Broadcom Corporation v. ITC
Broadcom Corporation v. ITC
Brian Matsui argues in front of the U.S. Court of Appeals for the Federal Circuit. Brian's argument starts at 22:30.
Unofficial transcript for users of mofo.com
Speaker 1 (00:01): The United States Court of Appeals for the Federal Circuit is now open and in session. God save the United States and this honorable court.
Judge Lourie (00:12): Good morning, ladies and gentlemen. We have six cases on the calendar this morning. One from the ITC, three from the TTab, two from the claims court, which are being submitted on the briefs and therefore not argued. Before we begin, when we ask counsel to be alert to the possibility, in fact likelihood of the judges asking questions. When we’re in court, you can see us leaning forward, ready to pounce. But in the courtroom—but on a telephonic conference, you cannot, and we often have lawyers continuing to ask when judges are asking questions. So please be alert to try to avoid that. Now first question, our first case is 20-2008, Broadcom Corporation v. the ITC and Renesas Electronics. Please proceed, Mr. Johnson.
Brian Johnson (01:22): Thank you. Good morning, and may it please the court. This is Brian Johnson representing Broadcom. Your Honors, the Commission was correct in finding that numerous respondents infringed Broadcom’s ’583 patent, and that all the asserted claims are valid, but wrong for denying Broadcom any remedy for failure to meet the domestic industry requirement. The claims at issue are directed to a system and for both domestic industry and infringement, Broadcom showed the claims or practice looking at the same type of product, a system-on-a-chip or SoC. The Commission correctly found infringement, but for domestic industry, the Commission defied precedent by requiring more evidence of unclaimed elements, like a memory. The Commission should be reversed because the products at issue, Broadcom’s new low-power SoC is an article protected by the patent, and the Commission cannot require Broadcom to prove more than what is in the claim. Broadcom made and sold its new low-power SoCs, every piece of it—hardware and software—and the SoCs practice every element of the claim. The Commission agrees for all but one limitation: 25C, a register controlled by a clock tree driver. Focusing on the claim, all of that has already been conceded. Broadcom develops the register, the clock tree driver, and the register is controlled by a clock tree driver. That’s how it was designed, and that’s all 25C requires. Ultimately, on this issue, the Commission should be reversed for two reasons. First, the Commission failed to ground its analysis of the technical prong in the claim as precedent requires. And second, the Commission failed to use infringement principles to analyze Broadcom’s industry in its use, development, and sale of it’s new low-power SoCs. On the finding of infringement, the Commission should be affirmed for two reasons. First the interview. Yes, Your Honor.
Judge Stoll (03:09): [Inaudible] I—sorry to interrupt your summary there, but I understand that you’re moving on to the other issue, and I just wanted to make sure I asked you some questions on that domestic industry question. I think that your point is that the language, at least one register that is controlled by a clock tree driver, doesn’t say anything about a memory nor does anything else in the claim. But in your system, I think it’s undisputed that the clock tree driver is located in the memory. So I have a question for you. It seems to me that the ITC interpreted this claim to require the language, at least one register that is controlled by a clock tree driver to actually require the clock tree driver, at least have the register be connected to one actively. Did you seek an interpretation of the language, at least one register that is controlled by a clock tree driver before the ITC? Cause I don't see where that term has been interpreted. And if you did seek a particular interpretation of that language, where would I find that in the Appendix?
Brian Johnson (04:14): Your Honor, we did not seek interpretation at the claim construction stage. The argument came up by the respondents that we did not demonstrate a complete system. And our response in our post hearing brief was that the complete system is not required because a memory is not a claim requirement. So that was in our post hearing brief. And it-
Judge Stoll (04:37): What [inaudible] that claim doesn’t require a clock tree driver? I mean, I agree with you it doesn’t require memory, but where is your argument that the language, at least one registered, that it’s controlled by a clock tree driver requires a register, but not a clock structure for a clock tree driver? Where is that argument? If you made it.
