IBM’s Patent Abandonment Strategy

by Dennis Crouch

It is well known that IBM receives more US patents than any other company. The company also abandons more patents than any other company.   Once a patent issues, the patent holder must pay maintenance fees in order to avoid abandonment.  Fees are due at 3½, 7½, and 11½ years after issuance and each subsequent fee is substantially higher. Thus, under the current schedule, the first fee is $1,130, the second fee is $2,850, and the third fee is $4,730.  In recent years, IBM has abandoned thousands of patents for failure to pay the fee due at 3½ years from issuance.  IBM’s rate of abandonment is more than double that of other major patent holders. 

The chart below shows the percent of IBM-assigned patents that that were abandoned withn 4–years of issuance for failiure to pay the maintenance fee due.  The most recent available annual information is for patents issued in 2007. (Maintenance fees for patents issued in 2008 are due in the upcoming months).


A company like Apple is on the other extreme regarding payment of these renewal fees. Based on my records, Apple has not abandoned a single application issued in the past ten years based upon failure to pay the first fee. Canon rivals IBM in terms of sheer number of patents but typically abandons fewer than 1% of its issued patents at the four-year mark.

What’s going on here? Apple and Canon are certainly paying maintenance fees on patents that would be found invalid on reexamination and not-infringed in court. On the other side, IBM could have predicted at the time of issuance that it wouldn’t pay the maintenance fee — so why did it pay the $1,000 issue fee?

55 thoughts on “IBM’s Patent Abandonment Strategy

  1. 55

    It’s true that it doesn’t necessarily make sense to pay maintenance fees on every single patent. But, as pointed out in this post, neither does it make much sense to file for patents in the first place, if you’re going to abandon them later. Maybe IBM is trying to cling to its status as the company with the highest patent issuance.

  2. 54

    Scrolling through and found this track – The first thought that came to mind was that this could be an extension of the IBM defensive publication strategy, maybe including a higher tier of inventions where there is uncertainty about whether they may or may have value or where the value may drop alot in a short amount of time…Anyway that was the first thought that came to mind without any examination of what technology is included here:)

  3. 53

    Just so I'm clear; are you (also) saying that companies buy patents without requiring knowing who has (already) licensed them?

    Some might, but I would not. The reason I would not is that the value of the patent depends on who already has a license. For example, if the patent covers $1 billion worth of business if every company was unlicensed, the patent might be worth X. But if all but $10 million of that 1 billion was already licensed, the patent would be worth virtually nothing.

  4. 52

    Licensable, given the fact that the patents are widely cross licensed by IBM already, there will probably be very little value to anybody else for an individual patent in IBM’s portfolio. I believe that any patents that have high-value potential identified by IBM and are maintained.

    “Also; do you Ned (or anyone) know what happens when the owner of a patent which has been licensed to one or more other parties doesn’t have the renewal fee(s) paid on it? Does this kind of thing happen?”

    This depends I would think on whether the patent license was an exclusive license, or whether the licensee had paid a lump sum fee based on the patent being in force for 20 years and with the expectation that the patent owner would license other firms in the industry. But, unless license had a prohibition on abandoning certain patents, IBM would not breach the contract by not paying the maintenance fees.

  5. 51

    1) Somewhat touched on earlier is that the process w/in a company creates an innovative culture.
    2) Any idea what divisions in an IBM more patents are being jetisoned from? A strong bet would be that they are being released from external client interfaces (more short term enterprises) rather than from internal product development
    3) Extra IP certainly makes IBM more of a poison pill

  6. 50

    Thanks Ned.

    Just so I’m clear; are you (also) saying that companies buy patents without requiring knowing who has (already) licensed them?

  7. 49

    What I mean/am wondering with that statement is if there might be value to others in the patents IBM’s not paying the renewal fees on.

    Shouldn’t (if they don’t already) they (and others with their “dropping” patents) make available and market them to others who might pay good money for at least some of these patents?

