Mlc Intellectual Property LLC v. Micron Technology, Inc. ( 2021 )


Menu:
  • Case: 20-1413   Document: 98     Page: 1   Filed: 08/26/2021
    United States Court of Appeals
    for the Federal Circuit
    ______________________
    MLC INTELLECTUAL PROPERTY, LLC,
    Plaintiff-Appellant
    v.
    MICRON TECHNOLOGY, INC.,
    Defendant-Appellee
    ______________________
    2020-1413
    ______________________
    Appeal from the United States District Court for the
    Northern District of California in No. 3:14-cv-03657-SI,
    Senior Judge Susan Y. Illston.
    ______________________
    Decided: August 26, 2021
    ______________________
    FABIO E. MARINO, Polsinelli PC, Palo Alto, CA, argued
    for plaintiff-appellant. Also represented by TERI HONG-
    PHUC NGUYEN.
    RUFFIN B. CORDELL, Fish & Richardson PC, Washing-
    ton, DC, argued for defendant-appellee. Also represented
    by MICHAEL JOHN BALLANCO, CHRISTOPHER DRYER,
    TIMOTHY W. RIFFE, ROBERT ANDREW SCHWENTKER, ADAM
    SHARTZER.
    WILLIAM F. LEE, Wilmer Cutler Pickering Hale and
    Dorr LLP, Boston, MA, for amici curiae Apple Inc., Dell
    Case: 20-1413    Document: 98      Page: 2    Filed: 08/26/2021
    2                         MLC INTELLECTIAL PROPERTY LLC V.
    MICRON TECHNOLOGY, INC.
    Inc., HP Inc., Intel Corporation. Also represented by
    BENJAMIN NOAH ERNST, MARK CHRISTOPHER FLEMING,
    LAUREN B. FLETCHER.
    ANDREW DUFRESNE, Perkins Coie LLP, Madison, WI,
    for amici curiae Computer & Communications Industry As-
    sociation, High Tech Inventors Alliance. Also represented
    by THOMAS ANDREW CULBERT, THERESA H. NGUYEN, Seat-
    tle, WA.
    PHILLIP R. MALONE, Juelsgaard Intellectual Property
    and Innovation Clinic, Mills Legal Clinic, Stanford Law
    School, Stanford, CA, for amici curiae Engine Advocacy,
    The R Street Institute. Also represented by ABIGAIL A.
    RIVES, Engine Advocacy, Washington, DC. Amicus curiae
    The R Street Institute also represented by CHARLES DUAN,
    R Street Institute, Washington, DC.
    ______________________
    Before NEWMAN, REYNA, and STOLL, Circuit Judges.
    STOLL, Circuit Judge.
    MLC Intellectual Property, LLC seeks interlocutory re-
    view of the United States District Court for the Northern
    District of California’s orders excluding certain opinions of
    MLC’s damages expert. For the reasons that follow, we af-
    firm the district court’s orders precluding MLC’s damages
    expert from characterizing certain license agreements as
    reflecting a 0.25% royalty, opining on a reasonable royalty
    rate when MLC failed to produce key documents and infor-
    mation directed to its damages theory when requested
    prior to expert discovery, and opining on the royalty base
    and royalty rate where the expert failed to apportion for
    non-patented features.
    Case: 20-1413     Document: 98      Page: 3   Filed: 08/26/2021
    MLC INTELLECTUAL PROPERTY LLC     v.                        3
    MICRON TECHNOLOGY, INC.
    BACKGROUND
    I
    MLC sued Micron for infringing certain claims of
    
    U.S. Patent No. 5,764,571
    . The ’571 patent, titled “Electri-
    cally Alterable Non-Volatile Memory with N-bits Per Cell,”
    describes methods of programming multi-level cells. The
    specification discloses that, in conventional single-bit per
    cell memory devices, a memory cell assumes either an “on”
    state or an “off” state, defining one bit of information.
    Thus, a memory device that stores n-bits of data requires n
    separate memory cells, meaning that the number of
    memory cells must increase on a one-for-one basis with the
    number of bits to be stored.
    The specification explains that an alternative approach
    to the single-bit per cell approach involves storing multi-
    ple-bits of data in a single memory cell, known as a multi-
    level cell. Prior approaches to multiple-bit per cell non-vol-
    atile memory have only used mask programmable read-
    only-memories (ROMs). This may be accomplished by var-
    ying the channel width or length of the memory cell “such
    that 2n different conductivity values are obtained which
    correspond to 2n different states corresponding to n-bits of
    data which can be stored on a single memory cell.” ’571 pa-
    tent col. 1 ll. 45–49. Another conventional ROM approach
    involves varying an ion implant for the threshold voltage
    “such that the memory cell will have 2n different voltage
    thresholds (Vt) corresponding to 2n different states corre-
    sponding to n-bits of data which can be stored on a single
    memory cell.” 
    Id.
     at col. 1 ll. 49–54. In these multi-bit
    ROM approaches, the 2n conductivity level must be deter-
    mined during the manufacturing process, and the memory
    can only be used for one data pattern. Thus, each time a
    data pattern needs to be changed, a new batch of semicon-
    ductor wafers must be processed.
    Conventional alterable multiple-bit per cell memories
    can store multiple levels of charge on a capacitive storage
    Case: 20-1413     Document: 98     Page: 4    Filed: 08/26/2021
    4                          MLC INTELLECTIAL PROPERTY LLC V.
    MICRON TECHNOLOGY, INC.
    element, such as dynamic random access memory (DRAM)
    or charge-coupled devices (CCDs). These approaches use
    volatile storage by providing “2n different volatile charge
    levels on a capacitor to define 2n different states corre-
    sponding to n-bits of data per memory cell.” 
    Id.
     at col. 2
    ll. 22–25. The problem with volatile storage is that a cell
    loses its data whenever power is removed, and cells must
    be periodically refreshed as they can lose charge over time.
    The ’571 patent purports to solve these problems in
    ROM and DRAM multiple-bit memories by disclosing a
    multi-bit semiconductor memory cell that has the non-vol-
    atile characteristics of ROM, as well as the electrically al-
    terable characteristics of a multi-bit per cell DRAM.
    Particularly, the specification describes a multi-bit per cell
    electrically alterable non-volatile memory (EANVM) where
    each cell stores information in Kn memory states, “where K
    is a base of a predetermined number system, n is a number
    of bits stored per cell, and Kn>2.” 
    Id.
     at col. 2 ll. 58–61.
    Moreover, the ’571 patent discloses programming the
    multi-level cell to a state corresponding to the input infor-
    mation and comparing the memory state of the multi-level
    cell with the input information, where the input infor-
    mation corresponds to a reference voltage.
    On appeal, MLC only asserts claim 30 of the ’571 pa-
    tent against Micron, which reads as follows:
    30. Apparatus for programming an electrically al-
    terable non-volatile memory cell having more than
    two predetermined memory states, comprising:
    a selecting device which selects one of a plurality of
    reference signals in accordance with information
    indicating a memory state to which said memory
    cell is to be programmed, each reference signal cor-
    responding to a different memory state of said
    memory cell;
    Case: 20-1413    Document: 98      Page: 5   Filed: 08/26/2021
    MLC INTELLECTUAL PROPERTY LLC    v.                       5
    MICRON TECHNOLOGY, INC.
    a programming signal source to apply a program-
    ming signal to said memory cell; and
    a control device to control the application of said
    programming signal to said memory cell based on
    the selected reference signal.
    