Brian Johnson (04:59): Your Honor, in the briefing below, we demonstrated¾our argument was that we were required to show a register and a register that was controlled by the clock tree driver. And we made that argument before the ALJ, and the ALJ in fact found that in our favor. And this is in Appendix 101, where the ALJ acknowledges that the registers referenced by Broadcom source code are controlled by a clock tree driver implemented in source code filed to code .c, and she cited our expert for support. She also found in our favor that the clock tree driver was located on the SoC hardware. Her particular finding of—that we didn’t practice this claim element referred to an external memory, which as Your Honor correctly demonstrates we say is not required by the claims. But as far as the physical presence on the SoC hardware, the ALJ did find that in our favor. And she did that-she did that on APPX 104, and she describes the inter-loop arc processor. That’s a processor within the SoC, and she says in her own words, it’s in the domestic industry products, and it writes a value to the registers in question. She cites to our experts-
Judge Stoll (06:13): [Inaudible] that page, Appendix 104, and could you point me to the language and about where on the page I should be looking? That would be helpful.
Brian Johnson (06:24): Yes, Your Honor. Let me open it up so I can find the page location.
Brian Johnson (06:40): So it’s at the very bottom of the page, Your Honor. And it’s with respect to claim element 25D, but we use precisely the same proof and evidence to disclose that there was a claim element 25D requires an—a processor that overwrites. And it is in fact that inter-loop processor overwriting that is a right to this register. Its in effect claim element 25D is similar to 25C. We use the same proof and it’s a bit of a paradox that the ITC found that we satisfied D without C. So it’s the very bottom there where she explains that our expert identified that our inter-loop arc processor in the domestic industry products writes a new value to the registers in question. What’s important is she cites to our expert at CX0006C at Q&A201. And that translates to Appendix 1763. That’s where our expert explains that the inter-loop processor actually executes code to code .c. That’s the clock tree driver firmware, as the ALJ found elsewhere.
Judge Stoll (07:49): The only concern I have with this page you’ve cited for me is the sentence that proceeds the one you’ve identified. It’s in this section above. It does relate to the limitation at issue and says Broadcom has failed to identify any actual articles practicing the clock tree driver limitation of claim 25.
Brian Johnson (08:08): Yes, Your Honor. And that relies on the discussion just before that, where the ALJ believed that we required demonstration of an external memory, which we do not, both within the claims and as a functional aspect of the product itself, because¾
Judge Stoll (08:28): [Inaudible] the memory that contains the clock tree driver software, right?
Brian Johnson (08:33): The [inaudible]—sorry, it may have cut out.
Judge Stoll (08:37): [Inaudible] line four on that page says Broadcom has not identified any particular set top box or any specific memory that contains the clock tree driver software. I’m just trying to understand. So I just want to make sure that we’re talking about the same thing that, you know, it's not that the ALJ just said a memory. It said a memory that contains the clock tree driver software, because that’s where the clock tree driver software is located in your domestic industry product, right?
Brian Johnson (09:06): No, in two responses to that it—the—it is located on the processor itself. So that is a physical place where it is located. And that’s the only portion that’s actually claimed. That’s again, 25D—I’m expanding the universe a little bit. And that is a physical location that is executed. When it’s executed, it physically resides there. So a memory isn’t required at all, but if a memory was required, then as we’ve explained in our briefing based on testimony from our fact witness, that it is located on an internal memory within the SoC as well. Our customers use it with an external memory, and the testimony that both the Commission and the Intervenors cite about external memory all have to do with our customer use. But our customer use is irrelevant. And it is the Broadcom new, low-power SoC is our domestic industry product, and we have shown both memory and processor physically located on the SoC hardware that contains that firmware. So the ALJ’s ruling that we haven't shown a specific memory, and she says, “external memory elsewhere is fixated on this idea that we need an external memory.” So this is either a misapplication of claim construction or a misapplication of this court's precedent in Microsoft v. the International Trade Commission. Because we believe that is the case that properly aligns to this court’s precedent. And we also believe that the ALJ was applying an argument that was primarily directed to a different patent, the ’027 patent. It does have an express memory requirement in it, and that’s important¾
Judge Stoll (10:49): One last question. On page A-12, I take it then you completely disagree with the statement by the ALJ that Broadcom does not dispute that the claim clock tree driver is stored on an external memory separate from BCM7260.