    Also; do you Ned (or anyone) know what happens when the owner of a patent which has been licensed to one or more other parties doesn’t have the renewal fee(s) paid on it? Does this kind of thing happen?

  8. 48

    Possible explanation, at least for Apple – Apple is notoriously secretive, and has no official “research division” (not like IBM) where there is a fair amount of interesting in publishing (I’ve co-authored papers with IBM employees). This means Apple may have more time between invention and disclosure, and may also submit patents that are more product-focused than something coming from a research division.

  9. 47

    Maybe the value of each patent makes a difference:

    Rank Inventiveness Index* Inventiveness (5 yrs)

    1 Microsoft 185,044
    2 Samsung 128,727

    3 Canon 109,650
    4 Hewlett-Packard 101,502
    5 Intel 96,610

    6 Hitachi 70,450

    7 Ricoh 57,501
    8 IBM 57,414
    9 Panasonic 51,577
    10 Seiko Epson 49,573
    11 Toshiba 48,221
    12 Medtronic 46,288
    13 3M 46,187
    14 Sony 43,747

    15 Texas Instruments 36,905
    16 Xerox 33,809
    17 Fujitsu 31,057
    18 Pfizer 29,329
    19 Exxon Mobil 28,426
    20 Sun Microsystems 27,497
    21 Apple 25,587

    22 Qualcomm 21,149
    23 General Electric 19,649
    24 Nokia 18,658
    25 Bristol-Myers Squibb 16,295

    *The Top Inventive Companies in this list were selected using the following process: Beginning with a list of the top 1,000 most influential companies in the world (based on recent annual revenue – trailing annual revenue from September 2008 to September 2009), the companies were then ranked and sorted according to a statistical measure of patent value, based on U.S. utility patents issued during a trailing 5 year period. Taking into account the relative quality of each patent (i.e., IPQ score), and time-adjusting for patent age and technology obsolescence, Ocean Tomo PatentRatings® determined the Top 25 Most Inventive Companies.

    Data Source: Ocean Tomo PatentRatings

  10. 46

    Bruce, Why not abandon after publication?

    1) Prestige in #’s of patents issuing
    2) Possible patent award for issued patent
    3) Licensing department pays annuities, has P/L budget (I believe IBM’s licensing department is so organized), while the patent department’s budget is set to get as many US patents as possible.

  11. 45

    There are numerous other ways to publically disclose inventions (e.g., publish technical paper at a conference) that cost next to nothing. Don’t forget, before the $1000 issue fee, multiple $1000’s are spent between inventors time and legal counsel to draft patent application, not to mention additional $1000’s to prosecute the patent with the USPTO before it is issued. If this were a purely defensive stategy, they should have their heads examined. Patents do not prevent you from getting sued. Patents do not necessarily prevent you from losing law suits. Lastly, 18 months after filing an application, it is published. At that point the prior art is established. This publication can be used just as effectively than the issued patent to establish prior art. So paying is issue fee has virtually no defensive benefit. Some of the other arguments (PR, etc.) make a little more sense.

  12. 44


    Comments on the fee structure somewhat capture your “conspiracy” in that the balance in the aggregate requirement combined with basically doubling the fee input (at both the prosecution and the maintenance stages) can only be achieved if the workload is estimated to be halved.

    The objective of weakening the US patent system is neither a new idea, nor a radical (read that as cuckoo-conspiracy) one.

  13. 43

    > A major factor is the PR value.

    Let us not forget intimidation value in preparing new licensees for their ordeal. I guess everyone here has heard of the Qualcomm patents wall.

    And speaking of Qualcomm which (unlike IBM) is a big PCT filer I would expect abandoning strategy to be a little different. Contrasting IBM with Qualcomm and comparing this to licenses registered with patent offices could be interesting.

  14. 42

    That might not be frustrating … yet. I had a case where the client was about to collapse so I called the lawyer handling the bankruptcy process. The patent portfolio was the only thing of any real value in the company.

