Sun has done this before and the re-centered on one HW/SW architecture. In the early nineties, Sun did the 386i, the Sun 3 line, MC680x0 based, as well as starting the SPARC line, all the while making the transition from SunOS 3 to SunOS 4. The support issues were horrendous. So bad that the field pre-sales engineers were closing service orders.
Shortly after that the ?all the wood behind one arrowhead? slogan came out and the Sun 3 line and the 386i line ended. You might see the same for SPARC, but in a much slower rate then occurred with the 680×0 based systems, Sun Has a great deal of IP in the SPARC line. As for Solaris, its days are numbered, even though Sun has just as much IP there as in SPARC. It’s main selling points are history and scalability, there is still a great love of Solaris in the marketplace, and is a stalwart OS today, but with Linux getting better at scaling and the hay day of the big SMP single memory space environment ending, Solaris could be a niche OS for things like 3+TB data warehousing applications.
In this style of application the latency of multi-OS solutions is untenable, and until the multi-OS interconnects get much faster making the single OS image across many boxes a viable solution, Solaris is still a good platform on which to deploy.