Brian Johnson (11:07): We would not dispute that some of our customers use it that way, but we don’t dispute that that’s how the product must operate because we do it with an internal memory, which is demonstrated in our testing documents that the HEVD low-power specification that we cite repeatedly. Appendix 1900 and 1938 are two of them that show specific tests with respect to the clock tree driver firmware actually running on the SoC hardware itself. And that’s not customer use, so that doesn’t require an external memory. So it depends how Your Honor interprets that. But the fact that if an external memory is required, we absolutely dispute that. But I don’t know that that was the ALJ’s finding. She seemed to believe that we needed to rely on our customer’s product, and that’s the Commission’s error.
Brian Johnson (12:03): In fact, Your Honor, I would like to focus on this [inaudible] a little bit before I move on to the next. I don’t have a lot of time here, but I’ll focus on Alec briefly. And this court stated that it’s essentially the same test as infringement. You look at the claims and you apply it to the product, and that’s precisely the right method, and the method applied in the Microsoft case, which is the only support that the Commission and Intervenors look at. In that case, the claims at issue were directed to a mobile device, and this court followed the procedure in Alec correctly, asking the question, “did Microsoft develop a mobile device to practice a patent?” And the answer was no because Microsoft made software alone, and it was required to look at third-party products. The Nokia mobile device was what Microsoft identified as a product that practiced the patent, and that’s why the focus became on the customer use. And so it’s an improper application of Microsoft that the Commission appears focused on the same. There’s an important distinction here. Broadcom makes everything, the entire claim system, in house, and we don’t have to rely on any component by any other company to achieve that. So the Microsoft case doesn’t apply here. Moving on very briefly to claim 17 infringement. We believe that the Commission should be affirmed there because the Commission effectively—or pardon me, the Intervenors’ challenge asks for a redo on claim construction, asking that this court undo an agreement that they made years ago. And even if permitted, their substantial evidence challenge would fail because they ignore most of the evidence that the Commission relies upon in that case. Your Honor, I see I’m over time, so I'll reserve the rest for rebuttal.
Judge Lourie (13:48): We will save you three minutes, Mr. Johnson. Ms. Herzbach.
Lynde Herzbach (13:57): Yes, may it please the court. I would like to start with the ’583 patent and the technical prong findings. And specifically, I’d like to spend some time on the clock tree driver limitation and whether it’s required. So there was no dispute before the Administrative Law Judge or the Commission that the clock tree driver was required as part of claim 25. And this is shown even during the claim construction of a different—at least one processor term, the ALJ found that during prosecution of the ’583 patent, the applicant faulted the examiner for failing to identify a clock tree driver disclosed in the prior art. And that’s at APPX 87, and Broadcom never sought review of that finding. Broadcom tries to compare the technical prong arguments to its infringement arguments. And I think that is important to look at. Broadcom argues for infringement based on accused products that include an SoC, the head unit that it’s installed in, all the way to the automobile that the head unit’s installed in. So it’s just a much larger universe than just the SoC that they’ve identified for their domestic industry product. And Broadcom actually argued for infringement that this claim 25 requires hardware and software. The required software is the clock tree driver, and that’s at APPX 2428. So there’s no dispute here that the clock tree driver is required. What the Administrative Law Judge found is that on the system identified by Broadcom, which was only its system-on-a-chip, that did not include the clock tree driver; therefore, it didn’t meet the claim limitation. So essentially Broadcom’s now arguing on appeal for a case that it didn’t make below. The Commission opinion specifically found that Broadcom waived any reliance on its firmware together with the identified SoC as the domestic industry product. And the file that was identified as part of that firmware-
Judge Stoll (16:01): [Inaudible] Counsel. This is Judge Stoll. I don’t see Broadcom carrying that argument forward. I see them relying on the SoC alone on appeal before us. Do you agree with that?
Lynde Herzbach (16:11): I don't agree with it just because of the way that they’ve relied on the additional—the firmware to kind of backfill where this clock tree driver is missing. So they’ve stated repeatedly that there was no dispute that it was actually implemented, and we disagree. We think that that’s what the ALJ found, you know, as Your Honor pointed to that finding in the initial determination at 104, that Broadcom failed to identify the actual articles practicing the limitation, because there was not implementation of the clock tree driver on the SoC.
Judge Stoll (16:45): Okay, thank you.