    Unfortunately the specialists in bankruptcy knew litte about IPR and let all applications lapse, preferring to concentrate on selling the office furniture to cover the expences.

    Now that I call frustrating.

  15. 41

    This strategy of filing and A publishing then abandoning is heavily used, all over the world, indeed for 102(e) purposes, so it does not surprise me one jot that IBM is using it too.

    Just look and see how many hundred thousand PCT cases get filed in China, and then published as a WO. You’ll be hard-pressed to find any of them that are carried forward any further, into the national phase.

    As Bruce observes, this 102(e) strategy depends on the PTO being capable of performing patentability searches that are fit for purpose. When did Mr Kappos first know that he was to be parachuted into the office that bears responsibility for all that stuff?

  16. 40

    Another thing to keep in mind here is the timing of the release of this information. It comes immediately on the heels of USPTO hearings on their new fee structure, where the proposed new fees significantly front load patent fees. In the past, there was a lot of subsidization of examination through maintenance fees. This was done to incentivize disclosure and to have the more successful patented inventions subsidize the less successful ones. Much less in the future, if the USPTO has its way.

    The reason that the timing is a bit suspicious is that the person running the hearings, listening to the testimony, and pushing the new fee schedule, is none other than former IBMer, Director Kappos.

    And, one possible motive might be that lower maintenance payment rates might detract from the argument for continuing the subsidization of examination through maintenance fees that we currently enjoy. I think that the idea is that with lower maintenance fee payment rates, there will be less revenue to shift towards examination, and thus, more up front money is needed there, which needs to be raised by increasing examination costs, at the expense of maintenance fees. Something like that.

    Just a thought for the conspiracy minded.

  17. 39

    Another possible suggestion for this is that IBM seems to have moved from using their own defensive publication program to utilizing patent applications for this purpose. This might explain part of why they are letting so many patents expire. After all, for defensive purposes, a 102(e) rejection is probably better than a 102(a) or maybe even 102(b) rejection, esp. in view of the problems incurred when trying to swear behind references. Besides, examiners seem much more zealous in searching the databases of patents and published patent applications than of publications.

    But, if that were the case, then why bother to prosecute the applications after publication? Looking at their abandonment rate through time might provide more information here, one way or another.

  18. 38

    What must also be kept in mind is that the AIA was intentionally designed to weaken our patent system by making patents easier to challenge and invalidate, and one of its primary backers was IBM (which is why it is not the least bit surprising that they got their man put in charge of the USPTO, so that they could use the agency to also lobby for the legislation – in apparent contravention of the Obama Administration lobbying rules).

    My guess is that we will ultimately see a lower filing rate for the company, after their backlog works its way through the system. We shall see.

    Why would they be doing all this? I would suggest that the lowered level of maintenance fee payments is a result of a forward looking estimate that their patent portfolio will be significantly less valuable in the future as a result of the AIA, and that they worked for the weakening of the patent system because they determined that the loss that they would likely see from a weakened patent portfolio would be more than offset by a reduction in royalties that they would have to pay to others, as well as patent infringement damages that they would incur.

    So, why not pay the maintenance fees anyway. After all, it appears to have been policy up until a decade ago to almost automatically pay the first maintenance fee (I know it was informal policy at the companies that I worked for up through that time). Esp. in view of the fact that they have long depended on the overwhelming size of their patent portfolio to extract sizable royalties from a large number of other companies.

    One suggestion is that this may have been tax driven – that they could write off the cost of less valuable patents when they went abandoned, thus reducing their taxes through increasing their deductible expenses. Not an expert here, but do have enough tax knowledge to see how this would work. Others here may be better equipped to opine.

  19. 36

    Exactly.  IBM has the same problem as Kodak, and so would anybody that actively licenses/cross licenses.  The patent value is exhausted.
    So, it is either pay the maintenance fee or not.  There is no real option to sell.

  20. 34

    … and if IBM’s not letting the market/s know which of their issued patents they’re going to abandon by not paying the renewal fees, they’re missing out on some potentially very big additional revenue.

