In the end, the real core issue was that the executives thought they were experts because the technologies looked so similar. They didn't learn the critical differences and the customers and employees paid the price.
Now Cisco didn't try to acquire a server company, which is the path that did sink AT&T's effort and is likely giving Oracle executives no end of heartburn. I still think it likely that Oracle is not fully reporting the level of pain from the Sun acquisition, given the shape Sun was in at the time.
Cisco acquired people from companies in the server business, mixed them with its own folks and continues to attempt to build a server business largely from the ground up. When you hire someone from another industry initially, you don't really know what to look for and senior executives, if successful, know how to interview well. That is often how they advance inside their firms after all. But in the larger firms, like an HP or IBM, duties are heavily distributed and very specialized to the company. For instance, an executive may understand very well a part of the server that HP or IBM builds, but typically won't need to understand all parts because of the massive layers and specialties inside the company. Unless they have been forced to branch out, and people really don't like relearning core skills, their experiences are very narrow.
Problems are also very localized, which means while the organization gains experience, much of the learning never makes it to all of the key individuals. This is because large companies are very turf-oriented and folks trying to understand another part of the business are often seen as threats and are discouraged from those efforts.
In short, if a vendor tries to build from the ground up by pulling from a large firm, you'll miss core skills and neither you nor the people you hire will recognize the gap because neither has learned to look for it.
However, if a company like Cisco tries to hire a complete team, it will likely be sued successfully for an attempt to steal a competitor's intellectual property because that team, collectively, may know all of the critical competitive secrets and the firm being raided my not be able to replace them. (In fact, when IBM did its first major layoffs in the early 1990s, it lost for a time the ability to build one of its core products and had to rehire a number of key assets in order to resume production). These tend to be lessons hard learned.
Wrapping Up: Anticipate the Risk
Whether we are talking Google, Cisco or another vendor going into a brand-new area, anticipate that they will talk a good game but they likely don't yet know where they are skill-deficient. This means that you should make sure you build in clauses that protect you in case of major failures to execute and place the burden of that execution on the vendor. While this may actually not be a bad practice, even with experienced vendors, it is critical with someone exploring the space. It is unlikely that IBM, HP or Dell will exit servers, but remember IBM did exit telephone and AT&T did give up on computers, so there is history here and a clause that keeps you whole in case they pull the plug would be wise. (Cisco did exit virtually all of its consumer efforts recently after all and a major mistake could cause the same result here.)
In the end, if a vendor wants to learn a new skill, make sure they either fund that lesson themselves or through one of your competitors; learning by doing is expensive and you certainly don't want to have to absorb that cost yourself.