    Id.
     at col. 15 ll. 10–22. The scope of claim 30 is narrower
    than the scope of the other independent claims in the
    ’571 patent. While the other independent claims are di-
    rected to a “multi-level memory device” or a “multi-level
    memory apparatus,” claim 30 is more narrowly directed to
    an “[a]pparatus for programming an electrically alterable
    non-volatile memory cell having more than two predeter-
    mined memory states.” Compare, e.g., 
    id.
     at col. 12 l. 6,
    with 
    id.
     at col. 15 ll. 10–12.
    II
    Micron manufactures and sells NAND flash wafers and
    packages. Flash memory is a type of non-volatile memory,
    and NAND flash memory is a low-cost, high-density
    memory option. As such, NAND flash memory is consid-
    ered the standard for storage-related applications. Both
    Micron’s NAND flash wafer, or bare die assembly, and
    NAND flash package may include multiple dies. Included
    in each die is a memory array, which may comprise both
    single-level and multi-level memory cells. Micron assem-
    bles and sells its products in a variety of ways, including
    individually as wafers or in completed assemblies as pack-
    ages. Wafers may be used to make NAND flash packages,
    while flash packages encapsulate sorted functional dies
    that are connected to external leads in a plastic package.
    Micron contends that while a wafer having a single die is
    the smallest saleable patent practicing unit, there are nu-
    merous other non-infringing features in Micron’s die, in-
    cluding “error correction hardware,” “data clocking
    hardware,” “addressing hardware,” “cache registers,” and
    “digital to analog converters.” J.A. 1242.
    Case: 20-1413    Document: 98      Page: 6   Filed: 08/26/2021
    6                         MLC INTELLECTIAL PROPERTY LLC V.
    MICRON TECHNOLOGY, INC.
    III
    In his expert report, MLC’s damages expert, Michael
    Milani, first provided his understanding of the technology
    relevant to the ’571 patent. He explained that, based on
    his discussion with MLC’s technical expert, Dr. Jack Lee,
    he understood that the ’571 patent relates to technology
    that enables multi-level cell and triple-level cell flash
    memory.
    Mr. Milani next addressed the flash memory market as
    a whole, explaining that as the market became more satu-
    rated, production in the market shifted to NAND flash
    memory devices. Mr. Milani opined that by 2006, the
    NAND flash market had become a commodity market, with
    competitors mainly competing on price.
    Mr. Milani next explained that MLC was formed in
    2006 by Jerry Banks and Robert Hinkley. In 1997,
    Mr. Banks assigned to BTG International, Inc. the rights
    to a sizeable patent portfolio (the “MLCIP Patent Portfo-
    lio”), which included the ’571 patent among forty other pa-
    tents. BTG subsequently granted non-exclusive licenses to
    practice the MLCIP Patent Portfolio to Renesas Electronics
    Corporation in November 2006, Hynix Semiconductor Inc.
    in April 2007, and Toshiba Corporation in April 2007. Af-
    ter BTG sued Samsung Electronics Co. for infringement of
    certain patents in the MLCIP Patent Portfolio, BTG and
    Samsung entered into a settlement and license agreement
    in December 2010. MLC reacquired the rights to the
    MLCIP Patent Portfolio in 2012.
    With this background, Mr. Milani attempted to recon-
    struct the hypothetical negotiation between MLC and Mi-
    cron. Mr. Milani began by opining that the hypothetical
    negotiation date occurred in the fourth quarter of 2006,
    around the time that Micron first began selling the accused
    devices. He further opined that the compensation period
    began on August 12, 2008, six years prior to the filing of
    Case: 20-1413    Document: 98     Page: 7    Filed: 08/26/2021
    MLC INTELLECTUAL PROPERTY LLC   v.                        7
    MICRON TECHNOLOGY, INC.
    the complaint, and continued through the expiration of the
    ’571 patent on June 9, 2015.
    Turning next to the royalty base, Mr. Milani opined on
    two separate approaches for determining the royalty base:
    (1) a comparable license approach and (2) the smallest
    saleable patent practicing unit (SSPPU) approach. As for
    the comparable license approach, Mr. Milani included all
    of the revenue associated with the accused products, rea-
    soning that the royalty rate associated with the compara-
    ble license agreements already apportioned for other
    components in those products. As to the SSPPU approach,
    Mr. Milani began with the premise that the SSPPU is a
    bare die. Mr. Milani did not apportion the revenue from
    sales of a bare die or wafer. For sales of a non-SSPPU prod-
    uct, Mr. Milani apportioned the revenue by limiting the
    cost of each non-SSPPU product to the average selling price
    of each die. Thus, Mr. Milani ultimately applied an appor-
    tionment factor of approximately 87.4% of the total accused
    product revenue for the SSPPU approach.
    Mr. Milani next considered each of the factors set out
    in Georgia-Pacific Corp. v. U.S. Plywood Corp.,
    
    318 F. Supp. 1116
    , 1120 (S.D.N.Y. 1970), to determine an
    appropriate royalty rate. As to the first factor—“royalties
    received by the patentee for the licensing of the patent in
    suit, proving or tending to prove an established royalty,”
    id.—Mr. Milani considered BTG’s license agreements with
    Hynix and Toshiba to be the most relevant license agree-
    ments to consider in a hypothetical negotiation. Mr. Milani
    acknowledged that the Hynix license agreement required a
    lump sum of $21 million for the entire MLCIP Patent Port-
    folio rather than specifying a running royalty, but nonethe-
    less relied on a “most-favored customer” provision in the
    license to derive a royalty rate. The “most favored cus-
    tomer” provision states:
    In the event that BTG grants a license under the
    Licensed Patents after the Effective Date, other
    Case: 20-1413     Document: 98     Page: 8    Filed: 08/26/2021
    8                          MLC INTELLECTIAL PROPERTY LLC V.
    MICRON TECHNOLOGY, INC.
    than a license granted in settlement of litigation,
    in which the royalty is less than 0.25%, then as its
    sole remedy, Hynix’s future payments (if any) shall
    be reduced so that Hynix, in total pays not more
    than 90% of the royalty rate paid by the new licen-
    see.
    J.A. 1084. Mr. Milani explained that he considered the
    0.25% royalty rate called for in this provision “to reflect a
    relevant consideration for evaluating a reasonable royalty
    and underst[ood] that rate was applied to Hynix worldwide
    sales.” J.A. 906 (emphasis added). Turning to the Toshiba
    license agreement, Mr. Milani acknowledged that Toshiba
    paid BTG $25 million, but opined that, “given the most fa-
    vored customer provision in the Hynix Agreement, and the
    fact both agreements were executed on the same day, it’s
    reasonable to presume BTG considered the royalty rate in
    the Toshiba Agreement to reflect a running royalty that is
    at least equal to the rate reflected by the Hynix Agree-
    ment.” J.A. 907.
    As to the third factor—“[t]he nature and scope of the
    license, as exclusive or non-exclusive; or as restricted or
    non-restricted in terms of territory or with respect to whom
    the manufactured product may be sold,” Georgia-Pacific,
    
