Sun was huge on thin clients around the turn of the century, the last great battle between thin and thick clients. In the first dotcom bubble, Sun sold a zillion Enterprise-series (E400/E4000/etc. big boxes) servers and wanted to keep that flow going. The market evidently did not agree with their plans.
The perspective that is lost to time is that back then Sun was kind of like AWS nowadays WRT mindshare, with the addition of hardware business, which is key to my point: likely when McNealy said that they underestimated 32bit he was eliding the problematic future they saw in the UltraSparc 3, that it was an engineering dead-end (which begat the T series), and, well...capitalists profit from externalizing their liabilities. "We're still awesome, it was 32bit's fault."
The perspective that is lost to time is that back then Sun was kind of like AWS nowadays WRT mindshare, with the addition of hardware business, which is key to my point: likely when McNealy said that they underestimated 32bit he was eliding the problematic future they saw in the UltraSparc 3, that it was an engineering dead-end (which begat the T series), and, well...capitalists profit from externalizing their liabilities. "We're still awesome, it was 32bit's fault."