Lynde Herzbach (16:47): Yes. So you know it—to establish the violation of section 337, Broadcom had to demonstrate there was a domestic industry related to articles protected by the ’583 patent. So here in their post-hearing brief, they include a specific section entitled Broadcom Domestic Industry Products, and that’s at APPX 528. And in that section, they specifically define that SoC part numbers. They don’t mention the firmware. So this case actually is like Microsoft. And while counsel for Broadcom has mentioned that the Microsoft patent included claims that required applications on the phone. That’s only true for this ’376 patent. And there are actually two patents for which Microsoft failed to satisfy the technical prong. Even regardless of the claim limitations in Microsoft, what matters is what’s defined as the domestic industry product. So here, there was the system-on-a-chip that had to meet the claim limitations for the claim 25 system. It simply doesn’t meet those claim limitations because it doesn’t include the clock tree driver. Just as in Microsoft, there’s insufficient proof here that the clock tree driver’s actually implemented on that domestic industry SoC. Broadcom essentially is asking this court to re-weigh a highly technical evidentiary record regarding the ’583 patent, but substantial evidence supports the Commission’s determination that the technical prong was not satisfied, and it should be affirmed. There was no specific finding—
Judge Lourie (18:33): Was the Commission’s determination the same as in the initial decision?
Lynde Herzbach (18:39): The Commission affirmed the findings that the technical prong was not met. It provided an additional analysis. And so that’s what I had touched on briefly that the Commission actually found a waiver of this argument that would incorporate the SoC and the firmware together. And that argument was waived and that’s not an abusive discretion on the Commission’s part. The deficiency in Broadcom’s identification of its domestic industry product was identified very early in the investigation. Even in respondent’s pre-hearing brief, they argued that Broadcom identified the SoC alone domestic industry product. And that argument was made at APPX 51054. So I know now they’re trying to say that the SoC includes the clock tree driver and that it is implemented on the processor, but that’s contrary to its own engineer’s testimony. So the engineer—Broadcom’s engineer testified before the ALJ that the clock tree driver’s not found on the domestic industry SoC. And as Your Honor pointed out, Broadcom did not dispute that point before the Administrative Law Judge.
Judge Stoll (19:53): Do you have a citation to the Appendix for that testimony?
Lynde Herzbach (19:58): I do. It’s explained in our brief at page 39, but the citation to it’s a little bit long in the hearing. It’s—give me one minute. It’s actually in the hearing testimony, so it’s like 10110 to 111, but it’s explained in our brief at page 39.
Lynde Herzbach (20:33): So based on that testimony that the clock tree driver was not present on the domestic industry product, the Administrative Law Judge found that the claim limitation wasn’t met. Now, Broadcom points to the other claim limitation regarding a processor. But as I’ve mentioned, they don’t dispute that this claim requires both hardware and software. So what’s missing here is the clock tree driver software that is a required part of the system, and it’s not present on their domestic industry product. I would like to touch on briefly the Administrative Procedure Act argument that they have for the ’752 patent, if the court has no further questions regarding the technical prong?
Judge Lourie (21:18): No, proceed.
Lynde Herzbach (21:19): Thank you, Your Honor. So regarding the ’752 patent, Broadcom made the argument that the Commission’s decision is not in conformance with the Administrative Procedure Act, and that’s contrary to the actual record below. So unlike the cases cited by Broadcom, there’s no new combination here. Claim 5 is filed invalid based on the combination of Sih and Foster. And there was additional process before the Commission that was not present in the patent trial and appeal board cases on which Broadcom relied. So Broadcom raised its ’752 patent arguments regarding the paragraph 44 citation and the reasonable expectation of success before the Commission’s petition for review. Broadcom’s petition did not seek a remand to the Administrative Law Judge for further evidence or ask to reopen the record. And the Commission granted the petition and reviewed those findings.
Judge Lourie (22:18): Right. Have you finished your thought, counsel?
Lynde Herzbach (22:23): I did, Your Honor. Thank you.
Speaker 2 (22:26): Okay. Thank you very much. We’ll hear from Mr. Matsui now.
Brian Matsui (22:32): Thank you, Your Honor. May it please the court. Brian Matsui for the Intervenors. Substantial evidence-
Speaker 2 (22:37): I might say this is a battle of the Brians.
