Sunday, September 18, 2011

Terms of potential Oracle-Google settlement: even pi billion dollars may not be enough

[NOTE] This is the second of three posts analyzing the prospects and possible terms of a settlement on the eve of court-ordered mediation between Oracle and Google. The third post is an "executive summary". [/NOTE]

As Larry Ellison and Larry Page meet at the U.S. District Court for the Northern District of California in an attempt to settle the lawsuit, many people wonder what the terms of a settlement could look like. I have previously described the status of Google's 20 affirmative defenses and will address the likelihood of an agreement in a subsequent post. This post is based on the assumption that an agreement will be announced tomorrow, later in the week, or at any other stage of the process. What would such a settlement look like?

Scope of settlement

Settlements can resolve parts or all of a legal dispute. They can also go beyond the formal dispute in order to address issues that would otherwise be somewhat likely to arise between the parties, or if the inclusion of additional issues makes it easier to reach a mutually acceptable agreement.

The ongoing lawsuit relates to the following issues:

  1. infringement allegations

    1. patent infringement (7 patents)

    2. copyright infringement (12 copied files, 37 API specifications)

  2. geographic scope

    1. infringement in the United States

    2. contributory infringement outside the United States

  3. period

    1. past (from launch of Android until settlement)

    2. future (everything following the settlement date)

Theoretically, the parties could settle with respect to a subset of the infringement allegations. For example, they could settle the copyright part and keep on fighting each other over patents. That is very unlikely since Oracle gets a certain "synergy effect" from the jury's review of the copyright part: it exposes Google as a reckless infringer and copycat, which in turn could affect the outcome with respect to Oracle's patent infringement claims.

If the parties could reach an agreement only on past damages but still had to fight each other over the future, or the other way round, the settlement would reduce the scope of the trial to only a limited extent. That is unlikely, too.

Concerning the geographic scope, it sometimes happens that patent litigation is settled with respect to patents valid in one jurisdiction while continuing to litigate elsewhere. In the present case, Oracle also brought contributory infringement allegations against Google with respect to the use of its software outside the United States. That's a tricky one since U.S. patents are valid only in the U.S. market. The judge allowed Google to proceed with a summary judgment motion, which may very well succeed.

Regardless of whether Oracle will make any headway with its foreign infringement theory in the California case, I'm sure that Oracle wouldn't want to let Google off the hook in the U.S. market without a license agreement of global scope. Oracle's leverage in the U.S. market is strong enough to work out a worldwide deal. Not only would additional litigation in other jurisdictions be time-consuming: Oracle also has fewer Java patents to assert there, and in some jurisdictions that could be tricky (for example, they may not all meet the European standard for "computer-implemented inventions").

Since Oracle doesn't have any obligation to settle, or to grant a license on particular terms, it would be perfectly legal for Oracle to reject a settlement proposal that has anything less than worldwide scope. Oracle may very well ask for the same license fee around the globe regardless of how many patents it has in any jurisdiction, and how valid they are. For Google, it would be a non-option to give up the U.S. market. Also, it wouldn't be in Google's interest to settle the U.S. case in a way that would bring about further litigation.

Barring a major surprise, a settlement (if and when one is reached) is going to cover all types of infringement allegations brought by Oracle, addressing past infringement as well as enabling Google to continue to sell Android going forward, and all of that on a worldwide basis.

But the parties may very well conclude that they need a deal that goes even beyond the scope of the present case:

So far, Google has not countersued Oracle. But it recently acquired patents from IBM -- two batches of more than 1,000 patents each. Some of those appear to relate to technologies of the kind that Oracle offers, such as databases and customer relationship management software. There are even some Java patents among them.

It wouldn't make sense for Oracle to let Google off the hook and work out a deal for the Android/Java situation only to be subsequently sued by Google over some other patents (such as some of those former IBM patents).

But if Oracle wants to have peace of mind concerning all of Google's current patents and the ones it will obtain in the near term (as it keeps applying for, and buying, patents), then Google couldn't put itself into a defenseless situation because otherwise Oracle might start another lawsuit over something other than Android's Dalvik virtual machine. Oracle owns more than 20,000 U.S. patents, some of which could read on different Google services.

So the outcome would have to be a cross-license. Such a cross-license wouldn't necessarily address the two companies's entire current and near-term patent portfolios, but it would have to be broad enough in scope to ensure that no new conflict arises anytime soon.

One-off payment vs. running royalties

Settlements of intellectual property infringement matters come in different shapes and forms, but they are always license agreements (sometimes structured as a "covenant not to sue") that also include provisions for the termination of the lawsuit and the allocation of the related legal fees.

There are many ways in which license fees can be structured. If you purchase packaged software in a store, you also buy a license, and it's typically a one-time cost. But some software licenses are based on subscriptions, generating recurring revenues for the vendor.

In a situation such as the one between Oracle and Google, it can work either way. If Google offered Oracle a huge amount, such as perhaps pi billion dollars, it's possible that Oracle would take it. Pi billion dollars would represent approximately 40% of the price Oracle paid for Sun last year. That would be a way to look at it -- but it's not a ceiling of any kind.

