface subsystems (USB, display) plus the standard computer/networking controllers (memory bus, PCI peripheral bus). ARM is just a core. There are more than 1000 companies who make specialized processors built around ARM cores. There is virtually unlimited choice when it comes to ARM-based chips. Define the specs: what would you like to have on your chip. Hardware-assisted security? CAN bus?  Bluetooth RF? Voila! There always is somebody making the ARM chip you need.

Number two: power consumption. We are working on software applications implementing wireless IPv6 protocol (the 6LoWPAN) with end-to-end IPsec security and many hardware - related functions, including PWM power control or illuminance metering via I2C sensors. The STmicro chip we've selected has everything on board and (surprise, surprise) it is ARM-based. Powered by the Cortex-M3 core, it can run for two years on a single AA battery (an that includes the wireless transmitter sending sensor data over the air every 10 seconds). Intel is several parsecs away from this number.

But based on the cyclical nature of things, I cannot fully rule out Intel will regain its aura. Especially with the new leader, they may try and successfully reinvent the company But then the question is if they want to spread themselves so wide and thin? Focus is very important, even for large companies. Sometimes they simply have to accept they've peaked and focus not to fall too fast too soon.

The bottom line is: people will not be buying many more Intel computers. The upside here is limited. On the other hand every thing can have ARM inside - currently only a handful do.