    318 F. Supp. at
    1120—Mr. Milani determined that, be-
    cause the Hynix agreement was based upon worldwide
    shipments, the “0.25% royalty rate reflected within the
    Hynix agreement” was a discounted royalty rate to account
    for the fact that the MLCIP Patent Portfolio was predomi-
    nantly made up of U.S. rights. J.A. 913. Specifically,
    Mr. Milani relied on testimony that the U.S. sales reflected
    only a third of a licensee’s total shipments to determine
    that the proper rate to apply to U.S. sales in this case would
    be 0.75%. Accordingly, Mr. Milani concluded that “the
    Hynix Agreement suggests a royalty rate of 0.75% is the
    proper rate to consider in connection with determining a
    reasonable royalty in a hypothetical negotiation.” 
    Id.
    Case: 20-1413    Document: 98      Page: 9    Filed: 08/26/2021
    MLC INTELLECTUAL PROPERTY LLC   v.                         9
    MICRON TECHNOLOGY, INC.
    Considering the seventh factor—“[t]he duration of the
    patent and the term of the license,” Georgia-Pacific,
    
    318 F. Supp. at
    1120—Mr. Milani recognized that the
    Hynix agreement would run from April 11, 2007, to the ex-
    piration of the ’571 patent, June 9, 2015, or approximately
    eight years. Nonetheless, he determined that this factor
    was neutral “relative to the rate reflected in the Hynix
    agreement.” J.A. 915.
    Mr. Milani also addressed factors nine and ten—“[t]he
    utility and advantages of the patent property over the old
    modes or devices, if any, that had been used for working
    out similar results” and “[t]he nature of the patented in-
    vention; the character of the commercial embodiment of it
    as owned and produced by the licensor; and the benefits to
    those who have used the invention,” Georgia-Pacific,
    
    318 F. Supp. at 1120
    . In particular, Mr. Milani explained
    that, in his view, “the features and benefits of the
    ’571 [p]atent and the existence or acceptability of potential
    non-infringing alternatives would have been implicitly ac-
    counted for in the negotiation between Hynix and BTG.”
    J.A. 919.
    Considering the twelfth factor—“[t]he portion of the
    profit or of the selling price that may be customary in the
    particular business or in comparable businesses to allow
    for the use of the invention or analogous inventions,” Geor-
    gia-Pacific, 
    318 F. Supp. at
    1120—Mr. Milani again stated
    that “the Hynix Agreement reflects a 0.25% royalty,”
    J.A. 921, and identified extrinsic evidence to support his
    contention that the Hynix agreement reflected a 0.25% roy-
    alty rate. Specifically, Mr. Milani relied on (1) a BTG
    “Briefing Paper” summarizing negotiations with Samsung
    and stating that the parties discussed a 0.25% royalty rate;
    (2) a BTG letter offering to license the MLCIP Patent Port-
    folio to STMicroelectronics, Inc. at a 0.25% royalty rate;
    (3) a BTG letter offering to license the MLCIP Patent Port-
    folio to Micron at a 0.25% royalty rate; and (4) a BTG mem-
    orandum prepared during negotiations with Acacia Patent
    Case: 20-1413    Document: 98      Page: 10   Filed: 08/26/2021
    10                        MLC INTELLECTIAL PROPERTY LLC V.
    MICRON TECHNOLOGY, INC.
    Acquisition LLC, which indicated that the Toshiba and
    Hynix agreements used an effective royalty rate of 0.25%.
    As to the thirteenth factor—“[t]he portion of the realiz-
    able profit that should be credited to the invention as dis-
    tinguished      from     non-patented      elements,      the
    manufacturing process, business risks, or significant fea-
    tures or improvements added by the infringer,” Georgia-
    Pacific, 
    318 F. Supp. at
    1120—Mr. Milani explained that
    he believed it reasonable to presume that at least fifty per-
    cent of the licensing value of the MLCIP Patent Portfolio
    was attributable to the ’571 patent, and so he adjusted the
    0.75% royalty to 0.375%. Mr. Milani relied on testimony
    from Dr. Lee, who stated that “the vast majority of the tech-
    nical value of the MLCIP portfolio is attributable to the
    ’571 patent,” J.A. 2194, notwithstanding the fact that the
    MLCIP Patent Portfolio consisted of forty-one patent as-
    sets, including twenty-eight granted U.S. Patents, two U.S.
    Patent Applications, ten foreign granted patents, and one
    foreign patent application.
    Considering the fifteenth factor—“[t]he amount that a
    licensor (such as the patentee) and a licensee (such as the
    infringer) would have agreed upon (at the time the in-
    fringement began) if both had been reasonably and volun-
    tarily trying to reach an agreement,” Georgia-Pacific,
    
    318 F. Supp. at
    1120—Mr. Milani applied the royalty rate
    of 0.375% to the royalty bases calculated under the compa-
    rable licensing approach and the SSPPU approach, arriv-
    ing at lump sum payments of $70,207,876 and $63,142,053,
    respectively.
    IV
    Micron filed a motion in limine to preclude Mr. Milani
    from mischaracterizing the Hynix and Toshiba agreements
    as reflecting a 0.25% royalty rate. In addition, Micron
    moved to strike portions of Mr. Milani’s expert report un-
    der Rule 37 of the Federal Rules of Civil Procedure as being
    based on facts, evidence, and theories that MLC disclosed
    Case: 20-1413    Document: 98      Page: 11    Filed: 08/26/2021
    MLC INTELLECTUAL PROPERTY LLC    v.                        11
    MICRON TECHNOLOGY, INC.
    for the first time in Mr. Milani’s expert report. Micron had
    asked for MLC’s damages theories—as well as any facts,
    evidence, and testimony regarding any applicable royalty
    rate—during fact discovery in its Interrogatory Nos. 6 and
    22 and during a Rule 30(b)(6) deposition of a corporate de-
    signee. Finally, Micron filed a Daubert motion, seeking to
    exclude Mr. Milani’s reasonable royalty opinion for failure
    to apportion out the value of non-patented features. The
    district court granted all three motions.
    First, the district court granted in part Micron’s motion
    in limine, holding that “[Mr.] Milani may not testify that
    the Hynix and Toshiba agreements contain or ‘reflect’ spe-
    cific royalty rates because the documents speak for them-
    selves and neither provides for an applicable royalty rate.”
    MLC Intell. Prop., LLC v. Micron Tech., Inc. (MLC I),
    No. 14-CV-03657, 
    2019 WL 2863585
    , at *13 (N.D. Cal.
    July 2, 2019). The court reasoned that Mr. Milani’s “testi-
    mony about the Hynix and Toshiba licenses containing a
    0.25% royalty rate is not ‘based on sufficient facts or data’
    and is not ‘the product of reliable principles and methods.’”
    
