Smalltalk "failed" more due to the success of Java than because there was anything wrong with Smalltalk as a technology (so did a host of other promising dynamic languages). It certainly helped that Java implementations were available for free. Java got lucky that because it could ride the applet track that went along with the internet boom. Fortunately for Java, after applets failed, people had figured out that you could use Java for developing server side apps. I think that you could say that Java and Smalltalk failed equally well for producing desktop / client applications. Many of those applications became web browser applications, so the failure wasn't so apparent.

Lots of technology and expertise has migrated from the Smalltalk world to Java, so I don't think that you can count Smalltalk as a failure on that count either.

Even now, I wouldn't count Smalltalk out. Recent interest in dynamic languages may result in a second life for some languages that are currently out of the limelight. Who knows, maybe in a few years, I'll be looking at search strings that read "Java/C# failed for a good reason".