  21. 33

    Ned, are you saying folks buy patents without knowing if–and to whom–they are already licensed to?

    Akin to buying a house without knowing if–and to whom–there’s one or more mortgages on it.

    That doesn’t sound logical.

    Wasn’t that one of Kodak’s problems in trying to sell their patents; that they were already licensed to virtually everyone who needed a license?

  22. 32

    Thanks, In-N-Out. There certainly are those who are naturally unhelpful. Fortunately, my experience was that most engineers were eager to help, subject of course to the demands of their “real” job.

  23. 31

    Leo, adverse interests.

    I suppose IBM might retain a license. But what if it wanted the right to include your patent in any future licensing deal it might make? What if you were suing a party that IBM wanted to license?

    I know that most IBM licenses encumber patents with some obligations so that one cannot simply sell the patents without also requiring the purchaser to assume some of those obligations. For example, if IBM licensed Apple for subject matter X, a patent that arguably covered subject matter X would remained licensed to Applie after the deal. But, because IBM would not be inclined to tell you exactly where any patent it would sell you is licensed, they would probably have to retain some control over who you could or could not sue/license. In the end, it probably makes no sense, at least with IBM and perhaps with any big company heavily engaged in licensing, for them to sell you a patent.

    It gets very complicated.

  24. 30

    I’ve worked in-house where (i) inventors were given awards upon first filing and first grant; and (ii) given awards only upon first filing but not upon grant.

    With the inventors who were really good folks to work with, it made no difference.

    But the award-upon-grant scheme did provide, for those ordinarily disinclined to be helpful, incentive to continue being helpful during the prosecution stage.

  25. 29

    Yeah I thought about that too, a friend of mine was telling me all about the ethic hijinks that come into play and what they were told about in class at lawlschool.

    That said, as long as you do the xfer right after issue and have them pay the fees before xfering it seems like it is no different than any other attorney buying any patent “property”. But meh, idk.

    If anyone does weigh in on this perhaps they might also weigh in on an examiner buying his own issued patent after it issues.

  26. 28

    Right… PR factor, and also, maybe IBM isn’t all that interested in suing people. Maybe they just want to not get sued themselves by making stuff public domain.

    My guess is at 4 years, they look and see if it has any commercial value for them (e.g. are they making money on it), and if not, they drop it and are satisfied with the PR.

  27. 27

    Fair points, but our primary concern was encouraging lots of disclosures from very busy engineers, so we also paid a small award just for the disclosure. The last thing we wanted was to have the engineers making patentability judgments on their own – we had a pretty experienced patent team make the filing decisions. We also wanted strong cooperation with the patent attorneys during the preparation phase – the carrot of a substantial immediate payment was quite effective there. Somehow I don’t see that an uncertain payment at an uncertain time as much as 5 years in the future is very motivating.

  28. 26

    I think a better approach is to give awards at filing. From a motivational standpoint that point is obviously much closer in time to the activities that you’re trying to encourage,

    Unless you’re also trying to encourage them to stay with their current employer during the pendency of their applications.

    It could also make sense to defer some of that filing bonus until their invention checks out as being patentable. Anybody can file an application on anything. What the company wants is patentable inventions, so that’s what it should pay for. We’ve seen what can happen at the PTO when you reward the wrong metric.

  29. 25

    Paul Morgan is correct.

    Based on the evidence, I think the licensing department pays the maintenance fees from its budget. It is on a P/L budget. It therefor prunes with a vengence.

    In contrast, the patent production department is paid to obtain as many US patents as it can. It does so with a vengence. It does not prune at all, it appears, in order to win the PR race with its competitors by producing the great number of patents per year.

    Other companies are not organized like IBM, else their behavior would be similar.

  30. 24

    I don’t know what IBM’s practice is, but the isue you mention is precisely why we did not give awards based on issuance, back when I was inhouse. I think a better approach is to give awards at filing.

    Complete agreement here.

