The suit seems to have been filed in 2011, and HP said:

"The Itanium product line was highly profitable for HP, generating over $2 billion in annual profits in 2010"

And, also "84 percent of Itanium servers ran Oracle’s database software."

So, even though Itanium was pretty well dead for new sales to new customers 2010, they were still selling enough of them to make $2B in profit (not revenue) the year prior. That is, sales to customers that were sort of stuck on Itanium/HP-UX, and not yet ready to switch to something else.

Certainly, an EOL announcement for Oracle on Itanium would have kicked some Itanium-Locked-In customers into action faster than they would have otherwise.

I'm not sure how to translate $2B/year profit on Itanium hardware into lost sales due to the Oracle announcement. But, $3B doesn't sound outrageous in relative terms. Sounds like 1.5 years of H/W sales ending earlier than projected due to the stark message in Oracle's EOL announcement.

On the flip side, $3B is a lot more than it would have cost Oracle to keep supporting the platform. Having worked on a product that at one time or another supported Windows (including IA64), VMS (including IA64), and thirty-one flavors of Unix (including HP-UX and Linux on IA64), the marginal cost of continuing to support a platform is not that high. Sure, if the demand isn't there, we drop it, and we have dropped a lot of platforms over the years, including VMS and everything IA64. If HP claimed the demand was there, and furthermore claimed a contract required us to keep going, it would be a no-brainer. I don't get it. I mean, it's Oracle, so I get it, but I don't get it.