Dedication from Mohan Subramaniam: Bala Iyer, my research collaborator and a dear friend of many years, sadly passed away recently. I would like to dedicate this article to his memory.
IBM’s purchase of Red Hat is the largest acquisition in the history of software. According to The Wall Street Journal, this $33 billion acquisition is expected to shore up IBM’s position in cloud computing services. Yet, this is an expensive bet to seek parity with other cloud service providers like Amazon, Google, and Microsoft. Just a few years back, IBM was betting the farm on artificial intelligence through its Watson platform. That strategy has yet to deliver promised results. Is it going to be any different this time with Red Hat?
First, it’s important to look at the appeal of this acquisition. Red Hat’s business model is built on the Linux operating system, which came to the fore in the 1990s. Over the years, Linux has emerged as a stable and reliable OS that supports a vast ecosystem of software developers that meets most computing needs of users and enterprises, spanning all kinds of industries across the globe. In addition to its popularity and widespread use in computing devices (from desktop to mobile to smart devices and household appliances), Linux is open source software, available for free.
So why did IBM place such a big bet on free software? Why is this any different from its earlier bet on AI-driven Watson?
To understand the change in IBM’s expectations, it’s important to understand the modern IT industry stack. This stack has three layers — infrastructure (consisting of hardware, operating systems, and middleware), applications, and a newly emerging AI and analytics layer. An AI engine that resides in this new layer is designed to solve specialized problems based on transactional data in the applications layer. For example, H&R Block’s applications layer may automate taxpayer information, calculate returns, file returns with the IRS, and prepare clients for refunds. Watson’s AI engine, with its intelligence on millions of tax codes, can identify unique tax saving opportunities or inadvertent errors that the standard algorithms in the applications layer may have missed. In doing so, it complements the applications layer in unique and specific ways.