    Id.
     (quoting Fed. R. Evid. 702). The district court rejected
    Mr. Milani’s reliance on the “most favored customer” pro-
    vision to establish that the Hynix and Toshiba agreements
    reflected a 0.25% royalty rate. 
    Id.
     The court reasoned that
    the most favored customer provision did not state that the
    0.25% royalty rate was applied to calculate the lump sum
    payment in either the Hynix or Toshiba license; nor did the
    licenses provide any insight as to how the lump sum pay-
    ments were actually calculated. 
    Id.
     The district court also
    noted that had the lump sum payments been calculated
    based on the actual term of the license (2007–2017), “the
    effective royalty rate would be less than 0.25%.” 
    Id.
    Second, the district court granted in part Micron’s mo-
    tion to strike portions of Mr. Milani’s expert report under
    Rule 37(c)(1). MLC Intell. Prop., LLC v. Micron Tech., Inc.
    (MLC II), No. 14-CV-03657 (N.D. Cal. July 12, 2019), ECF
    No. 672 (incorporating its reasoning in MLC I, 2019 WL
    Case: 20-1413    Document: 98      Page: 12    Filed: 08/26/2021
    12                         MLC INTELLECTIAL PROPERTY LLC V.
    MICRON TECHNOLOGY, INC.
    2863585, at *14–15). The court concluded that “MLC never
    disclosed the factual underpinnings of its claim that the
    Hynix and Toshiba licenses ‘reflect’ a 0.25% royalty rate,
    and that pursuant to Rule 37(c)(1), this failure is a separate
    and independent basis for excluding evidence and argu-
    ment that those licenses contain such a rate.” MLC I,
    
    2019 WL 2863585
    , at *14. The court further held that the
    extrinsic evidence relied on by Mr. Milani to opine that the
    Hynix and Toshiba licenses reflect a 0.25% royalty rate
    “was never disclosed by MLC and thus MLC may not rely
    on this evidence to assert that the Hynix and Toshiba li-
    censes ‘reflect’ a 0.25% rate.” 
    Id. at *15
    .
    Finally, the district court granted Micron’s Daubert
    motion to exclude the expert testimony of Mr. Milani re-
    garding the royalty base. MLC Intell. Prop., LLC v. Micron
    Tech., Inc. (MLC III), No. 14-CV-03657, 
    2019 WL 3070567
    ,
    at *1 (N.D. Cal. July 12, 2019). The district court initially
    noted that MLC defended Mr. Milani’s comparable license
    approach and his SSPPU approach to calculating the roy-
    alty base by arguing that the royalty rate from the Hynix
    license already addressed apportionment and thus appor-
    tionment was accounted for in the 0.25% royalty rate. 
    Id. at *2
    . The district court determined that there was “no ev-
    idence regarding the Hynix agreement that supports
    [Mr.] Milani’s opinion that a specific royalty rate derived
    from the Hynix agreement already accounts for apportion-
    ment of non-patented features in Micron’s accused prod-
    ucts and thus can be applied to all the revenue for Micron’s
    accused products.” 
    Id. at *3
    . The district court based its
    conclusion in part on the fact that the Hynix agreement did
    not actually contain a royalty rate and there was no expla-
    nation as to how the lump sum in the Hynix agreement was
    determined. 
    Id. at *2
    . Notably, it found relevant that
    Mr. Milani failed to compare Micron’s accused products to
    the licensed Hynix products, especially in view of the fact
    that the Hynix agreement covered worldwide rights to
    forty-one patents for all of the Hynix products. 
    Id.
     As to
    Case: 20-1413    Document: 98      Page: 13   Filed: 08/26/2021
    MLC INTELLECTUAL PROPERTY LLC     v.                      13
    MICRON TECHNOLOGY, INC.
    the SSPPU approach, the district court additionally criti-
    cized Mr. Milani’s approach because he failed to apportion
    for non-infringing features, such as “error-correction soft-
    ware and implementation of copy-back technology.” 
    Id. at *3
    . The court held that Mr. Milani’s failure to apportion
    for these non-patented technologies rendered his analysis
    unreliable and excludable. 
    Id.
    Thereafter, the district court certified the preceding
    three damages orders for interlocutory appeal pursuant to
    
    28 U.S.C. § 1292
    (b). MLC Intell. Prop., LLC v. Micron
    Tech., Inc. (MLC IV), No. 14-CV-03657, 
    2019 WL 5269014
    ,
    at *5 (N.D. Cal. Oct. 17, 2019). We granted MLC’s petition
    for permission to appeal. MLC Intell. Prop., LLC v. Micron
    Tech., Inc. (MLC V), 794 F. App’x 951, 953 (Fed. Cir. 2020).
    We have jurisdiction under 
    28 U.S.C. § 1292
    (b).
    DISCUSSION
    “When reviewing damages in patent cases, we apply re-
    gional circuit law to procedural issues and Federal Circuit
    law to substantive and procedural issues pertaining to pa-
    tent law.” Whitserve, LLC v. Comput. Packages, Inc.,
    
    694 F.3d 10
    , 26 (Fed. Cir. 2012) (quoting Wordtech Sys.,
    Inc. v. Integrated Networks Sols., Inc., 
    609 F.3d 1308
    , 1318
    (Fed. Cir. 2010)). The Ninth Circuit reviews “a district
    court’s evidentiary rulings, such as its decisions to exclude
    expert testimony and to impose discovery sanctions, for an
    abuse of discretion.” Ollier v. Sweetwater Union High Sch.
    Dist., 
    768 F.3d 843
    , 859 (9th Cir. 2014); see also Gen. Elec.
    Co. v. Joiner, 
    522 U.S. 136
    , 141 (1997) (“[A]buse of discre-
    tion is the proper standard of review of a district court’s
    evidentiary rulings.”).
    On appeal, MLC challenges all three of the district
    court’s exclusion orders. We address each in turn below.
    I
    MLC first argues that the district court erred in exclud-
    ing Mr. Milani’s opinion that the Hynix and Toshiba
    Case: 20-1413    Document: 98      Page: 14    Filed: 08/26/2021
    14                         MLC INTELLECTIAL PROPERTY LLC V.
    MICRON TECHNOLOGY, INC.
    agreements reflect a 0.25% royalty rate. We are not per-
    suaded. As a gatekeeper, the district judge was required
    to “ensure that any and all scientific testimony or evidence
    admitted is not only relevant, but reliable.” Daubert
    v. Merrell Dow Pharms., Inc., 
    509 U.S. 579
    , 589 (1993). In-
    deed, the Federal Rules of Evidence “leave in place the
    ‘gatekeeper’ role of the trial judge in screening such evi-
    dence.” Gen. Elec., 
    522 U.S. at 142
    . In this case, the dis-
    trict court properly determined that Mr. Milani’s
    “testimony about the Hynix and Toshiba licenses contain-
    ing a 0.25% royalty rate [was] not ‘based on sufficient facts
    or data’ and [was] not ‘the product of reliable principles and
    methods.’” 1 MLC I, 
    2019 WL 2863585
    , at *13 (quoting
    Fed. R. Evid. 702).
    We see no abuse of discretion in the district court’s ex-
    clusion of Mr. Milani’s testimony that the Hynix and
    Toshiba agreements contain or reflect a specific royalty
    rate of 0.25%. Instead of resting on an accepted scientific
    theory or technique, Mr. Milani’s testimony that he under-
    stood the Hynix and Toshiba licenses to use a 0.25% royalty
    is not sufficiently tethered to the evidence presented. Nei-
    ther the Hynix agreement nor the Toshiba agreement dis-
    closes any royalty rate. Rather than deriving a rate from
    the lump-sum payments and projected sales, Mr. Milani’s
    testimony rests on an inference from the most favored cus-
    tomer clause that goes well beyond what the clause implies
    and is incompatible with the Hynix agreement as a whole.
    As the district court pointed out, if a 0.25% royalty had
    1  MLC also argues that the district court erred in ex-
    cluding Mr. Milani’s expert testimony under the parol evi-
    dence rule, an alternative ground addressed by the district
    court in a footnote. See MLC I, 
    2019 WL 2863585
    , at *13
    n.14. Because we affirm on the primary ground identified
    by the district court, we do not reach the parol evidence
    issue.
    Case: 20-1413    Document: 98      Page: 15    Filed: 08/26/2021
    MLC INTELLECTUAL PROPERTY LLC    v.                        15
    MICRON TECHNOLOGY, INC.
    been applied to forecasts of revenue for the term of the li-
    censes (2007–2017), the lump-sum amounts would have
    been far greater than $21 and $25 million. See MLC I,
    