Brian Matsui (22:40): It is, Your Honor. Substantial evidence supports the Commission’s no violation determination. On issue after issue, the ALJ and Commission squarely addressed disputed issues and resolved them on the facts against Broadcom. Broadcom’s repeated contentions that the issues on appeal were undisputed simply cannot be reconciled with any reading of the record. And this court should reject as waived, just as the Commission did, Broadcom’s 11th-hour attempts to recast its case. And I think it’s important to emphasize the Commission’s waiver finding. This is a waiver finding that Broadcom waived by failing to challenge in its opening brief. And there should be great deference to that agency determination. At Appendix 19, the Commission specifically found, and it said, “the Commission specifically finds that Broadcom’s DI products with respect to the ’583 patent is only the SoC and does not include customer set top boxes or a larger system.” And then it goes on to say, “to the extent Broadcom argued in its petition that the DI products are a system or something other than the SoCs, Broadcom waived that argument because it was not presented to the ALJ.” Now, before this court, Broadcom is definitely arguing that it’s something more than the SoCs. And we just looked at page five of the reply brief that they filed. They say in the middle of that page, “as detailed below, Broadcom calls this system it’s new low-power SoC, which includes this register”—I don’t want to say it because it’s highlighted—“and new low-power firmware (clock tree driver).” So it’s expanded its argument to include a clock tree driver, which is not on the SoC. And it conceded that before the ALJ. As the Commission’s counsel just said at Appendix 10111, Broadcom’s own witness, Mr. Hellman, an engineer, conceded facts that support the board—the Commission’s finding that there’s no software installed on the chip to no clock tree driver. He said “it’s stored in external memory.” So right there, the Commission was more than reasonable to rely upon that concession to say that there’s no clock tree driver in the SoCs, and that’s substantial evidence. The mere fact that Broadcom might be able to point to conflicting evidence under the substantial evidence standard doesn’t mean that there should be a different conclusion. A couple other points I just would like to make—again, just to reiterate, Broadcom has always argued that there is a clock tree driver that is required. At Appendix 2417 in their brief to the Commission, Broadcom argued claim element 25C can be broken down into three components, at least (1) register, and (2) a clock tree driver where (3) the clock tree driver controls at least one register. So right there, they are admitting to the Commission that this clock tree driver is required. It needs to be in their article, and it was not there because their engineer conceded that it’s not. And this is not a claim construction dispute there. I disagree with Broadcom’s counsel saying that there was a requirement of memory. The issue isn’t that memory is required. It’s that a clock tree driver is required by the claims. And Broadcom’s witnesses said that that clock tree driver is going to be stored on memory. And so the fact that there’s not a clock tree driver, because there’s not memory, just underscores the failure of proof in this case.
Judge Lourie (26:32): You did note that it was in the external memory. In other words, it’s involved; it’s not absent.
Brian Matsui (26:43): He knows it’s in an external memory that is not the SoC. And so the problem they have is that they identify their domestic industry article as the SoC alone. And they did not include in it anything—any firmware that would be sort of on external memory. And so when you just identify the chip itself, that means that their article does not include a clock tree driver. And even if we were to sort of, you know, separately look at this is their waived argument—even if we were to sort of separately look at the more complete system, which they’re arguing now, at page five in the reply brief, the Commission addressed that and said that that evidence of testing, of development was just vague and unsubstantiated, and that’s at Appendix 23, where if you look at the Commission just said, Hellman’s so-called testimony, Hellman’s testimony was vague. There’s no evidence that specifically identifies-
Judge Stoll (27:49): Counsel, you say Hellman’s testimony was vague. Could you be more specific?
Brian Matsui (27:53): You’re right. I apologize. It says no evidence. I’m sorry.
Judge Stoll (27:56): [Inaudible] I think what you mean by vague is that it doesn’t even identify—it’s not related specifically, for example, to this register or this clock tree driver is directed more generally to how all sorts of different chips, not the SoC. Am I right about that?
