This from the article: According to a recent analysis of more than 30 million transactions, Zylo—a local firm that helps its customers track and manage software subscriptions—found enterprise companies are spending an average of more than $10,000 per employee per year on software. So, for every 100 employees, companies are spending $1 million on software subscriptions.
So what would an IT department, physical plant, and software purchase prices cost? This is an area where data scientists have had a big impact, pricing models. Between keeping the cost "just under" the similar cost to do it yourself, and adding the value of "you can double or halve your size without any additional HR costs", the SaaS model gets an enterprise to be closer to being precisely sized.
Think about how this worked in the past, you over provisioned infrastructure based on projected growth, you sold off at a loss infrastructure you no longer used, your business expense didn't track your actual size and always cost more. So in that model SaaS wins pretty handily.
Google Enterprise - replaces Web server, mail server, DNS server, chat server, and NAS devices.
Solidworks SaaS - replaces simulation resources with cloud resources.
Slack SaaS - chat servers, administration of same.
There is an IT concept, SUDs, which is short for "Servers Under Desks." That is a situation where an employee sets up a server do do something and it "lives" under their desk because they don't have an official place to put it. And then it becomes mission critical without anyone knowing and there is a big fire drill when something disrupts the flow (employee leaves, server crashes/fails, scaling requirement isn't met, Etc)
So what would an IT department, physical plant, and software purchase prices cost? This is an area where data scientists have had a big impact, pricing models. Between keeping the cost "just under" the similar cost to do it yourself, and adding the value of "you can double or halve your size without any additional HR costs", the SaaS model gets an enterprise to be closer to being precisely sized.
Think about how this worked in the past, you over provisioned infrastructure based on projected growth, you sold off at a loss infrastructure you no longer used, your business expense didn't track your actual size and always cost more. So in that model SaaS wins pretty handily.