Same could be said for SaaS products before the 2020's.
Money. What Xerox PARC was allowed to do was to develop technology that was not yet cost-effective.
I got a tour of PARC in 1975, when I took Bill McKeenan's summer computer architecture course at UC Santa Cruz. McKeenan had designed several Burroughs machines. He knew everybody in computer architecture, and we had really good field trips. Visited an an Intel wafer fab. (2 inch wafers! 1K bit memory! Soon, 4K!) Visited Amdahl, met Gene Amdahl, saw their enormous prototype mainframe, which was to be shrunk for production by making custom ICs for each board. Visited PARC, met Alan Kay and Adele Goldberg, and saw some of the first Alto machines. They were making tall CRTs in house and it wasn't going well yet, but they had some machines going. The first Ethernet was described as "an ALOHAnet with a captive ether." Kay described Smalltalk as treating everything as an active object, acting like a little computer, and working by message passing. (I was kind of appalled at that, because I thought he meant true threaded programming with real message passing. I was already doing that on UNIVAC mainframes, which had user-space threads, called "activities". My job at the time was getting the OS for a multiprocessor mainframe installation to stop crashing. So I was thinking concurrency hell and huge overheads as you sent 1 to a number to add 1 to it for each arithmetic operation. But no, that's not how OOP really works, even in Smalltalk. That's just a way to talk about it, which I did not get at the time.)
PARC was also an R&D center for Xerox copiers, so they had considerable mechanical and electronic fabrication capability in house. That was a huge advantage when they needed to build something.
We were told that Xerox had funded them heavily to push the technology regardless of cost, with the assumption that hardware technology would catch up. It was a good bet for the industry, but the payoff was further in the future than Xerox could really afford. Xerox lost their moat when their patents on xerography expired, and their margins dropped.
Think of fields or problems only a handful of people are working on. Usually because it is viewed as fruitless. Then somebody cracks it and sometimes a use is found for it. The rest is labeled "useless research".
The test would be: not only was there no precedent or suggestion for it, but it turned out to be epochal.
Today such a research environment exists only in our dreams, short of winning a MacArthur Fellowship. Most researchers today work in environments where their direction is heavily influenced by the people funding them. In order for researchers to pursue their own research vision and be paid for it, they must persuade a company or a funding agency that the pursuit of this vision will benefit the funders. Researchers also must justify their research activities and constantly produce deliverables such as patents, publications, and prototypes in order for their performance to be assessed by their funders.
While this more directed form of research has also definitely led to many innovations, the trouble is that only the research that fits the visions of those funding the work gets funded and thus proceeds. However, what about research ideas that don’t meet the interests of companies or government funding agencies? This is one of the reasons why we need to promote curiosity-driven research; society could be missing out on future revolutions in science and technology if our imaginations are limited to what we can imagine today instead of letting exploration take its course.
The Thiel fellowship program which is similar produced many billionaires from somewhat new technology, maybe not quite fundamental research. However some of them achieved it through ponzi schemes with world changing amounts of carbon burned for cryptomining waste heat.