Brian Matsui (28:18): It is. I mean, when I looked at all their citations in their reply brief and in their opening brief, it’s basically testimony that just talks about research and development, the fact that they work with customers to develop software and firmware. But none of that relates specifically to any sort of clock tree driver that would be used with their SoCs. And under the substantial evidence standard of review, the Commission could easily discount that type of evidence as not being specific enough to show that there is a clock tree driver in this case. I know that the counsel for Broadcom didn’t discuss the ’752 patent, but substantial evidence supports the Commission’s--the ALJ’s and Commission’s determination that those claims are unpatentable. I’m happy to ask any questions that the court might have.
Judge Lourie (29:22): That also was involved in the 1511 case, isn’t it?
Brian Matsui (29:28): That’s correct, Your Honor.
Judge Lourie (29:30): All right. Thank you, Mr. Johnson. Anything further?
Brian Matsui (29:36): No, Your Honor. We would ask that the Commission would be affirmed.
Speaker 2 (29:39): Mr. Matsui, sorry. Mr. Johnson has three minutes for rebuttal.
Brian Johnson (29:45): Yes. Thank you, Your Honor. I’d like to address a few points raised by the Commission and Intervenors. First, the Commission said that we are asking this court to re-weigh the evidence, and that is a complete mischaracterization of our position here. Our position is that this was a misapplication of the claim construction analysis. There is a requirement of a memory held by both the ALJ and the Commission, and we do not require that. And with respect to the actual facts, the ALJ’s findings alone are sufficient to demonstrate that absent this wrong requirement of the memory, Broadcom’s new, low-power SoCs practice every element to the claim. Now, the Commission made the argument that we waived use of firmware in establishing the technical prong. The Intervenors did not appear to join that in briefing, but may have today. Regardless, it is inaccurate. We presented both the hardware and the firmware. That’s how we defined our new low-power SoCs. That’s how it’s defined in all of our HEVD low-power specifications at Appendix 1900. But the most important citation perhaps is the ALJ’s findings, where she correctly recognized that we relied on firmware all along. At Appendix 101, she identifies correctly that our clock tree driver is source code filed to code .c, that’s software, and then at Appendix 104, she recognizes that the inter-loop processor runs that software. And she cites to our expert at CX6 Q&A201 describing the operation of our decode .c. The ALJ, in her own words, talks about what is a software overwrite and our description of firmware. So the idea that we did not present firmware below is not consistent with the record.
And finally, I’d like to address Mr. Hellman’s testimony. And I’d like to expand the citation a little bit because in full context, his testimony supports what Broadcom’s argument has been along. It’s in 10103.3-10110.2. And in that question and answer segment, Mr. Hellman testified that within the context of our customer usage, that the customers did not ran firmware and there wasn’t sufficient memory on the SoC itself to run all of the firmware that may be executed by customers. Then the questioner directed it specifically to the ’583 patent and asked about that. His response then was that it is moved on to internal memory and executed in the inter-loop arc processor. So to the degree a memory is required, and we fervently disagree with that proposition. But if it is, the internal memory is right there. That was his testimony in the context of the ’583 patent. And it’s executed on the inter-loop arc processor. And that-s consistent with what the ALJ found in Appendix 104. With respect to the--
Judge Stoll (32:39): This is Judge Stoll. Very quickly, I’m sorry, could you repeat those cites that you just gave? I did not get that.
Brian Johnson (32:44): Yes. It’s the Hellman cites were 10103.3 through 10110.2. The Appendix findings to the ID regarding the--
Judge Stoll (32:58): I just wanted the Hellman. I'm good. Thank you. Go—please proceed.
Brian Johnson (33:04): And finally, the—very quickly—the Commission’s findings with respect to the testimony of Mr. Hellman and whether it was specific enough, importantly, there was other testimony from Mr. Hellman in questions 71 to 76 specific to the ’583 patent that the Commission did not cite. And that is the evidence that shows our new low-power technology. It shows both hardware and software, and it happens to be the very specification our expert relied on to show satisfaction of all the claim elements, and very critically, it shows the testing of revision Q2 and revision R that we conducted on the internal memory and processor. So that’s a specific article protected by the patent. Thank you, Your Honors.
Judge Lourie (33:44): Thank you, counsel. As, as you observed, we have a clock driver here, too. Thank you to both counsel, all counsel, and the case is submitted.