The primary driver of profitability above the cost of capital is your intellectual property which includes names of Customers, Revenue, Prices, Costs, technical Engineering data and Manufacturing processes. To apply AI in large companies often requires the compute/storage power of the “Cloud”. A measure of the Cloud compute power is Amazon Web Services (AWS) with over 1 Million servers (each with 16 “logical” CPU), Microsoft Azure with 500,000 servers etc. This allows an exponential increase in the compute power “available for rent” as needed. The Cloud is thus a primary enabler of AI solutions to “unsolvable” Combinatorial problems (see “Lean Six Sigma in the Age of Artificial Intelligence”, pp 122-132). But can your intellectual property be hacked on these Clouds? This appears unlikely as AWS provides Cloud services to the DoD storing Top Secret information. (AWS News, Sept 15, 2017). However, some of their competitors say that AWS can be hacked. (Forbes, Oct 26, 2018).
If security is a matter of concern, it is now possible for a company to own its own “Cloud” at ever lower cost. For example, the Hewlett-Packard HPE ProLiant ML350 Gen10 Server now offers 32 Gigabytes of storage with Intel Xeon Scalable 4208 processors for less than $2500. Arrays of these servers allow even the largest company’s ERP, Engineering, etc. data to be locally stored. A local Cloud enables AI waste removal and cycle time reduction at scale without compromising data security and compliance. The arrays of Servers are pre-configured in Hewlett-Packard Enterprises (HPE) facilities. HP provides service and support. We will discuss your security issues as part of our “AIT Trial” process. The procurement and maintenance costs of a company owned Cloud are still significantly higher than the “rental” cost of AWS, Azure, etc due to peak vs steady state demand. The implementation of AI requires very high peak Cloud use during the training of Neural Networks compared to steady state “Big Data” ERP save/retrieve file requirements. Peak compute power is most economically supplied by AWS, Azure, etc.