    2019 WL 2863585
    , at *13.
    We reached a similar conclusion based on similar facts
    in Whitserve, where we determined that “multiple errors in
    [the expert’s] royalty rate calculation cause[d] his ultimate
    opinion regarding a reasonable royalty rate to be specula-
    tive.” 694 F.3d at 29. In that case, the expert identified
    two lump-sum payments in the range of approximately
    $2–3 million as evidence to support an increased royalty
    rate of 19%. Id. at 30. Because the expert offered no testi-
    mony as to how those lump-sum payments could be con-
    verted to any royalty rate, let alone a 19% royalty rate, we
    affirmed the district court’s determination that the lump-
    sum agreements did not support a 19% running royalty.
    Id. The same rationale supports the district court’s deter-
    mination here.
    We addressed the fundamental differences between
    lump-sum agreements and running-royalty agreements in
    Lucent Technologies, Inc. v. Gateway, Inc., 
    580 F.3d 1301
    (Fed. Cir. 2009). In Lucent, we explained that, “[f]or a jury
    to use a running-royalty agreement as a basis to award
    lump-sum damages . . . , some basis for comparison must
    exist in the evidence presented to the jury.” 
    Id. at 1330
    .
    Lucent cited four running-royalty licenses to support a
    lump-sum damages award. 
    Id. at 1329
    . Because Lucent
    provided almost no testimony for the jury to recalculate the
    value of the royalty agreements to arrive at the lump-sum
    damages award, we determined that the running royalties
    disclosed in the agreements did not support the award. 
    Id. at 1330
    . This case involves a similar problem, though in
    the mirror-image situation where lump-sum agreements
    are being asserted as a basis to infer a rate for running roy-
    alty. Because Mr. Milani did not provide mathematical
    analysis to derive the 0.25% royalty rate from the lump-
    sum payments in the Hynix and Toshiba licenses, the
    Case: 20-1413    Document: 98      Page: 16   Filed: 08/26/2021
    16                        MLC INTELLECTIAL PROPERTY LLC V.
    MICRON TECHNOLOGY, INC.
    district court could reasonably determine that those li-
    censes cannot support testimony that the lump-sum pay-
    ments were, in fact, based on that royalty rate.
    We acknowledge that Mr. Milani’s testimony may well
    have been proper had he merely asserted that he “con-
    sider[ed] the 0.25% royalty rate called for in the most fa-
    vored customer provision to reflect a relevant consideration
    for evaluating a reasonable royalty.” J.A. 906. But he
    crossed the line when he stated that he “under[stood] that
    [the 0.25%] rate was applied to Hynix worldwide sales” in
    calculating the lump-sum license payment of $21 million.
    
    Id.
     As the expert failed to do in Whitserve, Mr. Milani of-
    fered no testimony as to how the $21 million lump-sum
    payment could be converted to any royalty rate, let alone a
    0.25% royalty rate. See 694 F.3d at 30. Nonetheless,
    Mr. Milani repeatedly took the unsupported position that
    the 0.25% royalty rate represented the lump sum amounts
    in both agreements. J.A. 921 (the “Hynix Agreement re-
    flects a 0.25% royalty applied to worldwide shipments”);
    J.A. 926 (“I also recognize the lump-sum payments in-
    cluded in the BTG license agreements reflect the applica-
    tion of the 0.25% royalty rate reflected in the agreements
    to a royalty base comprised of estimated worldwide sales.”);
    J.A. 1145 (Milani Dep. 146:21–24) (“I would tell the jury
    that when this agreement was negotiated, that the parties
    to the agreement considered 0.25 percent to be the effective
    royalty rate to which they were agreeing.”). Accordingly,
    we agree with the district court that Mr. Milani’s charac-
    terization of the Hynix and Toshiba licenses reflecting a
    0.25% royalty rate is not reliable and therefore affirm the
    district court’s exclusion of his opinion.
    II
    We next turn to the district court’s order striking por-
    tions of Mr. Milani’s expert report under Rule 37(c)(1) of
    the Federal Rules of Civil Procedure. Under Rule 37(c)(1),
    when “a party fails to provide information or identify a
    Case: 20-1413    Document: 98      Page: 17    Filed: 08/26/2021
    MLC INTELLECTUAL PROPERTY LLC    v.                        17
    MICRON TECHNOLOGY, INC.
    witness as required by Rule 26(a) or (e), the party is not
    allowed to use that information or witness to supply evi-
    dence on a motion, at a hearing, or at a trial, unless the
    failure was substantially justified or is harmless.” Here,
    the district court explained that MLC failed to disclose in-
    formation required by Rule 26(e), including what it be-
    lieved was an appropriate royalty rate, that it believed the
    Hynix and Toshiba licenses reflect a 0.25% royalty rate,
    and the extrinsic evidence on which Mr. Milani relied in
    support of his belief. MLC I, 
    2019 WL 2863585
    , at *14.
    The court held that Mr. Milani could not opine that the li-
    censes reflected a 0.25% royalty rate where MLC had failed
    to disclose in discovery all of the evidence that Mr. Milani
    relied on in support of that opinion.
    On appeal, MLC primarily argues that it was not re-
    quired to disclose these specific facts and documents sup-
    porting its damages theory during fact discovery because it
    ultimately disclosed them during expert discovery. Specif-
    ically, MLC reasons that it provided adequate responses to
    Micron’s Interrogatory Nos. 6 and 22, and that anything
    more would have required it to disclose material desig-
    nated for expert discovery.
    Interrogatory No. 6 asked MLC to describe “the factual
    and legal basis and supporting evidence for the relief”
    sought by MLC, “including but not limited to [MLC’s] con-
    tention that [it is] entitled to damages (e.g., a reasonable
    royalty).” J.A. 2618. Besides boilerplate objections, MLC
    responded that its “calculation of damages will also be in-
    formed by, at least, the following documents,” and then pro-
    vided a list of ninety-three Bates-numbered documents.
    J.A. 2620. In a second supplemental response, MLC added
    that the “royalty rate will be based on at least the Georgia-
    Pacific factors, and will include but not [be] limited to con-
    sideration of relevant license agreements for the patented
    technology, including those identified in MLC’s prior re-
    sponse, as well as any prior negotiations between the par-
    ties regarding the patented technology.” J.A. 2621–22.
    Case: 20-1413    Document: 98      Page: 18    Filed: 08/26/2021
    18                         MLC INTELLECTIAL PROPERTY LLC V.
    MICRON TECHNOLOGY, INC.
    Interrogatory No. 22 asked MLC to “[i]dentify all facts,
    evidence, and testimony regarding any applicable royalty
    rates that [MLC] intend[s] to rely upon at trial and describe
    in complete detail why those royalty rates are applicable.”
    J.A. 1186. MLC responded that the “royalty rate will be
    based on at least the Georgia-Pacific factors, and will in-
    clude but not [be] limited to consideration of license agree-
    ments for the patented technology, including but not
    limited to” seven bates-numbered documents. J.A. 1187.
    In its order granting Micron’s motion, the district court
    noted that MLC’s response to Interrogatory No. 6 failed to
    identify the Hynix license, the Toshiba license, its reason-
    able-royalty theory, and any of the extrinsic evidence relied
    on by Mr. Milani to support his opinion that both the Hynix
    and Toshiba licenses reflect a 0.25% royalty rate. MLC I,
    
    2019 WL 2863585
    , at *8. Likewise, the district court noted
    that MLC’s response to Interrogatory No. 22 failed to iden-
    tify the Toshiba license, a specific royalty rate, that it be-
    lieved the Hynix or Toshiba licenses supported a 0.25% (or
    0.75%) royalty rate, and any of the extrinsic evidence relied
    on by Mr. Milani to support his opinion that the Hynix and
    Toshiba licenses reflect a 0.25% royalty. 
    Id. at *9
    . The
    district court also noted that, when asked about the royalty
    rate used in the Toshiba and Hynix agreements, MLC’s
    30(b)(6) corporate designee testified that MLC had no un-
    derstanding of what royalty rate was in the agreements.
    