  31. 23

    That thought occurred to me as well, but it strikes me as a practice that might run afoul of ethics rules. Anybody have any thoughts on this?

  32. 22

    I don’t know what IBM’s practice is, but the isue you mention is precisely why we did not give awards based on issuance, back when I was inhouse. I think a better approach is to give awards at filing. From a motivational standpoint that point is obviously much closer in time to the activities that you’re trying to encourage, plus this eliminates any financial concerns of the employee with respect to later decisions regarding whether prosecution should be continued in the face of adverse office actions. The inventors are happy to cooperate with prosecuting attorneys for the sake of the pride pride/prestige that you mention.

  33. 20

    So why doesn’t the humorless Democrat as him for a piece of the action to continue if it bothers him so much. Why an atty or a law firm wouldn’t want a piece of something that they really liked, or for that matter the whole firm liked is beyond me. What pray tell would it cost YOU? You could even tweak it so you would be even happier.
    That’s akin to a house painter hiring a “House Painter.”

  34. 19

    “A painter paints and can say at the end of the day, ”

    A patent attorney drafts and at the end of the day can say: Look I have facilitated my client advancing the useful arts.

    That is, presuming you did in fact advance the useful arts.

    If you don’t like advancing the useful arts, patents are not your bag babe.

  35. 18


    If you love the app so much then get the folks to assign it over to your shell corp with you paying them a few thousand dollars for it. Of course assure them a license.

  36. 16

    Its about producing a tangible product… based on intangible property, but still.

    No, it’s about providing a service to your client.

    Pin up the Notice of Allowance on your wall, if it makes you feel better. Pay the issue fee out of your own pocket if you like. But I can’t think why that would be better to have than billed time and a happy client.

    Unless it troubles you to have a client sensible enough to stop spending money when it stops making business sense. But who among us wouldn’t want the likes of IBM as a client?

  37. 15

    IANAE points out the obvious, but what do you have to show for all that work, other than billed time and a happy client?

    Its about producing a tangible product… based on intangible property, but still.

    A painter paints and can say at the end of the day, “Look at what I have accomplished!” A surgeon operates and can say at the end of the day, “Look at the size of that excised tumour.” A senator goes to congress.. Never mind – that’s a bad example.

  38. 14

    Recently got a first-action allowance in a case for a corporate client that went belly-up a few years ago. That’s pretty frustrating.

  39. 13

    “On the other side, IBM could have predicted at the time of issuance that it wouldn’t pay the maintenance fee — so why did it pay the $1,000 issue fee?”

    In addition to the obvious prestige factor of boasting a large number of “issued patents,” IBM’s employee-inventors get some benefit out of having patents issued in their names, even if the patents soon thereafter become abandoned. I have no firsthand knowledge, but most companies give the inventor a small monetary reward upon issuance. Inventors also perceive some professional prestige in having a patent issued in their name. Maybe IBM doesn’t want to irk its inventors by routinely denying them the monetary reward and the prestige of a patent in their name.

  40. 12

    Your client has business objectives. You’re supposed to provide a service in support of those objectives. If the technology is no longer relevant to the client, to the point that even an RCE or a continuation wouldn’t save it, that’s no reflection on you, and you should respect your client’s business decision.

    You also have business objectives. You’ve billed the drafting and prosecution of an entire application to allowance, the client was presumably happy enough with your services to pay your bills all along the way, and all you missed out on was the pittance you charge for sending the PTO money, for a reason that is nobody’s fault.

    Call it a win and move on.

  41. 11

    Is there nothing more frustrating than drafting and prosecuting and application, for years, only for the client to tell you to abandon AFTER getting allowance?

  42. 10

    Apple and Canon are certainly paying maintenance fees on patents that would be found invalid on reexamination and not-infringed in court.

    Are you saying that Apple and Canon know which patents would be found invalid, and are maintaining them anyway? Or that issued patents that would ultimately be invalidated can’t still be valuable? Seems tenuous, especially in view of the recent article (by someone else) about IV’s fifteen patents from a dozen sources that would be intimidating even if each had a 90% probability of being invalid.

