Bloomberg reports that another round of talks between software giants Google Inc. (GOOG) and Oracle Corp. (ORCL) ended at an impasse.  Oracle Corp. is currently suing Google for using pieces of its patented Java2ME (J2ME) code in Android via the Apache Project's Java Standard Edition "Harmony" (J2SE).  

Google CEO Larry Page sat with Oracle CEO Larry Ellison for 10 hours trying to work out an acceptable licensing agreement.  The talks ended with no deal reached, with both parties agreeing to talk to court officials about "when further discussions will take place and whether the further attendance of Mr. Ellison and Mr. Page will be required."

I. Oracle's Plan For Double-Dip License Fees

The case represents a substantial risk for both firms.

Working in Oracle's favor is the fact that the judge has suggested that Google's infringement of the Java technology was "brazen".  But Google has some advantages of its own.  It dug up documents from the former CEO of Sun Microsystems -- the company Oracle acquired the Java intellectual property from in 2010 -- praising Google's use of Java in Android, and casting doubt on whether Sun was opposed to the unpaid use.  

Furthermore, Google succeeded in convincing the judge to toss out the $6.1B USD that Oracle wanted -- a figure which included speculative future damages.  Oracle has since revised its estimate downwards to $2.3B USD, a figure that still includes a $1.2B USD speculative damages total for 2012, which Google contends is unfair.

If the pair can't settle up, the case will likely go to trial.  If that happens Oracle will be seeking one thing -- a permanent injunction banning U.S. Android sales.  Unlike Apple, Inc. (AAPL) who is seeking a similar injunction in hopes of permanently preventing U.S. Android sales, Oracle's "permanent" injunction bid would be in hopes of forcing Google into a lucrative licensing deal.  Oracle could ask for a royalty as high as $15-20 per device license, on top of the identical fee it already charges Android handset makers like Samsung Electronics Comp., Ltd.'s (SEO 005930).

This is an important aspect to understand of why Oracle's campaign is more harmful to Android, than Microsoft's licensing efforts -- Oracle is double-dipping, seeking fees both at the OS and handset levels, where as Microsoft is content to stick with handset licensing.

Google offers Android licenses to handset manufacturers for free, though it does earn a certain amount of revenue via its cut of app sales and Android-targeted licensing.  The question becomes how much of that revenue can Oracle take before Android starts to become a losing proposition for Google.

If Google is unable to beat Oracle in court, it will surely regret rejecting Sun's 2006 offer to license Java2ME to Android for $100M USD.

II. Google Could Remove Java from Android

One possible way Google could shirk the licensing fees, though, is to remove all the J2SE/J2ME code, leaving a naked interface for partners to implement their own build of J2ME.  As most Android partners -- such as Samsung and Google subsidiary Motorola Mobility – are J2ME licensees, Oracle likely couldn't do anything about this scheme.  The one major player who would be hurt by such an arrangement would be HTC Corp. (SEO:066570).  The Taiwanese manufacturer, believed to be the second largest Android handset maker, does not license J2ME.  Thus it would likely be summarily sued by Oracle, should it try to plug in the technology.

If Google adopts such a strategy (and there's no clear reason why it wouldn't other than to protect HTC), it would still have to pay the damages for past infringement, but it would be spared from the ongoing double-dip licensing scheme.  That would mean a big loss for Oracle, as the license fees comprise the majority of the money Oracle hopes to pocket in future years as a result of the case.

One thing to recall is that Google is sitting on a $39B USD cash surplus [source], so it can afford to stomach some damages.  But licensing fees?  Those it should be far more concerned about, as they affect its ongoing bottom line.