    Id.
     at *9–11.
    On appeal, MLC asserts that it did identify both the
    Hynix and Toshiba licenses, as well as several of the ex-
    trinsic documents Mr. Milani relied on. MLC explains that
    it just used different Bates numbers than the ones refer-
    enced by Mr. Milani in his report. Micron does not chal-
    lenge that MLC identified both the Hynix and Toshiba
    licenses. Instead, it asserts that MLC failed to identify a
    number of other documents, including the extrinsic evi-
    dence relied on by Mr. Milani to show that the Hynix agree-
    ment reflects a 0.25% royalty rate—notably, the documents
    Case: 20-1413    Document: 98      Page: 19     Filed: 08/26/2021
    MLC INTELLECTUAL PROPERTY LLC     v.                        19
    MICRON TECHNOLOGY, INC.
    related to BTG’s negotiations with Samsung, STMicroelec-
    tronics, and Acacia. MLC concedes that these documents
    were not disclosed in its responses to Interrogatory Nos. 6
    and 22, but asserts that at least some of the documents
    were disclosed in response to other interrogatories.
    The Ninth Circuit affords district courts “particularly
    wide latitude” in applying Rule 37(c)(1) to exclude infor-
    mation that a party failed to provide under Rule 26. In-
    genco Holdings, LLC v. Ace Am. Ins. Co., 
    921 F.3d 803
    , 821
    (9th Cir. 2019). While we acknowledge the district court’s
    factual error in finding that MLC did not identify the
    Toshiba and Hynix licenses, we nonetheless determine that
    the district court did not abuse its discretion in finding that
    MLC did not properly disclose its claim that the Hynix and
    Toshiba licenses reflect a 0.25% rate, as well as the extrin-
    sic documents relied on by Mr. Milani to show that the
    Hynix agreement reflects a 0.25% royalty rate. The district
    court acted well within its discretion when it excluded
    Mr. Milani’s opinion that the Hynix and Toshiba licenses
    reflect a 0.25% rate and the extrinsic documents under
    Rule 37(c)(1) as a result of MLC’s failure to supplement its
    discovery responses to provide this information. As the dis-
    trict court emphasized, because the Hynix and Toshiba li-
    censes are lump-sum agreements that do not contain
    specific royalty rates, absent a disclosure by MLC, Micron
    would have no way of knowing that Mr. Milani would opine
    that the agreements reflect a 0.25% royalty rate, particu-
    larly given the Rule 30(b)(6) testimony indicating that
    MLC did not know the royalty rate in the Hynix and
    Toshiba agreements. 
    Id.
     Further, we agree with the dis-
    trict court that, had MLC disclosed this information, Mi-
    cron could have sought fact discovery regarding this
    contention. Finally, it is worth noting that the Hynix and
    Toshiba agreements and the extrinsic evidence were pro-
    duced by MLC and in its possession from the outset of the
    case, largely mitigating any timing and fairness concerns
    Case: 20-1413    Document: 98      Page: 20    Filed: 08/26/2021
    20                         MLC INTELLECTIAL PROPERTY LLC V.
    MICRON TECHNOLOGY, INC.
    preventing MLC from knowing its contentions about a rea-
    sonable royalty until after close of fact discovery.
    We find unpersuasive MLC’s argument that, under
    Rule 26(a)(2) of the Federal Rules of Civil Procedure—
    which governs the disclosure of expert testimony—it was
    outside of the district court’s discretion to require MLC to
    identify which documents Mr. Milani would be using, how
    Mr. Milani would interpret those documents, and how
    Mr. Milani would use them to derive his reasonable royalty
    opinion. According to MLC, all that the discovery rules re-
    quire is that those theories be developed in the expert’s re-
    port. We reject this narrow reading of Rule 26, as well as
    the implied narrow reading of the district court’s discretion
    to interpret discovery rules. Rule 26 explains that the dis-
    closures required under section (a)(2) are in “addition to
    the disclosures required by Rule 26(a)(1).” Fed. R. Civ. P.
    26(a)(2) (emphasis added). And Rule 26(a)(1)(A)(iii) re-
    quires parties seeking damages to provide in their initial
    disclosures “a computation of each category of damages” as
    well as “the documents or other evidentiary material, un-
    less privileged or protected from disclosure, on which each
    computation is based.” 2
    2   The Northern District of California has previously
    explained that, with regard to reasonable royalty, though
    Rule 26(a) “does not require a full exposition of the type
    required at trial or in an expert report,” it “does expressly
    require an initial computation and disclosure of the evi-
    dence that will be relied on to the full extent the patent
    plaintiff could or should know of it.”            Brandywine
    Commc’ns Techs., LLC v. Cisco Sys., Inc., No. C 12-01669,
    
