AI book: Life 3.0 & Cloud FPGA customizable HW

podcast interview with MIT Physicist Max Tegmark on Life 3.0: What We Do Makes a Difference
@ Singularity.FM

Life 3.0: Being Human in the Age of Artificial Intelligence: Max Tegmark: 9781101946596: Amazon.com: Books

The concept is interesting:

• Life 1.0: improvements by evolution, random changes & natural selection
• Life 2.0: improvements by learning: chancing "software" 
• Life 3.0: improvements by design: intentional optimizing of "hardware" 

While for human and other live biological species this concept is still a challenge,
for "cloud based" computers this is already a reality!

Besides options to select HW for the task, such as faster IO, more memory, GPU,

they also offer FPGA: modifiable processors, where microcode can be changed to optimize for specific tasks, such as database, web servers, AI etc, it is possible to speed up 10 or more times.

FPGAs and the New Era of Cloud-based 'Hardware Microservices' - The New Stack

Amazon and Microsoft already offer such option on their clouds:

Developer Preview – EC2 Instances (F1) with Programmable Hardware | AWS News Blog

Inside the Microsoft FPGA-based configurable cloud
"Azure Accelerated Networking... The bandwidth between two VMs inside Azure, even with a 40-gigabit network adapter on each VM, is only around 4Gbps per second; with FPGA-accelerated networking, that goes up to 25Gbps, with five to ten times less latency"

For Google FPGA optimization seems to be not sufficient:
they prefer complete custom HW design, in particular for AI tasks. 

Google answered some of our questions about its fancy new AI chip - Recode

"Google's answer: FPGAs are much less power efficient than ASICs due to their programmable nature. The TPU has an instruction set, so as TensorFlow programs change or new algorithms are developed they can run on the TPU."