There is room for many opinions on this topic. I am very grateful to HP (and in the end HPE) for a fun and engaging 37 year career, and for standing by me at a few difficult spots during that time.
Here’s what I would observe, from an R&D perspective:
Hewlett Packard was the finest small-project hardware incubator there ever was, in a business model where projects or products could fail independently of each other, feeding a business/manufacturing/selling process tuned for modest volume in each of maybe a thousand products (plus options and accessories) sold for 3–5x parts cost.
Think about the open lab stock (and Bill Hewlett personally cutting the lock off when a well intentioned employee tried to protect the asset), the G-job culture in which side projects like Wozniak’s were the norm and the company not only funded parts for you but would look the other way if 10% of your work hours were spent on “personal” projects. Where someone like Chuck House could bootstrap a business on the sly.
PC’s broke the business model, with much lower margins which have spread to other parts of the computer business.
Computer systems suddenly found, instead of 10 engineer teams which could fail independent of each other, 1000’s of engineers on 100’s of teams all of whom had to succeed in order for the product to meet customer expectations. I remember as a very junior engineer being told about then-CEO John Young walking around talking to senior managers about what it meant that the computer business was set up as independent divisions with independent P&L’s, whose pricing was arbitrary since the customer only cared about the bottom line, not how their revenue dollar was divided among HP’s computer divisions.
Now, think about running a software business in a hardware development reward mechanism / measurement paradigm. The approach handles firmware developed for and embedded in a particular product design well. But when it comes to running a software business, instead of saying we have this code base to develop, mature, tune, and evolve over a period of decades; we have to get the foundation right; and we have to be thoughtful about accumulating technical debt; instead we have a series of independently funded 1-to-3-year efforts, each of which must over its maybe 3-5 years of development+sale life show profitability, with the previous software releases treated as a sunk cost and the future software releases not yet funded. (HP/HPE is not this stupid, but that is the measurement system the computer business inherited from instruments.)
It simply wasn’t the right measurement system for a Microsoft, Amazon, Facebook, or Google. And concentrated on taking risk at the product level, not the large scale corporate risks Apple took with the stores supporting iPod and then iPhone.