    2012 WL 5504036
    , at *2 (N.D. Cal. Nov. 13, 2012). Thus,
    to the extent possible, this initial disclosure should include
    a claimed royalty rate and the evidence that supports such
    a rate, “even though subsequent discovery may eventually
    warrant a modification of the calculation.” 
    Id.
    Case: 20-1413    Document: 98      Page: 21     Filed: 08/26/2021
    MLC INTELLECTUAL PROPERTY LLC     v.                        21
    MICRON TECHNOLOGY, INC.
    Thus, in our view, the district court was within its dis-
    cretion in determining that, though MLC was not required
    to disclose its expert opinions during fact discovery, it was
    still required to disclose (1) its view that the Hynix and
    Toshiba license agreements reflect a 0.25% royalty rate
    and (2) the extrinsic evidence Mr. Milani relied on to sup-
    port that view in response to Micron’s reasonable requests
    for all facts, evidence, and testimony regarding any appli-
    cable royalty rates that MLC intended to rely on at trial.
    The Ninth Circuit has previously enforced sanctions for
    failing to timely disclose damages theories. In Ingenco
    Holdings, LLC v. Ace American Insurance Co., the district
    court did not permit Ingenco “to rely upon any computation
    or evidence of their [damages claims] beyond that provided
    in their initial disclosures” for failing to produce these dam-
    ages theories until the day before discovery cutoff at a
    Rule 30(b)(6) deposition. No. C13-543, 
    2016 WL 4703758
    ,
    at *2, *5 (W.D. Wash. Sept. 7, 2016). The Ninth Circuit
    affirmed this exclusion, explaining that because Ingenco
    failed to respond to Ace’s interrogatory requesting this
    damages information, and because Rule 26(a)(1)(A)(iii) re-
    quires timely disclosure of damages information, the dis-
    trict court properly excluded the damages information
    under Rule 37. Ingenco, 921 F.3d at 821–22; see also Elliott
    v. Google, Inc., 
    860 F.3d 1151
    , 1161 (9th Cir. 2017) (affirm-
    ing the district court’s exclusion of evidence where the evi-
    dence was not properly disclosed during fact discovery);
    Am. Cas. Co. of Reading, Pa. v. Baker, 
    22 F.3d 880
    , 886 n.3
    (9th Cir. 1994) (explaining that “the district court was
    within its discretion by precluding expert testimony as a
    sanction for the [] failure to seasonably supplement its in-
    terrogatory responses”); Silvia v. MCI Commc’ns Servs.,
    Inc., 787 F. App’x 399, 400 (9th Cir. 2019) (affirming the
    district court’s exclusion of Silvia’s damages theory for fail-
    ing to timely supplement or correct her disclosures and dis-
    covery responses under Rule 26(e)(1)(A)).
    Case: 20-1413    Document: 98     Page: 22    Filed: 08/26/2021
    22                        MLC INTELLECTIAL PROPERTY LLC V.
    MICRON TECHNOLOGY, INC.
    MLC’s argument that it need not disclose factual un-
    derpinnings and evidence underlying its damages theory
    prior to expert discovery undermines a district court’s dis-
    cretion to encourage early discovery. District courts have
    the discretion to encourage parties to provide discovery of
    damages theories prior to expert discovery. Doing so pro-
    motes judicial efficiency, informs settlement discussions,
    and helps parties determine the resources that will be de-
    voted to a case based on its potential value. Consistent
    with these goals, several district courts have adopted local
    rules requiring parties to provide this information early in
    the litigation. For example, the Northern District of Cali-
    fornia amended its local rules in 2017 to require each party
    asserting infringement to disclose its damages “theories of
    recovery, factual support for those theories, and computa-
    tions of damages within each category,” within fifty days
    after service of the invalidity contentions. U.S. Dist. Ct.
    N.D. Cal. Patent L.R. 3-8. 3
    MLC’s argument also undermines Rule 33 of the Fed-
    eral Rules of Civil Procedure. Rule 33 states that “[a]n in-
    terrogatory is not objectionable merely because it asks for
    an opinion or contention that relates to fact or the applica-
    tion or law to fact.” Fed. R. Civ. P. 33. As we recognized in
    Woods v. DeAngelo Marine Exhaust, Inc., contention inter-
    rogatories—like Interrogatory Nos. 6 and 22 here—“serve
    an important purpose in helping to discover facts support-
    ing the theories of the parties. Answers to contention in-
    terrogatories also serve to narrow and sharpen the issues
    thereby confining discovery and simplifying trial prepara-
    tion.” 
    692 F.3d 1272
    , 1280 (Fed. Cir. 2012) (citing Fed. R.
    Civ. P. 33 advisory committee’s note (1970 Amendment,
    Subdivision (b)). We have recognized that answers to con-
    tention interrogatories evolve over time as theories of
    3  We note that the district court did not rely on the
    local rules in its Rule 37(c)(1) ruling in this case.
    Case: 20-1413    Document: 98      Page: 23    Filed: 08/26/2021
    MLC INTELLECTUAL PROPERTY LLC    v.                        23
    MICRON TECHNOLOGY, INC.
    liability, defense, and relief begin to take shape and that
    answers to those interrogatories may not come into focus
    until the end of discovery. See Woods, 692 F.3d at 1280
    (citing O2 Micro Int’l Ltd. v. Monolithic Power Sys., Inc.,
    
    467 F.3d 1355
    , 1365 (Fed. Cir. 2006)). But Rule 26(e) ex-
    pressly requires that as theories mature and as the rele-
    vance of various items of evidence changes, responses to
    interrogatories, and particularly contention interrogato-
    ries, must be corrected or supplemented to reflect those
    changes.
    Finally, we are unpersuaded by MLC’s argument that
    the district court’s order required MLC’s corporate de-
    signee in the Rule 30(b)(6) deposition to divulge privileged
    information when asked about MLC’s view of the royalty
    rate in the Hynix agreement. Likewise, we do not agree
    that the district court erred by requiring MLC to disclose
    privileged information in its interrogatory responses. In
    striking portions of Mr. Milani’s expert opinion on reason-
    able royalty under Rule 37(c)(1), the district court ex-
    plained that “Micron repeatedly asked MLC . . . for the
    factual basis of its reasonable royalty claim and about its
    reliance on the Hynix license in particular – and MLC con-
    sistently failed to disclose its contention that the Hynix li-
    cense ‘reflected’ a 0.25% royalty rate that should be applied
    to this case.” MLC I, 
    2019 WL 2863585
    , at *14.
    A request for information regarding the factual basis of
    MLC’s reasonable royalty claim does not seek privileged in-
    formation. Attorney-client privilege “only protects disclo-
    sure of communications; it does not protect disclosure of the
    underlying facts” of those communications. Upjohn Co.
    v. United States, 
    449 U.S. 383
    , 395 (1981); see also Mur-
    doch v. Castro, 
    609 F.3d 983
    , 995 (9th Cir. 2010) (explain-
    ing that a witness may be questioned “about the underlying
    facts” because “attorney-client privilege protects only a
    communication between an attorney and a client, not the
    facts that are communicated”); Castaneda v. Burger King
    Corp., 
    259 F.R.D. 194
    , 197 (N.D. Cal. 2009) (explaining
    Case: 20-1413    Document: 98       Page: 24   Filed: 08/26/2021
    24                        MLC INTELLECTIAL PROPERTY LLC V.
    MICRON TECHNOLOGY, INC.
    that though documents may be privileged, “the facts con-
    tained within them may not be” and that attorney-client
    privilege does not protect the disclosure of underlying
    facts); accord In re Pioneer Hi-Bred Int’l, Inc., 
    238 F.3d 1370
    , 1374 (Fed. Cir. 2001). That the Hynix and Toshiba
    licenses reflect a 0.25% rate, along with the statements in
    the extrinsic documents supporting this rate, are nothing
    more than facts underlying MLC’s damages theory. We
    therefore affirm the district court’s order striking portions
    of Mr. Milani’s expert opinion.
    III
    Finally, we affirm the district court’s grant of Micron’s
    Daubert motion to exclude Mr. Milani’s expert opinion on
    reasonable royalty for failure to apportion. We agree that
    Mr. Milani did not properly apportion either the royalty
    base or the royalty rate to account for the patented tech-
    nology. 4 We have repeatedly held that when the accused
    technology does not make up the whole of the accused prod-
    uct, apportionment is required. See, e.g., Exmark Mfg. Co.
    Inc. v. Briggs & Stratton Power Prods. Grp., LLC, 
    879 F.3d 1332
    , 1348 (Fed. Cir. 2018). “[T]he ultimate combination
    of royalty base and royalty rate must reflect the value at-
    tributable to the infringing features of the product, and no
    more.” Finjan, Inc. v. Blue Coat Sys., Inc., 
    879 F.3d 1299
    ,
    1309 (Fed. Cir. 2018) (alteration in original) (quoting Er-
    icsson, Inc. v. D–Link Sys., Inc., 
    773 F.3d 1201
    , 1226
    (Fed. Cir. 2014)). This is so even where the proposed roy-
    alty base is the smallest saleable patent practicing unit or
    SSPPU. VirnetX, Inc. v. Cisco Sys., Inc., 
    767 F.3d 1308
    ,
    1329 (Fed. Cir. 2014).
    4   While we disagree with the district court to the ex-
    tent it suggested that the only way to apportion is through
    the royalty base, Mr. Milani also did not apportion using
    the royalty rate.
    Case: 20-1413    Document: 98     Page: 25    Filed: 08/26/2021
    MLC INTELLECTUAL PROPERTY LLC    v.                       25
    MICRON TECHNOLOGY, INC.
    Here, the sole asserted claim is claim 30, which is di-
    rected to an “[a]pparatus for programming an electrically
    alterable non-volatile memory cell,” and the accused tech-
    nology does not make up the whole of the accused die or
    wafer. Neither of Mr. Milani’s damages theories—compa-
    rable license or SSPPU—apportioned for the non-patented
    aspects of the accused dies or wafers. Accordingly, the dis-
    trict court did not abuse its discretion in granting Micron’s
    Daubert motion.
    We turn first to Mr. Milani’s and MLC’s contention
    that because the licenses are “comparable,” there is de facto
    no need to apportion. See J.A. 893 n.195 (“In other words,
    the royalty rate associated with the comparable license
    agreements already apportions for other components and
    technologies included in the infringing product.”). We have
    previously approved the use of comparable licenses to ac-
    count for apportionment. See Bio-Rad Labs., Inc. v. 10X
    Genomics Inc., 
    967 F.3d 1353
     (Fed. Cir. 2020); Elbit Sys.
    Land & C4I Ltd. v. Hughes Network Sys., LLC, 
    927 F.3d 1292
     (Fed. Cir. 2019); Commonwealth Sci. & Indus. Rsch.
    Org. v. Cisco Sys., Inc. (CSIRO), 
    809 F.3d 1295
     (Fed. Cir.
    2015). As we have explained, “[w]here the licenses em-
    ployed are sufficiently comparable, this method is typically
    reliable because the parties are constrained by the mar-
    ket’s actual valuation of the patent.” CSIRO, 809 F.3d at
    1303     (footnote    omitted)    (citing   Georgia-Pacific,
    
