It was not all all obvious that Windows would become so important, and they knew they needed a new operating system platform for it anyhow. So it does make some sense to me - depending on the licensing terms.
No, not entirely. I visited the Corvallis operation in 1989. It may have been recent, but I got the impression it was some time before that. If it was late 1988/early 1989, would that make more sense?
The HP operation I was in, the Australian Software Operation, was seeking corporate charter agreements to commercialise the cross- platform development tools for user interfaces which I had been working on since 1987, which implemented my inventions from back as far as 1983. Some of those inventions later found their way into the modern web browser, courtesy of a friend from there who later worked at Netscape. But that was after HP had closed the ASO in 1990, and we bought the UIMS technology and founded OSA to bring it to market. Some of that team are still working together!
HP divisions operated almost like separate companies. Just because the relationship was dysfunctional in one area says nothing about other areas. There was very little central coordination or management, and the kind of charter war we engaged in was new to the company, because the idea that one division should own a market segment (and not suffer competition from another division) was new. I circled the globe twice in a year, getting these agreements in place. Then we got shut down; the only profitable software-only operation that HP *had ever had*, and Australia's largest exporter of software. HP decided it was a hardware company, just at the time that "open systems" meant that hardware was commoditized. Buy SUN, buy IBM, buy HP, it didn't matter because it was all Unix. Bloody idiots, the bad management decisions rapidly escalated to Fiorina-scale, and we all know how that ended.
Windows was certainly not (yet) considered crown jewels, and they were looking for an O/S path forward. Rightly so, given that it took them more than a decade to get NT to where it could be used to launch a compatible replacement for MSDOS and Windows.
I hadn't heard about Monterey, but that happened well after Pyramid had built a fantastic SMP Unix kernel. Those machines didn't have terribly fast cores, but they scaled almost linearly to 24 cores, and provided primitives that made the new RDBMS products fly. In fact, it was Pyramid's work - specifically the kernel profiling of RDBMS behaviour done by Ken McDonnell - and collaboration with the RDBMS vendors, that drove the shakeout in that market; the products rapidly converged on optimum behaviour (locks, disk transfers, etc) for the TP benchmarks. Vendors which didn't take Pyramid's advice rapidly vanished because they were uncompetitive.
Anyhow, my point was that a lot of folk had tried to hack SMP support into Unix, but they all (except Pyramid) tried to do it on-the-cheap, and that meant way too many global locks and poor scaling. I don't know why Pyramid succeeded, but I suspect the focus on RDBMS product behaviour gave them a rich source of expectations.
It was definitely after the Xenix sale. But "building Windoze on top of Xenix" was their goal; they just wanted it done by experts in Unix and X. SCO had the Xenix license, but lacked the X Windows expertise that HP Corvallis had.
I'd be curious to learn more, or to communicate with Bob Miller again. Not an easy name to track down, of course!
With some digging, I might even be able to find my email from the time.
Clifford Heath.