    On the other side, IBM could have predicted at the time of issuance that it wouldn’t pay the maintenance fee — so why did it pay the $1,000 issue fee?

    Why would IBM be able to predict such a thing? The whole point of maintenance fees (other than to raise money for the PTO) is that things change over a four-year timeframe, and the value of patents may change based on new information.

    Besides, in some years they abandoned fewer than 10% of their patents, so it’s not like they’re getting everything issued and then throwing everything away at the first opportunity.

    It’s great that IBM is making sound business decisions about its patent portfolio. I can’t even imagine how sophisticated their patent administration would have to be, to make those decisions about so many patents with such confidence in the thick of such a profitable licensing program.

  43. 9

    I have not seen any studies justifying the latter rationale, but I have not really looked.

    Nor have I seen (or looked) for any studies that show that IBM’s system is actually an effective system for saving money by doing maintenance fee pruning systematically.

    Do you have insight into this actual “systematic” efficiency pruning system? What are the parameters used for pruning decisions? Are they truly “efficient” or merely “short sighted”/ How does one tell?

    Methinks there be more than just a little bias flavoring the post above.

  44. 8

    Good point – applications for which non-publication was requested that do not issue have no defensive value at all against later patents, whereas those that issue are fully effective as prior art from their original filing date for everything they show or describe, even if the allowed claims have no value.

  45. 7

    This business of being top of the list of those issued with the most patents. How much is that worth to IBM on Wall street, I wonder.

    And have they made a rod for their own back? If the number of patents issued to IBM should get smaller this year, what will that do to their share price?

  46. 5

    Plebian musings alone will not suffice.

    Climb the mountain to have the view of an eagle

  47. 4

    Further on this subject of maintenance fees, some of the worst offenders in confusing the increasing number of issued patents with the much smaller number of patents that are actually in force [after the great extent of early terminations by non-payments of the maintenance fees, especially the second and third] are academics writing economic articles about patents.

  48. 3

    The $1000 initial fee is a tiny part of a patent’s cost, and (speaking as someone who works for a large computer corporation), you can only figure out so much of a patent’s value between the time you figure something out and start to spend big (lawyer and engineer time) money on it. You thought it was good enough to spend many thousands of dollars on the filing; even if your estimate of its value declines by 90%, the $1000 issue fee is probably still a good bet. Wait 3.5 years, your estimate of its value might decline enough to not make it worth an additional slug of money. That’s one explanation; sometimes patents might get attached to projects that get cancelled — if the project is not worth funding, why would the associated patents be worth funding? (At one tech company where a friend worked, 60% of projects were cancelled before they could generate a shipping product.)

  49. 2

    “On the other side, IBM could have predicted at the time of issuance that it wouldn’t pay the maintenance fee — so why did it pay the $1,000 issue fee?”

    I agree with Paul, and I wonder if IBM has ever failed to pay an issue fee given the sheer number of patents it obtains each year.

    Also, if (for whatever reason) the application was never published, they at least get the benefit of a defensive disclosure for an incremental $1000 …

  50. 1

    A major factor is the PR value. Many corporations think that a large number of issued patents will convince the stock market that they are a high tech company worthy of a higher stock price.
    There is much less attention paid to patents being abandoned for non-payment of maintenance fees. Typically that is for technical obsolecence. Typically these days there are a lot of years passing between an application filing date and even a first patent maintenance fee date.
    IBM has an unusually effective system for saving money by doing maintenence fee pruning systematically. Other companies do not, and/or have somehow convinced themselves that others are tracking, and impressed by, the number of their patents remaining “in force” [not just issued]. I have not seen any studies justifying the latter rationale, but I have not really looked.
    Finally, some companies have separate licensing organizations that worry that a patent on which maintenance fees have not been paid might turn out years later to be valuable after all. Especially if those fees do not come out of their budget, but rather some other department’s budget.

Comments are closed.