    318 F. Supp. at 1120
    ). We reject the view that the Hynix
    and Toshiba agreements are comparable licenses. As the
    district court properly explained, “there is no evidence re-
    garding the Hynix agreement that supports [Mr.] Milani’s
    opinion that a specific royalty rate derived from the Hynix
    agreement already accounts for apportionment of non-pa-
    tented features” in this case. MLC III, 
    2019 WL 3070567
    ,
    at *3.
    We agree with the district court that the cases in which
    we have held that damages can be based on the terms of a
    comparable license that already values the patented
    Case: 20-1413    Document: 98     Page: 26    Filed: 08/26/2021
    26                        MLC INTELLECTIAL PROPERTY LLC V.
    MICRON TECHNOLOGY, INC.
    technology involve very different facts than those pre-
    sented here. For example, in CSIRO, we held that the dis-
    trict court properly accounted for apportionment when it
    relied on proposed royalty rates from prior negotiations be-
    tween the parties to the litigation involving the same pa-
    tent and accused technology. 
    809 F.3d 1302
    –04. We
    explained that “the district court did not violate apportion-
    ment principles in employing a damages model that took
    account of the parties’ informal negotiations with respect
    to the end product.” 
    Id. at 1304
    . We expanded on this anal-
    ysis in Elbit. There, we rejected Hughes’s argument that
    Elbit’s damages evidence, and hence the jury award, was
    counter to our precedent on apportionment. Elbit, 927 F.3d
    at 1301. Elbit’s damages expert relied on a prior settle-
    ment agreement for slightly different technology to support
    his reasonable-royalty determination. Id. at 1300. Like
    MLC’s expert, Elbit’s expert opined that apportionment “is
    essentially embedded in [the] comparable value” from the
    prior agreement. Id. at 1301 (alteration in original). Un-
    like MLC’s expert, however, Elbit’s expert appropriately
    accounted for the differences between the technology at is-
    sue in the settlement agreement and the accused technol-
    ogy.    Id. (explaining that Elbit’s expert’s testimony
    “allowed the jury to find that the components at issue . . .
    were comparable to the components at issue” in the prior
    agreement). Likewise, in Bio-Rad, though Bio-Rad’s ex-
    pert did not adjust the royalty rate of the comparable li-
    censes, he did account for the differences between the
    accused technology and the licensed technology. 967 F.3d
    at 1377 (explaining that Bio-Rad’s expert “assess[ed]
    whether the importance of [the] technology to the particu-
    lar license was similar to the hypothetical negotiation,” re-
    lying in part “on the reports, testimony, and conclusions of
    other witnesses”).
    In this case, Mr. Milani provided no evidence or expla-
    nation for how the 0.25% royalty rate he derived from the
    Hynix agreement accounts for apportionment of Micron’s
    Case: 20-1413    Document: 98     Page: 27    Filed: 08/26/2021
    MLC INTELLECTUAL PROPERTY LLC   v.                        27
    MICRON TECHNOLOGY, INC.
    accused products. Specifically, unlike the expert in Elbit,
    Mr. Milani conducted no assessment of the licensed tech-
    nology versus the accused technology to account for any dif-
    ferences.     See 927 F.3d at 1300 (“[P]rior licenses or
    settlements need to be ‘sufficiently comparable’ for eviden-
    tiary purposes and any differences in circumstances must
    be soundly accounted for.” (quoting VirnetX, 767 F.3d
    at 1330)). Mr. Milani’s general characterization of the
    flash memory market as a whole as a commodity market
    does not satisfy this requirement of establishing that a li-
    cense is, in fact, comparable. Moreover, unlike the agree-
    ment in CSIRO, the Hynix agreement is not a license for
    the same single patent. See 809 F.3d at 1298. To the con-
    trary, the Hynix agreement granted a license to a portfolio
    of forty-one U.S. and international patents and patent ap-
    plications, and only one of those forty-one patents is at is-
    sue in the hypothetical negotiation. For these reasons, we
    see no abuse of discretion in the district court’s determina-
    tion that Mr. Milani’s comparable license theory does not
    properly apportion for the value of the patented technology.
    We are also not persuaded by MLC’s argument that it
    need not further apportion beyond the single-component
    SSPPU because the asserted claims are directed to a
    memory device as a whole. Contrary to MLC’s suggestion
    in its briefing, claim 30 is the sole claim at issue in this
    appeal. Because claim 30 is an “[a]pparatus for program-
    ming an electrically alterable non-volatile memory cell
    having more than two predetermined memory states,”
    ’571 patent col. 15 ll. 10–12, it is not commensurate in
    scope with the SSPPU, which also contains “error correc-
    tion hardware,” “data clocking hardware,” “addressing
    hardware,” “cache registers,” and “digital to analog con-
    verters.” J.A. 1242. Accordingly, we affirm the district
    court’s Daubert order excluding MLC’s expert testimony re-
    garding a reasonable royalty for failure to apportion.
    Case: 20-1413   Document: 98    Page: 28   Filed: 08/26/2021
    28                       MLC INTELLECTIAL PROPERTY LLC V.
    MICRON TECHNOLOGY, INC.
    CONCLUSION
    We have considered MLC’s remaining arguments and
    find them unpersuasive. For the foregoing reasons, we af-
    firm the district court’s orders excluding the reasonable
    royalty opinion of MLC’s damages expert.
    AFFIRMED