Even pi billion dollars would likely be a less lucrative deal for Oracle than a long-term license deal based on a per-unit royalty. In July, the IDG News Service quoted an analyst who claimed to know that Oracle was demanding 15 to $20 per device from Android handset makers. Shortly thereafter, Google claimed that there were more than 550,000 activations of Android devices per day, with significant week-on-week. 365 days times 550,000 activations would already be a run rate of more than 200 million devices a year. If you multiply that number with the lower end of the range of what Oracle is asking device makers for ($15), it would take only about a year to exceed pi billion dollars.

That example doesn't factor in further growth, for which there's plenty of potential. Oracle's management owes it to its shareholders not to leave any money on the table. At the same time, it has some discretion when negotiating a settlement. Asking device makers for $15-20 per device is one thing -- but when you can sign a high-volume deal with Google and don't have to shake down each and every device maker, you can go down quite a bit and it may still make business sense.

Even with a drastic reduction of the per-unit amount, a multi-year license deal with a running royalty would still be hugely more profitable than a one-off payment over the hypothetical amount of pi billion dollars.

Oracle has the financial strength and also a reasonable likelihood of winning its lawsuit that it can afford to insist on a per-unit royalty even if a multi-billion dollar one-off payment was offered. At the same time, Google has limited income per copy of Android with its advertising-based revenue model. Even if Oracle went down to half, a third, a quarter or a fifth of the amount it purportedly wants to receive per copy, it wouldn't be possible for Google to operate Android profitably while having to meet such a running royalty obligation.

Even a tenth of the rumored range, or $1.50 to $2.00 per unit, would be very hard for Google to make work without modifying the Android business model and passing its license fees on to device makers. Based on the current business model, Google doesn't even have a way to limit the distribution of Android. Oracle would likely hold Google responsible for all distribution, not just the part that's based on license agreements with Google. So if some vendor in China took Android on open source terms and sold hundreds of millions of units over the years, Google might not have any noteworthy revenue from that activity, but it might have to pay.

Oracle doesn't have any reason to destroy Android. However, Oracle ideally hopes to impose its commercial licensing model and doesn't care whether this may require Google to change its business model.

The sky (or the market) is the limit for Oracle's demands

When reading my theories about why pi billion dollars might not be enough to settle this case, some may wonder how the settlement could even exceed the $2.6 billion figure of Oracle's original damages claims, let alone any new damages claims, which the judge said should start at $100 million since that amount was the cost of an Android-related deal Sun offered Google years ago.

Just to be precise, any of those damages amounts could be tripled based on willful infringement, for which there are pretty strong indications. But even any post-tripling amount still wouldn't be a hard limit.

We're talking about two completely different processes that arrive at a number:

  • If a court determines damages, it has to follow certain legal rules. A key part of the consideration there is what the parties would have agreed upon in a hypothetical negotiation prior to the launch of the infringing product. Also, the plaintiff bears the burden of proof, and sometimes there are damages that are very real but hard to quantify and substantiate.

  • If Oracle is handed a permanent injunction, and if Google can't work around it without major complications for Android, there's far more leverage given that Android is successful and that well over 200,000 android apps have already been written in Java. It's not Oracle's fault that Google is in this situation. It's Google's own recklessness and bad decisions (such as the one to reject Sun's offer) that have led to this.

It would be a mistake to think that the judge's perspective on damages limits Oracle in any way. Since Google doesn't adhere to the Java standard (the "fragmentation" issue), it isn't entitled to a license on any particular terms. Oracle can always say no. That's the law. A patent is a time-limited, scope-limited monopoly. Unless a company makes a commitment to grant licenses, or has an obligation under antitrust law, it can use a patent as an exclusionary right.

This is what the Patent Act -- specifically, 35 U.S.C. § 271(d) -- says:

No patent owner otherwise entitled to relief for infringement [...] of a patent shall be denied relief or deemed guilty of misuse or illegal extension of the patent right by reason of [...] (4) [having] refused to license or use any rights to the patents [...]

Recently (in investigation no. 337-TA-769), the ITC staff also quoted the following from various Federal Circuit decisions:

"[C]ourts have noted that the patentee begins with substantial rights under the patent grant. [...] Those rights include the right to suppress the invention while preventing others from using it, to license or refuse to license the invention, to charge such royalties as the leverage of the patent monopoly permits, and to limit the scope of the license to a particular field of use."

(emphasis mine)

As the quoted passage says, it's all about leverage. If a patent can be worked around at a fairly limited cost, one can't charge far more for a license than what the cost of a workaround would be. Also, if Oracle (totally hypothetically speaking -- this isn't going to happen) asked for so much money that Android would be placed out of business, Google might decide to give up that business area.

Between Oracle and Google, the key question is whether Oracle's leverage is sufficient to force Google to modify its business model only in order to be able to pass along the license fees Oracle seeks to collect. I will get to that again in my "executive summary" post on the negotiating situation and the prospects of a settlement.

[NOTE] This was the second of three posts analyzing the prospects and possible terms of a settlement on the eve of court-ordered mediation between Oracle and Google. The third post is an "executive summary". [/NOTE]

If you'd like to be updated on the smartphone patent disputes and other intellectual property matters I cover, please subscribe to my RSS feed (in the right-hand column) and/or follow me on Twitter @FOSSpatents and Google+.

Share with other professionals via LinkedIn: