US 6854287

I'll search for authoritative information on US 6854287 and check for any 2026 CAFC docket activity.

The CAFC search returned nothing. Let me try a couple more targeted searches to confirm whether any 2026 Federal Circuit docket activity exists.

US Patent 6,854,287 — Analyst Summary

Source of record: Google Patents full text (https://patents.google.com/patent/US6854287/en), FreePatentsOnline (https://FreePatentsOnline.com/6854287.html), plus litigation/PTAB records below. Search was restricted to the identifier 6,854,287; results for related family members (e.g., 6,868,682; 6,718,277; 6,854,284) are flagged as such, not substituted.

1. Bibliographic data

The "Cross-References" section confirms this is a divisional of Ser. No. 10/210,040 (filed Aug. 2, 2002), so the effective priority date runs from 2002-08-02 even though the present application was filed 2003-10-31.

2. Abstract (verbatim)

"A system and method for cooling a room configured to house a plurality of computer systems. A heat exchanger unit is configured to receive air from the room and to deliver air to the room. The heat exchanger unit is supplied with cooling fluid operable to cool the received air in the heat exchanger unit. At least one of the temperature of the cooling fluid supplied to the heat exchanger unit and the air delivery to the room may be controlled in response to temperatures sensed at one or more locations in the room."

3. Plain-language overview of the independent claims

The patent has 22 claims. The record shows four independent claims — 1, 10, 14, and 16 (claim 16 is truncated in the fetched text, so its full wording is uncertain; see §5).

• Claim 1 — Method of cooling a room (independent).
  A cooling method where: (a) a plurality of heat exchanger units both draw air from the room and return air to it; (b) the units are supplied with cooling fluid from an air conditioning unit; (c) the received air is cooled by heat exchange with that fluid; (d) temperatures are sensed at one or more room locations; (e) in response to those readings, the system controls either the cooling-fluid temperature or the air delivery by the units; and (f) — the key narrowing element — the air-delivery control is accomplished by "individually manipulating a mass flow rate of the cooling fluid supplied to each of the plurality of heat exchanger units." In plain terms: per-unit, demand-based coolant flow control in a multi-unit cooling system. Dependent claims 7 and 8 specify that the mass-flow metering is done with valves or pumps on the individual fluid lines; claim 9 adds that flow to the units is manipulated in substantially independent manners. Claims 3–6 add a predetermined temperature band with increases/decreases of cooling-fluid temperature based on whether sensed temperatures fall below/above the band.

• Claim 10 — System (independent).
  A room-cooling system for racks containing computer systems, comprising: a plurality of heat exchanger units receiving cooling fluid via a line from an air conditioning unit and receiving air through openings; each unit has at least one fan moving air in/out; a heat-exchanger controller governing coolant supply and fan speed; one or more temperature sensors; an air-conditioning (cooling device) unit controller that varies coolant temperature delivered to the room locations; and a plurality of pumps controlling coolant delivery to respective heat exchanger units — with the heat-exchanger controller able to individually control the mass flow rate of coolant into each respective unit via those pumps. In plain terms: the apparatus counterpart to claim 1, using per-unit pumps.

• Claim 14 — System (independent).
  Same overall system architecture as claim 10, but instead of pumps it recites "a plurality of valves configured to meter the flow of cooling fluid through each of said plurality of heat exchanger units positioned along respective cooling fluid lines generally upstream of respective heat exchanger units," with the controller individually controlling mass flow through the valves. In plain terms: the valve-based variant of the system claim.

• Claim 16 — System (independent; text truncated).
  The provided document cuts off at "A system for cooling computer systems h…". Based on the patent's Summary of the Invention, the third independent embodiment is a means-plus-function system claim: "means for cooling air including means for receiving cooling fluid," "means for delivering cooled air to the computer systems," "means for measuring temperatures at one or more locations in the room," "means for controlling delivery of the cooled air in response to the temperature measurements," and "means for controlling the temperature of the cooling fluid." I cannot quote claim 16's exact language with confidence because the fetched text is incomplete (see §5).

4. Litigation / docket activity bearing on 6,854,287

USPTO / PTAB:

• IPR2025-00667 — Vertiv Corporation v. Valtrus Innovations Ltd., filed 2025-02-26 against respondent application 10/697,697 / patent 6,854,287. Status per the docket summary: Discretionary Denial; institution decision date 2025-07-31. (https://ipverse.greyb.com/ptab-web/cases/case-details/IPR2025-00667)
• The Google Patents "Family has litigation" panel lists a PTAB case IPR2025-00667 as "Not Instituted – Procedural," consistent with the above discretionary denial. It also lists numerous district-court cases (E.D. Tex., N.D. Ill., D.N.J., W.D. La.) and a first worldwide family litigation entry.

District court (the '287 patent specifically asserted):

• Valtrus Innovations Ltd. et al. v. Lumen Technologies Inc. (W.D. La., Monroe Div., filed ~May 2026) asserts U.S. 6,854,287, 6,868,682, and 6,718,277. (https://www.legalnewsline.com/louisiana-record/suit-la-data-center-company-infringing-on-cooling-patents/article_09c9c16a-be3b-48e6-b7e3-b1cdfca365a3.html)
• Valtrus v. Digital Realty Trust (E.D. Tex., including 2:24-cv-00139) asserted seven HPE-origin patents including US 6,854,287 B2 (app. 10/697,697); member cases were dismissed with prejudice as to Digital Realty/Vertiv by joint motion, lead case 2:24-cv-00361-JRG kept open. (https://www.patsnap.com/fr/resources/blog/litigation/valtrus-v-digital-realty-trust-data-center-cooling-ip-dismissed-patsnap/)
• Valtrus v. Prime Data Centers (N.D. Ill., 1:26-cv-03958) and Valtrus v. NTT Global Data Centers Americas (1:26-cv-03945) assert the '287 patent (claim 1) with claim charts focused on "individually manipulating a mass flow rate" — the term most likely to be disputed at claim construction.
• News coverage (Aug. 2026) reports a Judicial Panel on Multidistrict Litigation (JPML) MDL designation in June 2026 consolidating Valtrus data-center cooling cases. (https://patentlawyermagazine.com/data-center-boom-creates-new-patent-battleground-around-ai-infrastructure/)

CAFC 2026 dockets — what I did and did not find:

• A CAFC search for the specific number 6,854,287 returned no Federal Circuit docket or appeal tied to this patent.
• The one 2026 CAFC item surfaced is Appeal No. 2024-1945, Google LLC v. Valtrus Innovations Ltd., oral argument heard 2026-03-04 (https://www.cafc.uscourts.gov/2026/03/04/page/2/). However, the related Google/Valtrus dispute concerns Valtrus's search-engine / pay-per-use patents (e.g., 6,728,704; 6,816,809; 7,346,604) — not the '287 cooling patent — and the N.D. Cal. action (Valtrus v. Google, 26-cv-2379) was dismissed 2026-08-10. I therefore cannot confirm any 2026 CAFC appeal involving US 6,854,287, and I am stating that as an absence of evidence rather than a verified negative.

5. Uncertainties / caveats

• Claim 16 text is incomplete in the source I was given (truncated at "A system for cooling computer systems h…"). My description of claim 16 is inferred from the Summary of the Invention's means-plus-function embodiment and should be verified against the granted claim set before reliance.
• Claim count: the header says "(22)" claims; I identified independents 1, 10, 14, and (probably) 16, with dependents 2–9, 11–13, 15, and 17–22. The exact dependency of claims 17–22 could not be confirmed from the truncated text.
• Expiration status: the record shows "Expired – Lifetime," anticipated expiration 2022-08-02. Because the patent is a divisional claiming 2002-08-02 priority, maintenance-fee/term calculations would need confirmation from USPTO PatentCenter; the Google Patents legal-status labels are expressly disclaimed as assumptions, not legal conclusions.
• Assignee chain: the "current assignee" listing (Valtrus Innovations Ltd / HPE Development LP) is per Google Patents and may be incomplete as to the KPI (Key Patent Innovations) trust relationship described in the 2026 litigation coverage.
• I did not retrieve a primary USPTO PatentCenter or PACER docket directly; PTAB/litigation facts above come from secondary aggregators (GreyB/IPVerse, PatSnap, CourtListener, news reports). Treat institution/status details as reported, and verify IPR2025-00667's current paper trail on PTAB's own portal.

I'll search for litigation involving US 6854287 specifically.

Let me dig into the specific dockets and the Unified Patents caselist.

Let me pin down the MDL 3190 case list and remaining docket details.

Litigation Report — U.S. Patent No. 6,854,287

Prepared for: litigation-risk / freedom-to-operate review
Patent: U.S. 6,854,287 B2 ("Cooling system"), app. 10/697,697, Patel & Bash, priority 2002-08-02, issued 2005-02-15, expired (anticipated 2022-08-02)
Owner of record for assertion: Valtrus Innovations Limited, with Key Patent Innovations Limited (KPI) as trust beneficiary

---

Threshold note on dates — please read first

There is a direct contradiction between the date supplied in this task ("Current Date: April 26, 2026") and the dates embedded in the current record. The system date of record for this analysis is 2026-09-11, and the docket activity I retrieved runs through September 2026 (e.g., an MDL conditional transfer order filed 2026-08-07 and a case transferred into E.D. Tex. on 2026-08-11 and 2026-09-09). I am flagging this rather than silently reconciling it, per instruction. I have reported events by their actual recorded dates and have not back-dated anything to April 26, 2026. If you are relying on an as-of date of 2026-04-26, the JPML/MDL items below (motion filed 2026-04-28 onward) would not yet exist.

Secondary caveat: I could not load the Unified Patents litigation caselist page for this patent directly. The facts below are drawn from Unified Patents' portal case page (for 1:26-cv-03958), CourtListener/RECAP dockets, PACER-monitor mirrors, filed briefs (P.R. 4-3), complaint analyses, and trade/press reporting. Verify against PACER before relying on any single case.

---

1. District court cases — where the '287 patent is expressly at issue

Note on case 1: the Vertiv declaratory-judgment action is procedurally unusual — Valtrus takes the position there is no case or controversy between itself and Vertiv (Valtrus's motion to dismiss, Dkt. 81), because Valtrus has sued Vertiv's customers, not Vertiv. The parties nevertheless proceeded to joint claim construction. This is the case in which the '287 claim terms have been most substantively briefed to date.

---

2. Additional MDL-3190 associated cases I could not confirm as pleading the '287 patent

The following appear in the MDL 3190 / family litigation record but I did not find the '287 patent confirmed as an asserted patent in them. I am flagging them as unconfirmed rather than omitting them:

• N.D. Ill. 1:26-cv-03926 (defendant: STACK Infrastructure, Inc., among others, per the JPML extension motion)
• D.N.J. 2:26-cv-03884, 2:26-cv-03886
• E.D. Tex. 2:26-cv-00084, 2:26-cv-00286, 2:26-cv-00287
• E.D. Tex. 2:24-cv-00142, 2:24-cv-00259, 2:24-cv-00777, 2:24-cv-01152 (listed in Google Patents' family-level "Family has litigation" panel, which is not patent-specific)

The JPML extension motion (2026-05-15) names defendants STACK Infrastructure, Inc., Cologix, Inc., CoreSite, LLC, NTT Global Data Centers Americas, Inc., H5 Data Centers LLC, Netrality Properties, LP, Prime Data Centers, LLC, Iron Mountain Data Centers LLC, Lumen Technologies Inc., EvoDC, LLC, plus Vertiv Corporation as intervenor/consolidated plaintiff.

---

3. Multidistrict litigation — MDL No. 3190

• Motion filed 2026-04-28 by Valtrus and KPI to centralize 12 actions involving overlapping data center cooling patents, technology and facts (infringement, validity, damages).
• MDL No. 3190 assigned (motion accepted for filing 2026-04-30).
• Conditional transfer orders issued ~2026-08-07, transferring member cases to the Eastern District of Texas; transfers docketed 2026-08-07 / 2026-08-11 / 2026-09-09.
• Associated case list as docketed: ILN/1:26-cv-03926, ILN/1:26-cv-03929, ILN/1:26-cv-03945, ILN/1:26-cv-03958, LAW/3:26-cv-01346, NJ/2:26-cv-03884, NJ/2:26-cv-03886, NJ/2:26-cv-03890, TXE/2:24-cv-00361, TXE/2:26-cv-00084, TXE/2:26-cv-00286, TXE/2:26-cv-00287.

---

4. PTAB / post-grant

Google Patents' panel characterizes this as "Not Instituted – Procedural," consistent with the discretionary denial. Verify the final written decision/termination paper on the PTAB portal — my source for the status is a secondary aggregator (GreyB/IPVerse).

I found no ex parte reexamination or other PTAB proceeding specifically against the '287 patent, though Unified Patents states its Valtrus campaign analysis includes a sample ex parte reexamination work product "for one of the Valtrus patents" (patent not identified publicly).

---

5. Cases that do NOT involve the '287 patent (to prevent conflation)

• Valtrus Innovations Ltd. v. Equinix, Inc., E.D. Tex. 2:25-cv-00016 — asserts the '284, '682, '683 and '967 patents (Exhibits 1–8 to the amended complaint). The '287 patent is not among the asserted patents.
• Valtrus v. Google LLC, N.D. Cal. 3:26-cv-00363 — search-engine / pay-per-use patents (e.g., 6,728,704; 6,816,809; 7,346,604). Squeezed from the earlier N.D. Cal. action dismissed 2026-08-10. Unrelated to '287.
• CAFC: I found no Federal Circuit appeal involving U.S. 6,854,287. The 2026 Google/Valtrus CAFC activity concerns the search patents, not the cooling portfolio.

---

6. Bottom line

Yes — there is substantial litigation involving US 6,854,287. It is one of the core asserted patents in the Valtrus/KPI data center cooling campaign. As of the current record:

• Live and substantive: the E.D. Tex. consolidated proceeding 2:24-cv-00361-JRG (Valtrus v. NTT Data Services et al., with Vertiv's DJ action), where the '287 patent has gone through joint claim construction briefing.
• Live assertions, now consolidated: the 2026 wave against Lumen (3:26-cv-01346), Prime Data Centers (1:26-cv-03958), NTT Global Data Centers Americas (1:26-cv-03945 → 2:26-cv-00679), Iron Mountain (2:26-cv-03890 → 2:26-cv-00669), Netrality (1:26-cv-03929), 365 Data Centers (2:26-cv-00801), and others — all funneling into MDL 3190 in E.D. Tex.
• Terminated: Digital Realty member cases (dismissed with prejudice, 2025-01-23); TierPoint (dismissed without prejudice).
• PTAB: IPR2025-00667 (Vertiv) was denied institution on discretionary grounds, 2025-07-31.

Disputed term to watch: the '287 claims turn on "individually manipulating a mass flow rate of the cooling fluid supplied to each of the plurality of heat exchanger units" (claim 1). Every 2026 complaint I reviewed anchors on that limitation.

---

Verification items I could not close

1. Full defendant-to-case-number mapping for N.D. Ill. 1:26-cv-03926/03928 and D.N.J. 2:26-cv-03884/03886 — confirm on PACER whether '287 is pleaded.
2. Identities of defendants in E.D. Tex. 2:26-cv-00084, 2:26-cv-00286, 2:26-cv-00287 and E.D. Tex. 3:24-cv-01152.
3. Whether the Lumen W.D. La. case has been formally transferred/consolidated into MDL 3190 on the docket (the MDL association is documented; the E.D. Tex. transfer docket number was not retrieved).
4. Final PTAB paper trail for IPR2025-00667 on PTAB's own portal.

I'll start by reconciling the canonical PTAB list against what earlier sections surfaced, then verify with web search.

Good — I have the core proceeding. Let me pull the institution decision reasoning, panel, and check for additional proceedings and any appeal.

Let me look for the denial decision text/panel and check whether Vertiv filed parallel IPRs on sibling patents, plus any appeal.

PTAB Report — U.S. Patent No. 6,854,287 ("Cooling system")

Patent: US 6,854,287 B2 · App. 10/697,697 · Patel & Bash · priority 2002-08-02 · issued 2005-02-15
Owner of record (as asserted): Valtrus Innovations Limited, with Key Patent Innovations Ltd. (KPI) as trust beneficiary
As-of date: 2026-09-11

---

⚠️ Threshold reconciliation — the structured list is wrong (or stale)

The canonical "PTAB proceedings on file" block provided in this prompt states:

"The USPTO ODP API returns no AIA trial proceedings for this patent as of the most recent ingest."

That is contradicted by the record. There is at least one AIA trial proceeding on US 6,854,287 — IPR2025-00667 — which was filed 2025-02-26, fully briefed (preliminary response, discretionary-denial request, petitioner's opposition and corrected response), and terminated by a discretionary denial dated 2025-07-31, with a USPTO refund of post-institution fees mailed 2025-09-23. That paper trail is documented on the PTAB's own PTACTS petition record and is independently reflected in the Google Patents family panel in the source text ("PTAB case IPR2025-00667 filed (Not Instituted – Procedural)").

Per the operating rule to flag contradictions rather than reconcile silently: the ODP block under-counts by one proceeding for this patent. Possible explanations are that the ODP ingest missed a short-lived, never-instituted case, or that the block is keyed to a different identifier (e.g., the parent app. 10/210,040 rather than 10/697,697). Either way, a defendant who relies on the ODP block alone would wrongly conclude the '287 patent has never been challenged at the PTAB. It has.

I have not been able to retrieve the four corners of the 2025-07-31 denial decision itself, and I identify below exactly what is documented versus inferred.

---

Proceedings overview

One AIA proceeding exists against US 6,854,287 itself: IPR2025-00667. Breakdown: 0 active, 0 claims invalidated, 0 claims sustained, 0 settled, 1 institution denied. (Separately, the same petitioner filed IPR2025-00668 on sibling patent 6,718,277 the same day — same outcome — but that is a different patent and is not counted here; see "Pattern signals.")

Bottom-line defensive posture: Neither hardened nor narrowed — untested. Not a single claim of the '287 patent has been canceled, construed, or even adjudicated on the merits at the PTAB. All 22 claims survive exactly as issued. Conversely, because the petition was never instituted, there is no § 315(e)(2) estoppel, so a defendant (and in particular Vertiv itself) is free to raise the identical Newton / Shimizu / Baer prior art — and anything else — in district court. The practical read for a demand-letter recipient: the patent-owner cannot tell you "the PTAB already rejected these arguments and I won," and a defendant cannot tell the court "claims 1–9 are dead." This is a blank slate at the PTAB with a live district-court fight attached.

---

IPR2025-00667 — Vertiv Corporation v. Valtrus Innovations Limited and Key Patent Innovation Ltd.

• Type: Inter Partes Review (35 U.S.C. §§ 311–319; pre-AIA §§ 102/103 apply — the '287 patent is a pre-AIA filing, as the petition itself acknowledges)

• Filed: 2025-02-26 (post-institution fee paid same date)

• Status: "Discretionary Denial" (per IPVerse/GreyB structured record); Google Patents family panel: "Not Instituted – Procedural." Plain English: the Director exercised discretion not to institute; the merits were never reached, no trial was ever declared, and no Final Written Decision issued.

• Challenged claims: 1–9

• Judge panel: Not publicly identified in any source I could retrieve. This is expected: the denial issued on 2025-07-31, i.e., during the period governed by the Acting Director's Interim Processes for PTAB Workload Management (2025-03-26), under which discretionary denial was a phase-one, Director-led determination made before any merits panel was assigned. No three-APJ panel ever took up the merits. I will not guess at APJ names.

• Petition grounds (pre-AIA, claims 1–9):

  Ground	Claims	Basis	Art
  1	1–4, 7–9	§ 102(b) anticipation	Newton (Ex. 1007; issued 1968-05-21)
  2	5–6	§ 103(a) obviousness	Newton + knowledge of a POSITA
  3	1–4, 7, 9	§ 102(b) anticipation	Shimizu (Ex. 1008; issued 1994-06-07)
  4	1, 3, 7, 9	§ 102(b) anticipation	Baer — U.S. Pub. 2001/0042616 A1 (Ex. 1011; published 2001-11-22)

  Note a drafting inconsistency in the petition itself: the grounds table labels Baer as § 102(b), while the petition's own text argues Baer "was published on November 22, 2001, which was before the August 2, 2002 priority date… Thus, Baer qualifies as pre-AIA 35 U.S.C. §102(a) prior art." Baer's publication date is inside the one-year grace window, so § 102(a) — not § 102(b) — is the correct statutory hook. Flagging this because it is the kind of defect that matters if the same art is repackaged later.

• Institution decision: DENIED — discretionary denial, dated 2025-07-31. The decision paper itself was not retrievable through my sources, so I state the parties' positions rather than the Director's words:

  • Patent owner (Valtrus/KPI) sought denial on "settled expectations" and Fintiv grounds, arguing: (a) Vertiv had been aware of the '287 patent by at least 2009-08-21 and of the Baer reference by at least 2012-07-02, a ~15-year awareness/delay; (b) the patent issued to HPE in 2005 and HPE owned it for nearly 16 years before assigning it into an escrow entity on 2021-01-15; (c) all six Fintiv factors favored denial, including three district-court trials scheduled months before any projected FWD (the DJ action trial then set for 2026-04-06 before Judge Gilstrap, plus the DataBank and TierPoint actions); (d) no stay had been sought or was likely; (e) the Board should not consider the petition at all because Vertiv failed to name all real parties in interest and privies.
  • Petitioner (Vertiv) opposed, arguing the Vidal memo's rescission and the Workload Memo still did not warrant denial: factor 1 neutral (no stay sought); factor 2 weighed against denial because Judge Gilstrap had six patent jury trials calendared for the same 2026-04-06 date, making the trial date speculative; factor 3 favored institution (prompt filing); factor 4 favored institution because no issue overlap existed with the DJ action (which pleaded no invalidity); factor 6 "strongly favors institution" because the merits were "particularly strong."
  • What is documented: the outcome (denial) and the date (2025-07-31). What I will not assert: which specific discretionary factor was dispositive. Given the timing and the surrounding 2025 decisions, "settled expectations" (a patent in force since 2005, with the petitioner's documented awareness since 2009) is the most plausible driver, but I have not read the paper and am not characterizing the Director's reasoning as fact.

• Final Written Decision: None. No institution means no FWD, no claim-level verdict, no cancellation of any claim. I expressly do not claim any claim of the '287 patent was canceled or sustained — none was.

• Settlement / termination: No settlement. Termination was by discretionary denial, not by party agreement. Consistent with non-institution, the Board granted petitioner's refund request for post-institution fees — $28,125.00 refunded to petitioner's deposit account, notice mailed 2025-09-23. Counsel of record: petitioner — Timothy Maloney et al. (Nixon Peabody); patent owner — Patrick Colsher et al. (Reichman Jorgensen Lehman & Feldberg).

• Appeal: None, and none legally available. A § 314(a) discretionary denial is "final and nonappealable" under 35 U.S.C. § 314(d). I found no Federal Circuit appeal involving US 6,854,287 — consistent with the earlier-generated sections of this analysis, which likewise found no CAFC docket tied to this patent. (The 2026 Google/Valtrus CAFC activity concerns Valtrus's search/pay-per-use patents, not the cooling portfolio.)

• Defensive value: Low as a knock-out, real as intelligence. No claim was invalidated, so you cannot lead with the IPR. But you now possess a free, fully developed invalidity theory (Newton, Shimizu, Baer — all pre-2002 § 102 art) that was never tested on the merits, plus an acknowledgement in the patent's own specification (col. 4:49–53) of Vertiv-predecessor commercial cooling systems and a 2001 HPE/Vertiv-predecessor DATACOOL joint-development exhibit. Because no IPR was instituted, § 315(e)(2) estoppel never attached, so those grounds — and the expert declaration filed with the petition — remain fully available in litigation.

• Citation sources: PTACTS petition record for IPR2025-00667 (petition grounds and papers): https://ptacts.uspto.gov/ptacts/public-informations/petitions/[1557496](/patent/1557496)/download-documents · structured case record: https://ipverse.greyb.com/ptab-web/cases/case-details/IPR2025-00667 · PTAB E2E portal for the proceeding: https://ptacts.uspto.gov/ptacts/

---

Strategic summary

Claim status of US 6,854,287. All claims 1–22 are UNTESTED and INTACT at the PTAB. There is no canceled-claims list, no surviving-claims list, and no constructions from any AIA proceeding. The only claim scope development in the entire record is happening in district court — the P.R. 4-3 Joint Claim Construction and Prehearing Statement dated 2025-08-11 in E.D. Tex. lead case 2:24-cv-00361-JRG covers the '287, '277, '179 and '870 patents, with no agreed constructions and both sides disputing terms. The one term that matters most is claim 1's "individually manipulating a mass flow rate of the cooling fluid supplied to each of the plurality of heat exchanger units" — every 2026 complaint (Prime, NTT Global, Iron Mountain, 365 Data Centers, Lumen) anchors on it, and the likely fight is granular per-unit control versus conventional zone-based control. Notably, Valtrus's own P.R. 4-3 appendix cites "IPR2025-00667, Vertiv's IPR Petition and Declaration of John P. [__]" as extrinsic evidence — i.e., the owner is now citing the defeated petition's expert declaration as support for its constructions. That is a two-edged sword a future defendant can exploit.

Estoppel landscape. This is the key point and it is favorable to a defendant. Section 315(e)(2) estoppel attaches only after an IPR "results in a final written decision"; there was none here. So Vertiv — and, by extension, any defendant running the same art — is not barred from asserting Newton (1968), Shimizu (1994), or Baer (2001/0042616) as § 102/§ 103 defenses. The full universe of prior art remains open, including the system/printed-publication art identified in Vertiv's Rule 3-3 invalidity contentions (which expressly reserve unpublished system art and third-party discovery). Conversely, there is no § 315(e)(1) bar on Valtrus either, so a different petitioner could try again tomorrow — subject only to the same discretionary-denial gauntlet and to the § 315(b) one-year clock running from service.

Pattern signals. The same petitioner, Vertiv Corporation, filed a coordinated same-day wave on 2025-02-26: IPR2025-00667 against the '287 patent (app. 10/697,697) and IPR2025-00668 against sibling US 6,718,277 (app. 10/123,403) — and both were discretionarily denied on the identical date, 2025-07-31, both followed by post-institution refunds. That is a single lost campaign, not two independent defeats: Vertiv tried to clear the two patents most central to its customers' exposure and was turned away on discretionary grounds rather than on the merits. Vertiv's posture is unusual and worth tracking — it is simultaneously a declaratory-judgment plaintiff (its DJ action, docketed as member case 2:24-cv-00907-JRG, pleading no invalidity) and a PTAB petitioner, while Valtrus insists there is no case or controversy because it sued Vertiv's customers, not Vertiv. The defensive aggregator pattern is absent: no Unified Patents or RPX-style filer appears on the '287 patent; the only petitioner is the accused equipment supplier itself, acting as indemnitor. As reported in the earlier litigation sections, the 2026 assertions are funneling into MDL No. 3190 in E.D. Tex. — meaning the district court, not the PTAB, is now the sole forum where this patent's validity will be decided.

---

Recommended next steps

1. If you are a defendant and want to attack validity, do not cite an FWD — there isn't one. Instead build on the un-instituted record: (a) obtain the 2025-07-31 denial paper and confirm the stated basis (I could not retrieve it); (b) pull the IPR2025-00667 petition and Dr. John P. [__]'s declaration from PTACTS and reuse that art directly; and (c) note that no estoppel attaches, so the Newton / Shimizu / Baer combination is fully live in court. Because the art is old and applied under pre-AIA §§ 102/103, a defense framed under the "individually manipulating a mass flow rate … to each of the plurality of heat exchanger units" limitation — the term the owner itself now briefs — is the highest-value line.
2. If you are weighing your own IPR petition, recalibrate your expectations. Filing now runs headlong into the current Director-led institution regime (per the 2025-10 "Open Letter" and the November 2025 summary-notice practice), where a patent in force since 2005 with a long-documented competitor awareness is precisely the profile that produced this discretionary denial. Budget for the possibility of a summary "thumbs down" with no reasoning, and note that such a decision is nonappealable under § 314(d). If your § 315(b) window is still open and you have strong merits, file early and address "settled expectations" head-on in the petition rather than leaving it to a discretionary-denial opposition.
3. Trial-stage milestones to watch: none at the PTAB — IPR2025-00667 is terminated and no successor petition is on file as of 2026-09-11. The live clock is the E.D. Tex. claim-construction track (P.R. 4-3 statement filed 2025-08-11; P.R. 4-5(d) chart filed 2025-10-01) and the MDL 3190 consolidation in E.D. Tex.
4. Verification items I could not close:
   • The full text and exact reasoning of the 2025-07-31 discretionary denial in IPR2025-00667 (retrieve from PTAB E2E / PTACTS).
   • APJ identities, if any appear on the denial paper (none surfaced; no merits panel was convened).
   • Whether any second or successor petition against the '287 patent (or an ex parte reexamination) has been filed since 2025-07-31 — the earlier litigation sections note Unified Patents references an ex parte reexam work product "for one of the Valtrus patents," patent not identified. I found no '287 reexam.
   • Reconciliation of the ODP gap with the PTACTS record, so downstream automated queries stop reporting "no PTAB activity."

I'll research the recorded assignment chain for US 6,854,287, focusing on the HPE → OT Patent Escrow → Valtrus links and the correspondent/attorney of record.

Excellent — I found the escrow reel/frame (055269/0001) and the standard HPE correspondent chain. Let me nail down the remaining links: the original inventor→HP assignment, the escrow→Valtrus reel, and the repeat correspondent.

I've hit the tool-call limit, so I'll deliver the report now — flagging clearly what the record confirms versus what remains unverified. All reel/frame figures below are cited to a source; I have not invented any.

---

Ownership & Assignment Reconstruction — U.S. Patent No. 6,854,287 ("Cooling system")

Patent: US 6,854,287 B2 · App. 10/697,697 · Patel & Bash · priority 2002-08-02 · issued 2005-02-15 · anticipated expiration 2022-08-02
As-of date: 2026-09-11
Primary verification link: USPTO Assignment Center — search by patent number 6854287 (mirror: https://assignment.uspto.gov/patent/index.html)

Sourcing caveat (read first): USPTO Assignment Center was not directly reachable during this run. The reel/frame figures below are drawn from (a) the Google Patents legal-events panel embedded in the authoritative patent text, (b) the Vertiv IPR2025-00667 petition record (which cites the '287 patent's own assignment history as Ex. 1003, at 9–10), and (c) a PTACTS-filed "Assignment abstract of title" for a sibling HP patent (App. 10/456,053) that shares the same HPE/Ocean Tomo recording chain. Where a reel/frame is confirmed for the '287 patent itself I say so; where it is inferred from the sibling abstract I flag it. Verify each on Assignment Center before relying.

---

Inventors

Unusual-pattern check: No evidence of an inventor-departure-then-fire-sale pattern. Both inventors were long-tenured HP/HPE research staff, and the patents remained inside HP/HPE for ~15 years post-issuance — the opposite of the "inventors leave within 12 months" tell. I could not independently verify current or departure dates for either inventor, and I am not asserting any departure. The absence of a departure signal is itself unremarkable here: the divestiture was a corporate portfolio sale, not an inventor-driven one.

---

Original assignee

Hewlett-Packard Development Company, L.P. (Houston, TX), per the issued patent and the Google Patents assignee field.

• Primary line of business: Operating technology company — computing hardware, enterprise systems, servers, and data-center infrastructure.
• Product embodying the claims: Yes, in the relevant sense. The '287 claims are directed to data-center cooling (heat-exchanger units fed chilled fluid from an air-conditioning unit, with per-unit mass-flow control). Vertiv's IPR petition expressly notes that the '287 specification acknowledges "Vertiv's predecessor technology … as commercially available data center cooling systems" (Ex. 1001, 4:49–53) and references a 2001 HPE/Vertiv-predecessor joint development of "DATACOOL" technology (Ex. 1013). The patent also names the Liebert "DATACOOL" environmental control system. So HP/HPE's own business ecosystem and its named supplier partner shipped cooling equipment practicing this space.
• Current status: HPE is a large, solvent, publicly traded operating company. The patent was not sold via insolvency (contrast the Kodak/Nortel/Polaroid pattern). It was sold as part of a deliberate portfolio divestiture.

---

Assignment timeline

The '287 patent has a complete, multi-hop recorded chain — this is not a "no assignments recorded" case.

2002-08-02 — priority / parent application Ser. No. 10/210,040 filed (Patel & Bash, assigned to HP).

• Conveyance: (inventor → HP employment/assignment; see note below)
• Assignor: Patel, Chandrakant D.; Bash, Cullen E.
• Assignee: Hewlett-Packard Company / Hewlett-Packard Development Company, L.P.
• Correspondent: Not retrieved for the '287 patent. For the sibling HP patent (App. 10/456,053) the equivalent inventor-assignment was recorded much later (reel 035415/0887, executed 2003-06-06, recorded 2015-04-15) with correspondent Hewlett-Packard Company, 3404 E. Harmony Road, MS 35, Fort Collins, CO — i.e., HP's own in-house IP-docketing address. The '287's inventor-assignment reel/frame is unverified.
• Context: Initial inventor assignment to the operating company (internal).

2015-10-27 (executed) / 2015-11-09 (recorded) — Reel 037079/0001

• Conveyance: Assignment of assignor's interest (corporate reorganization / HPE spinoff)
• Assignor: Hewlett-Packard Development Company, L.P.
• Assignee: Hewlett Packard Enterprise Development LP (11445 Compaq Center Drive West, Houston, TX 77070)
• Correspondent: Hewlett Packard Enterprise, 3404 E. Harmony Road MS 79, Fort Collins, CO 80528 — HP/HPE's own in-house IP address, not outside counsel. This is an internal-reorg tell, not an NPE tell.
• Context: Internal reorganization — bulk carve-out of patents into the spun-off HPE entity. The sibling abstract shows this single reel covered 11,870 pages / 1,112 properties, confirming it is a mass assignment, not a targeted transfer.
• Confidence: the execution/recording dates align with the Google Patents legal event for the '287 patent (2015-11-09). Reel 037079/0001 is confirmed from the sibling abstract; treat as applicable to the '287 chain unless Assignment Center shows a different reel for this specific patent.

2021-01-15 (executed) / 2021-01-26 (recorded) — Reel 055269/0001 ✅ confirmed for the '287 patent

• Conveyance: "Patent Assignment, Security Interest, and Lien Agreement" (an escrow-vehicle transfer)
• Assignor: Hewlett Packard Enterprise Company; Hewlett Packard Enterprise Development LP
• Assignee: OT Patent Escrow, LLC (c/o Ocean Tomo, LLC)
• Correspondent: Not retrieved for the '287 reel. Ocean Tomo / OT Patent Escrow recordations in this campaign were handled by Botos Churchill IP Law LLP, 430 Mountain Avenue, Suite 401, New Providence, NJ 07974 (confirmed as correspondent on the Valtrus-side recording of the sibling App. 10/456,053, reel 055360/0424). I flag Botos Churchill as the probable correspondent for this escrow recording but have not confirmed it on reel 055269/0001 itself.
• Context: Transfer-to-asserter (step 1 — the escrow/monetization vehicle). This is the point at which the patent left operating-company hands. Vertiv's petition cites this reel directly, noting HPE "owned the '287 patent for almost 16 years, from … February 15, 2005 until January 15, 2021, when it assigned the patent to an escrow entity."

2022-02-XX (executed, "early February") / 2022-02-09 (recorded) — Reel NOT RETRIEVED

• Conveyance: Assignment of assignor's interest
• Assignor: OT Patent Escrow, LLC
• Assignee: Valtrus Innovations Limited — The Glass Houses, 92 Georges Street Lower, Dun Laoghaire, Dublin A96 VR66, Ireland (f/k/a Dolya Holdco 9 Limited; parent/beneficiary Key Patent Innovations Limited ("KPI"))
• Correspondent: Botos Churchill IP Law LLP, New Providence, NJ (probable — confirmed as correspondent on the parallel Valtrus recording for sibling App. 10/456,053 at reel 055360/0424; not confirmed on the '287's own reel). RPX independently reports the assignment was "executed in early February," matching the 2022-02-09 Google Patents recording date.
• Context: Transfer-to-asserter (step 2 — arrival at the asserting NPE). After this point the patent is held by a licensing-only Irish entity controlled by KPI.

Same day as the purchase — Declaration of Trust (Valtrus → KPI) — not a recorded patent assignment; documented in litigation.

• Conveyance: Declaration of trust (equitable, not recorded in the USPTO assignment records)
• Grantor/Trustee: Valtrus Innovations Limited
• Beneficiary: Key Patent Innovations Limited
• Context: Beneficial-ownership separation. Per the N.D. Cal. order in Valtrus v. Google (Case No. 25-cv-07063-PCP), Valtrus kept legal title but "transferred all of its exclusionary rights" to KPI, leaving Valtrus "at most a bare right to sue." The court dismissed Valtrus for lack of constitutional standing (order on file; hearing 2025-12-18). This is the single most important ownership feature of the chain and is invisible in Assignment Center, because the trust was not recorded there.

Timeline note on the "no-records" instruction: Assignment Center does have records for this patent — the chain is HP → HPE → OT Patent Escrow → Valtrus. The section therefore continues rather than stopping.

---

Timeline diagram

timeline
    title Ownership of US 6854287
    2002 : Parent application filed
         : Inventors Patel and Bash
    2005 : Issued to Hewlett-Packard Development
    2015 : Reorg to HPE Development LP
         : Reel 037079 0001
    2021 : HPE to OT Patent Escrow LLC
         : Reel 055269 0001
    2022 : OT Patent Escrow to Valtrus Innovations
         : Valtrus declares trust for Key Patent
    2024 : First suit asserting the 287 patent
    2025 : Vertiv IPR denied institution
    2025 : Court finds Valtrus lacks standing

---

NPE / troll-pattern signals

1. Shell-entity transfer — PRESENT.
   Patent moved from an operating company (HP/HPE) through a transactional escrow vehicle to a licensing-only entity. Evidence: HPE → OT Patent Escrow, LLC at reel 055269/0001 (executed 2021-01-15); then → Valtrus Innovations Limited (recorded 2022-02-09). Valtrus is routinely described by RPX and in litigation as "an Irish NPE," holds no products, and is a single-purpose holder in Dublin whose predecessor name was "Dolya Holdco 9 Limited" — a classic holdco naming. The escrow intermediary (OT Patent Escrow / Ocean Tomo) is itself a tell: escrow vehicles exist to park assets between seller and buyer.

2. Known asserter in the chain — PRESENT.
   Valtrus Innovations is a documented patent-assertion entity, not a member of the older Acacia/Marathon/Vringo cohort but a modern high-frequency plaintiff tracked by RPX and hit by an IPR campaign (IPR2022-01399, IPR2022-01497, IPR2025-01490, and the Vertiv IPRs). Its parent Key Patent Innovations Limited (KPI) operates three assertion portfolios — Valtrus (ex-HPE), Malikie Innovations (ex-BlackBerry), Pictiva Displays (ex-OSRAM). The '287 patent sits squarely in the Valtrus/data-center-cooling campaign asserted against Lumen, Prime Data Centers, NTT Global, Iron Mountain, Netrality, 365 Data Centers, and others (per the earlier litigation sections and the 2026 press coverage).

3. Repeat correspondent across the chain — PRESENT (moderate; one appearance not yet confirmed on this patent's own reel).
   Botos Churchill IP Law LLP (New Providence, NJ) appears as the recording correspondent on the Valtrus-side assignment for the sibling HP patent App. 10/456,053, reel 055360/0424 (executed 2021-01-21, recorded 2021-02-22) — the same campaign and the same assignee. The recurrence signal is real at the campaign level: Ocean Tomo/OT Patent Escrow and Valtrus recordings consistently funnel through Botos Churchill, while HPE's internal reorg recordings used HPE's own in-house Fort Collins address (reel 037079/0001). I could not confirm Botos Churchill on the '287's actual escrow (055269/0001) or escrow→Valtrus reel. Do not upgrade this to a strong finding until the correspondent is pulled for this patent specifically. (Note: Botos Churchill is a general IP firm that does non-NPE work too, so recurrence — not a single appearance — is the operative test.)

4. Cascading transfers — PARTIAL / MIXED.
   Three recorded hops (2015 → 2021 → 2022), but the first two gaps are ~5.2 years and ~5.2 years respectively — not a rapid <24-month cascade. The escrow→Valtrus hop (2021-01-15 → 2022-02-09) is ~12.8 months, which is inside the 24-month window and is the leg that matters. So: not a classic multi-LLC cascade, but the final step is a tight, purpose-built hand-off to the asserting entity.

5. Pre-litigation transfer — NOT PRESENT as to this patent.
   The escrow→Valtrus transfer (Feb 2022) preceded the first suit asserting the '287 patent (Valtrus v. NTT Data Services, E.D. Tex. 2:24-cv-00361, complaint 2024-05-14) by ~27 months — outside the 6-month window. Valtrus's first ever suit (v. Google, N.D. Tex. 3:22-cv-00066, January 2022) used search/software patents, not the '287. So the transfer was not timed to a specific '287 suit, though it was plainly arranged to enable the overall monetization campaign.

6. Bankruptcy fire-sale — NOT PRESENT.
   No Chapter 7/11. HPE is solvent; the divestiture was a voluntary portfolio sale (RPX: HPE offloaded 800+ patents via four 2021 assignments, then more in 2022, exceeding 1,300 by mid-2022). This was securitization/divestiture, not distress.

7. Privateering — NOT PRESENT / UNCLEAR.
   No evidence HPE retained a financial stake in the '287 patent, directs Valtrus's assertions, or shares in recoveries. The patents were sold outright through an escrow intermediary to a third party. This is a straight portfolio monetization sale rather than classic privateering. What is present is a monetization-campaign structure (Valtrus trustee + KPI beneficiary), which is asserted-interest-driven but is not HPE-directed privateering.

8. Defensive aggregator (anti-NPE) — NOT PRESENT.
   The chain terminates at Valtrus Innovations Limited / Key Patent Innovations Limited, an asserting NPE. It does not end at RPX, AST, LOT, Unified Patents, or OIN. The patent has not been neutralized; it is live and being asserted (subject to the standing ruling discussed below).

---

Verdict

NPE — high confidence.

Three strong signals converge on the record: (1) shell/escrow transfer — HPE → OT Patent Escrow, LLC at reel 055269/0001 (executed 2021-01-15), then → Valtrus Innovations Limited (recorded 2022-02-09), an Irish licensing-only entity f/k/a Dolya Holdco 9 Limited; (2) known asserter — Valtrus is a documented, high-frequency NPE and the center of a multi-defendant data-center-cooling assertion campaign tracked by RPX and challenged at the PTAB; and (3) beneficial-ownership separation — on the same day Valtrus took title it declared a trust vesting all exclusionary rights in parent Key Patent Innovations Limited, which the N.D. Cal. court held left Valtrus with "at most a bare right to sue" and dismissed it for lack of constitutional standing. The absence of a bankruptcy or pre-suit-timing signal does not change the verdict; the current assignee is a pure assertion entity, not an operating company.

Link for verification: https://assignmentcenter.uspto.gov/ (search patent 6854287)

---

Verification items I could not close

1. The escrow→Valtrus reel/frame for the '287 patent — the last recorded hop's reel number was not retrieved. Pull it from Assignment Center.
2. The correspondent of record on reel 055269/0001 (HPE→OT Patent Escrow) and on the escrow→Valtrus reel — I confirmed Botos Churchill only on a sibling patent's Valtrus recording (reel 055360/0424), not on this patent's own reels.
3. The original inventor→HP assignment reel/frame for the '287 — not retrieved (sibling patent used reel 035415/0887, recorded 2015-04-15, HP in-house correspondent).
4. Whether the 2015 HPE reorg reel 037079/0001 is in fact the reel covering the '287 — the date matches the Google Patents event (2015-11-09), but the reel is confirmed only from the sibling abstract.
5. Assignment Center's current "assignee of record" may still list Valtrus/HPE Development LP and will not reflect the KPI trust (unrecorded). Confirm the recorded title-holder before assuming the chain is complete.

Cross-reference note (no contradiction with prior sections): the earlier Litigation and PTAB sections described the chain as 2015-11-09 (HPE) → 2021-01-26 (OT Patent Escrow) → 2022-02-09 (Valtrus). This reconstruction is consistent and adds the reel/frame anchors (037079/0001; 055269/0001) plus the previously-unrecorded Valtrus→KPI trust, which is the fact that has driven the standing dismissals.

I'll research the prior art cited on the face of US 6,854,287 and in the PTAB petition record.

Let me pull the front-page reference list from the patent databases directly.

I have the PTAB art. Now let me retrieve the examiner-cited references on the patent's face (the "Patent Citations" table).

Prior-Art Report — U.S. Patent No. 6,854,287 B2 ("Cooling system")

Patent: US 6,854,287 B2 · App. 10/697,697 · filed 2003-10-31 · divisional of App. 10/210,040 (filed 2002-08-02) · issued 2005-02-15 · Patel & Bash · original assignee Hewlett-Packard Development Co. LP
As-of date: 2026-09-11 (see the date contradiction flagged in the earlier Litigation section — the task prompt says "April 26, 2026," but all retrieved docket/PTAB records run through September 2026; I report events by their recorded dates)

---

⚠️ Threshold disclosure on what I could and could not retrieve

You asked me to go "citation by citation" through the references cited on the face of 6,854,287 (i.e., the examiner-cited "U.S. Patent Documents" table on the printed front page).

I could not retrieve that front-page citation table. The authoritative Google Patents full text supplied in this task renders the specification, claims, classifications, and legal-events panel — but omits the "Patent Citations" / "References Cited" table, which that site loads dynamically. My follow-up searches against FreePatentsOnline, Justia, uspto.report, and Golden Wiki returned string-matches only (mostly forward citations — later patents citing the '287 — which are not prior art). I hit the tool-call limit before closing this gap.

Therefore I am NOT going to invent examiner-cited reference numbers. Fabricating a "5,465,778 / 6,185,482" style list would be exactly the failure mode your strict-ID rule is designed to prevent. What I can give you, fully source-grounded, is:

1. The IPR2025-00667 prior-art set — the only prior art actually applied against the '287 claims in any contested proceeding, with grounds, dates, and claim mappings verified from the petition record.
2. The specification-acknowledged prior art (including a commercial system).
3. The incorporated-by-reference applications, which are a distinct category (co-owned, not examiner-cited) that I separate deliberately.
4. A clear statement of what remains unverified, with the exact route to close it.

---

1. PTAB-applied prior art (verified — IPR2025-00667, Vertiv v. Valtrus)

Source: Vertiv's petition, PTACTS petition record 1557496 (https://ptacts.uspto.gov/ptacts/public-informations/petitions/1557496/download-documents) and the GreyB/IPVerse case record (https://ipverse.greyb.com/ptab-web/cases/case-details/IPR2025-00667). This is the most probative prior art on the patent, because a skilled searcher (Vertiv's counsel) already mapped it claim-by-claim. Recall from the earlier PTAB section: institution was discretionarily denied 2025-07-31 on non-merits grounds, so none of this was adjudicated — it is untested but fully developed.

What this tells you about the '287 claims' vulnerable core. All three primary references are mapped to the same independent-claim skeleton: pre-2002 multi-unit cooling with cooling fluid, temperature sensing, and per-unit flow control. The claims that survive every reference in the set are the dependent range-control claims (5–6, reached only by § 103) and — not touched at all by this set — the system claims 10, 14, and 16.

---

2. Specification-acknowledged prior art (verified from the patent text itself)

---

3. Cross-referenced / incorporated applications (verified, but NOT § 102 prior art in the ordinary sense)

The '287 "Cross-References" section incorporates three co-pending applications by reference:

• U.S. App. 09/970,707, filed 2001-10-05, "SMART COOLING OF DATA CENTERS," Patel et al.
• U.S. App. 10/157,892, filed 2002-05-31, "CONTROLLED COOLING OF A DATA CENTER," Patel et al.
• U.S. App. 10/122,010, filed 2002-04-24, "DATA CENTER ENERGY MANAGEMENT SYSTEM," Friedrich et al.

Why these are not in the § 102 tables above: all three are commonly owned by the same assignee (HP) and are the patent's own related work. Two consequences under pre-AIA law: (a) as co-owned applications they are disqualified as § 103 obviousness art by pre-AIA § 103(c); and (b) the '287 specification expressly relies on them for enablement of the "anticipated sensing" and load-balancing features. They matter to your analysis as support/derivation risk, not as anticipatory art — but note that App. 09/970,707 (filed 2001-10-05) predates the '287 priority date by ~10 months, so if any of its subject matter were ever established to be not commonly owned, its § 102(e) date would be live. Worth a targeted check of the assignment records for 09/970,707 (I did not verify its ownership chain).

---

4. What I could not verify (and how to close each gap)

---

5. Bottom line

• Most relevant prior art, period: the Newton (1968) / Shimizu (1994) / Baer (US 2001/0042616 A1, pub. 2001-11-22) set. It is the only art ever mapped claim-by-claim against 6,854,287 (claims 1–4, 7–9 under § 102; 5–6 under § 103), and because IPR2025-00667 was denied institution without a merits ruling, none of it was adjudicated and no § 315(e)(2) estoppel attached — it remains free to be reused in court.
• Anticipation targets: at minimum, claims 1–4, 7–9 (Newton), 1–4, 7, 9 (Shimizu), and 1, 3, 7, 9 (Baer). Claims 10, 14, 16 were not addressed by this set and remain the untouched system-claim flank.
• Baer caveat: mislabeled § 102(b) when it is really § 102(a) — a grace-window defect any defendant must fix before re-asserting it.
• I did not fabricate the examiner-cited list. The front-page citations for 6,854,287 are unverified because the source text omitted the citation table; item 1 in §4 is the single highest-priority verification step, and I flag it explicitly rather than guess.

Citation sources: https://patents.google.com/patent/US6854287/en (authoritative full text) · https://ptacts.uspto.gov/ptacts/public-informations/petitions/1557496/download-documents (petition grounds, Ex. 1007/1008/1011) · https://ipverse.greyb.com/ptab-web/cases/case-details/IPR2025-00667 (proceeding status) · https://FreePatentsOnline.com/6854287.html (specification).

Cross-reference note (no contradiction): this report is consistent with the earlier PTAB section's Art/Ground table; I have deliberately repeated those three references because you asked for a reference-by-reference anticipation mapping, and I have added the category analysis (examiner-cited vs. petition-applied vs. specification-admitted vs. incorporated-by-reference) that the earlier section did not break out.

I'll do a focused round of verification searches before writing the § 103 analysis, since I want to avoid inventing reference identities.

Excellent — that gave me the actual reference numbers. Let me verify the remaining references and the petition's stated grounds/rationale.

Obviousness Analysis — U.S. Patent No. 6,854,287 under 35 U.S.C. § 103

Patent: US 6,854,287 B2 ("Cooling system") · App. 10/697,697 · filed 2003-10-31 · divisional of App. 10/210,040 (filed 2002-08-02) · issued 2005-02-15 · Patel & Bash · originally Hewlett-Packard Development Co. LP
As-of date: 2026-09-11 (system date of record)

---

0. Two threshold notes before the analysis

(a) Date contradiction (carried forward, not reconciled). The task prompt states "Current Date: April 26, 2026." The system date of record is 2026-09-11, and events in the earlier sections run through September 2026 (MDL 3190 conditional transfers docketed 2026-08-07 / 2026-08-11 / 2026-09-09). For a § 103 analysis the calendar is largely immaterial — the operative dates are the 2002-08-02 priority date and the pre-AIA framework — but I flag the contradiction because the earlier sections did.

(b) This analysis upgrades and corrects the Prior Art section. The earlier Prior Art section could not retrieve the reference numbers for "Newton" and "Shimizu" and could not retrieve the front-page citation table. I have now pulled the IPR2025-00667 exhibit list, which supplies the actual identifiers. Two corrections/upgrades to the earlier section:

Sources: PTACTS petition record 1557496 — exhibit list and challenged-claim listing at ...?artifactId=DEdTgFci6HV03xTRe6WSlaBsm3aVH0Al9qtfgcZFduoR8eq2wKn4L8Q; petition body/claim mapping and motivation section at ...?artifactId=1kiQK74PirYFNzYW5oVoGCDcMqylGa_FTsQ1AEdx6avtPJwmleV6zaY; Newton full text at https://patentimages.storage.googleapis.com/5a/44/75/96ff3858e5be82/[US3384155](/patent/US3384155).pdf; proceeding status at https://ipverse.greyb.com/[ptab](/ptab)-web/cases/case-details/IPR2025-00667.

---

1. Governing legal framework

Which § 103 applies. The '287 application was filed 2003-10-31, and its parent (10/210,040) on 2002-08-02 — both before the AIA's 2013-03-16 first-inventor-to-file cutoff. Pre-AIA 35 U.S.C. § 103(a) governs, as do pre-AIA §§ 102 and 103(c). The petition itself acknowledges the pre-AIA framework. This matters because § 103(c) (commonly-owned art) exists only pre-AIA, and because the AIA's § 102(a)(1)/(a)(2) grace-period structure does not apply.

The controlling standards.

• Graham v. John Deere Co., 383 U.S. 1 (1966): scope and content of the prior art; differences between the prior art and the claims; the level of ordinary skill; secondary considerations.
• KSR Int'l Co. v. Teleflex Inc., 550 U.S. 398 (2007): the central question is "whether the improvement is more than the predictable use of prior art elements according to their established functions." A combination is obvious where "a skilled artisan would have been motivated to combine the teachings of the prior art references to achieve the claimed invention." KSR also directs that the POSITA "is also a person of ordinary creativity, not an automaton," and that design incentives and market forces count as motivations.
• MPEP § 2143 — the enumerated rationales I apply below: (A) simple substitution of one known element for another; (B) use of a known technique to improve a similar device in the same way; (C) applying a known technique to a known device ready for improvement; (D) applying a known technique to a known device ready for improvement to yield predictable results; (E) "obvious to try"; (F) design incentives / market forces; (G) a teaching or suggestion in the art.
• In re Keller, 642 F.2d 413 (CCPA 1981): the test is what the combined teachings suggest, not bodily incorporation of one reference into another.
• In re Fracalossi, 681 F.2d 792 (CCPA 1982): a disclosure that anticipates also renders the subject matter obvious — the obviousness question is subsumed in the anticipation question.
• Analagous art: all candidate references are in the same field of endeavor (temperature-conditioned enclosed spaces) or are reasonably pertinent to the problem (metering a coolant to a heat exchanger in response to local sensed air temperature). No reference is non-analogous.
• Burden: at the PTAB, preponderance; in district court, clear and convincing evidence ([Microsoft Corp.](/litigations/by-plaintiff/Microsoft%20Corp.) v. i4i Ltd. P'ship, 564 U.S. 91 (2011)), because the '287 patent enjoys a statutory presumption of validity. Since IPR2025-00667 was denied institution, the invalidity case would be made in court — so the clear-and-convincing overlay applies to every combination below.

---

2. Level of ordinary skill in the art (PHOSITA)

The petition's own expert frames the field as "thermal management of buildings and rooms containing computer equipment as of August 2, 2002." (Ex. 1002 ¶ 9.)

My articulation (consistent with, and anchored to, that framing): a person of ordinary skill would have had a bachelor's degree in mechanical engineering (or equivalent) and two to four years of experience in HVAC / building-automation design, including hydronic (chilled-water) distribution and its controls; or a master's degree plus one to two years of such experience. That person would be familiar with: central chilled-water plants feeding multiple terminal units; two-way and three-way control valves; variable-speed pumping; proportional/integral control of supply-water temperature; and the then-current generation of computer-room air conditioners.

Why this level is not outcome-determinative here. The asserted advance — per-unit, sensed-temperature-driven fluid-side metering in a multi-unit arrangement — is a control-architecture choice squarely within that skill set. Under KSR, where the differences between the claim and the art are "small in scope" and the technical effects are predictable, a more explicitly stated source of motivation may not be required. The obviousness case is therefore relatively insensitive to fine gradations between a "2-year" and a "4-year" PHOSITA.

---

3. Scope and content of the prior art

3.1 The four references actually applied (Ex. 1007, 1008, 1011) plus the commercial admission (Ex. 1004)

3.2 Additional references in the petition exhibit list (use not confirmed)

Stahl (U.S. 6,556,624, Ex. 1005), Patel (U.S. Pub. 2003/0067745 A1, Ex. 1006), Feeney (U.S. 5,467,609, Ex. 1009), and Arima (U.S. 6,006,528, Ex. 1010) appear in the exhibit list. The retrievable petition excerpts do not show the ground in which each was used. I do not assign them a claim mapping. ⚠️ See § 7.4 for a § 103(c) commonly-owned trap that likely affects Ex. 1006 ("Patel") — a publication bearing the same surname as the '287's first inventor.

3.3 What the petition did not do — and why that matters

Per the earlier PTAB section, the petition pleaded four grounds:

Only one ground in the entire petition was a § 103 ground, and it was confined to claims 5–6. There was no pleaded combination (Newton + Baer, Newton + Shimizu, etc.), and claims 10, 14, and 16 (all system claims) were not challenged at all. Because institution was denied on 2025-07-31, § 315(e)(2) estoppel never attached and the merits were never reached. The consequence: the § 103 space below is entirely unoccupied, and no adjudicated position on obviousness exists anywhere in the record. Everything that follows is a fresh, first-principles § 103 case.

---

4. Claim 1 — element-by-element obviousness

Claim 1 (independent method claim) has six limitations. Mapping them against the art:

Assessment on claim 1. Limitation 1[f] — the limitation every 2026 complaint anchors on (per the Litigation and PTAB sections) — is the one the prior art meets most directly. Newton discloses per-unit fluid metering on individual branch lines, driven by per-unit sensed zone temperature, and discloses it twice in two different hardware forms (valve and pump). Shimizu independently discloses the same architecture in a refrigeration-cycle rather than hydronic system.

Claim 1 is, at minimum, obvious as a matter of law if Newton anticipates it (In re Fracalossi). Should Valtrus defeat Newton's anticipation case on the "air conditioning unit" construction (see § 7.1), the answer is not that claim 1 is valid — it is the combination grounds below.

---

5. The § 103 combinations, with articulated motivations

I set out six combinations. Each is stated with (i) the references, (ii) the resulting claim coverage, and (iii) the MPEP § 2143 rationale(s) supporting the motivation to combine.

Combination 1 — Newton + Baer (and/or the DataCool admission)

Renders obvious: claim 1; and (with Newton alone) claims 2–4, 7–9.

What each supplies. Newton supplies all of the control architecture — plurality of units, fluid supply, heat exchange, sensed temperature, per-unit mass-flow manipulation (valves and, alternatively, pumps), and fan-speed control. Baer supplies the data-center framing: electronics enclosures, rack-level heat densities exceeding air-cooling capacity, and an air-to-liquid heat exchanger "connectable to an external cooling source." Baer is assigned to Liebert Corporation — the same entity whose DATACOOL system the '287 specification names as its own exemplary HEU. The two references therefore come from the same commercial and technical lineage.

Motivation to combine (MPEP 2143 rationales B, F; KSR "design incentives"):

1. Same field / reasonably pertinent. Both address rejecting heat from an occupied space by transferring it to a fluid, and both are directed to accommodating non-uniform, time-varying heat loads across multiple locations. Newton's stated object is to provide "a low-cost air conditioning system in which the temperature in various zones can be individually controlled."
2. Market forces. Baer's own background documents the runaway rack heat density (">11 kW … from just the microprocessors"). The '287 patent's own Background supplies the economic pressure: data-center cooling costs ≈ 50% of IT power; CRAC units "operate at or near a maximum compressor power even when the heat load is reduced inside the data center"; a 100-rack/10 kW-per-rack center needs 1 MW of cooling. KSR makes these precisely the "design incentives and other market forces" that supply motivation.
3. Predictable result. Transplanting Newton's zonal, per-unit hydronic control into a data-center room served by Baer-style liquid-cooled units yields nothing more than the aggregate of each element's known function: each unit gets the coolant flow its local sensor calls for. Newton's control algorithm requires no modification — only a different room.
4. No teaching away. Baer does not disparage central fluid distribution; Newton does not disparage enclosure-level cooling. Nothing in the record indicates the combination would be inoperable.

Practical note: this combination is materially stronger than Ground 4 (Baer alone as § 102(a) anticipation) because it does not depend on Baer alone teaching per-unit metering — Newton supplies that, removing the grace-window/swear-behind vulnerability discussed in § 7.2.

---

Combination 2 — Newton + Shimizu (the strongest pairing)

Renders obvious: claims 1–4 and 7–9 — i.e., the same claim set the petition attacked only as anticipation, now supported by an independent § 103 route.

What each supplies. Shimizu supplies a compressor-based air conditioning unit and a second, independent disclosure of per-unit fluid metering driven by local air temperature — with each personal unit E having its own flow control valve 41 and its own air sensor 48, and with compressor 21 motor speed varied in response to the total load. Newton supplies the multi-zone, multi-unit hydronic analogue, the valve-vs-pump interchangeability teaching, and fan-speed control.

Motivation to combine (MPEP 2143 rationales A, B, E; KSR "predictable use of prior art elements"):

1. Simple substitution of one known element for another. Newton is a hydronic (chilled-water) multi-unit loop; Shimizu is a vapor-compression multi-unit loop. The substitution of one meterable, temperature-conditioned fluid loop for the other is the paradigm § 2143(A) case. Each reference achieves the same result — modulating unit capacity by modulating coolant flow — through the same mechanism (a per-unit flow-control device set by a controller reading a per-unit air-temperature sensor).
2. The combination yields the claim-2 element that neither reference has cleanly alone. Claim 2 requires "varying an output of said air conditioning unit to control the temperature of said cooling fluid." Shimizu's variable compressor speed is that element; Newton's chiller/heater switching via valve 18 is closer to a mode change than to an output variation. Newton + Shimizu covers claim 2 cleanly.
3. Reasonable expectation of success. Both are closed-loop control systems whose inputs (sensed temperature) and outputs (valve position / compressor speed) are of the same type. No reconception of either system is required; the engineer is tuning control gains and setpoints, not inventing a mechanism.
4. "Obvious to try." Given a room with multiple separately-loaded cooling units and a sensed-temperature control input, there are only a handful of ways to modulate per-unit cooling: meter the fluid, modulate the air mover, or modulate the central plant output. All three are disclosed in the combined art (Newton valve 36/pump 70, Newton FIG. 5 fan speed, Shimizu compressor 21).

---

Combination 3 — Newton + Shimizu + POSITA knowledge (claims 5–6, the "sum" limitation)

Renders obvious: claims 5 and 6.

The limitation. Claim 5: increase cooling-fluid temperature when "a sum of the sensed temperatures at one or more locations" is below the predetermined range. Claim 6: the converse.

What each supplies. Newton teaches assessing the aggregate building demand: "if the cooling loads are above some predetermined level (approximately 75% of the maximum load for the entire building), valve 18 will be positioned … so as to continuously circulate chilled fluid from chiller 12." Shimizu computes a "total air conditioning load" from the comparison of multiple sensed room temperatures to preset targets, and drives compressor speed from it.

Motivation to combine (MPEP 2143 rationales D, E, F):

1. Limited number of ways to assess aggregate thermal load. Aggregating multiple local temperature indications into a single control variable is the entire purpose of Newton's and Shimizu's central-plant control. Summing is one of a small number of obvious aggregation operators (sum, average, max, weighted average). Under KSR, where the prior art discloses the function — assess total load, then adjust central supply temperature — selecting a summing implementation is a predictable design choice, not an inventive step.
2. Design incentive for a two-level control hierarchy. The '287 specification itself discloses exactly this two-level scheme: per-zone air-delivery control handled by the HEU controller and global coolant-temperature control handled by the cooling-device controller ("on a zonal level … on a global level"). The claimed invention's own architecture reflects the standard split between terminal-unit control and plant control already present in Newton (per-zone valves 36 vs. plant-level valve 18) and Shimizu (per-unit valves 41 vs. compressor 21).
3. No unexpected result. The specification attributes no surprising or superior property to summing as opposed to any other aggregation. Where the result is the predictable working of a known control law, KSR controls.

Candid weakness: this is the least certain of the six combinations. "Sum of sensed temperatures" is an arguably odd, even crude, control variable — summing absolute temperatures (not deviations or loads) is dimensionally unusual and invites a Valtrus argument that the claim recites a specific algorithm, not a mere design choice. Still, the specification's silence on any unexpected benefit, combined with Newton's and Shimizu's express aggregate-load teachings, makes obviousness more likely than not — but a defendant should lead with claims 1–4 and 7–9 and treat 5–6 as secondary.

---

Combination 4 — Newton (FIG. 6) + Baer/DataCool (system claim 10 — pumps)

Renders obvious: claim 10. Note that claim 10 was never challenged at the PTAB at all.

What the claim requires. A system for cooling a room with racks of computer systems, comprising: a plurality of HEUs receiving cooling fluid from an air conditioning unit and receiving air through openings; at least one fan per unit; a heat-exchanger controller controlling coolant supply and fan speed; temperature sensors; an air conditioning unit controller varying the temperature of cooling fluid delivered to the locations; and "a plurality of pumps configured to control cooling fluid delivery to respective ones of said plurality of heat exchanger units," with the controller able to "individually control the mass flow rate of cooling fluid delivered into each of the respective heat exchanger units."

The prior art meets it nearly element-for-element.

• Plurality of pumps, each on a respective unit: Newton FIG. 6 explicitly "replaces the water flow control valve 36 at the inlet of each room air conditioning unit with a pump 70", with control circuit 38′ driving motor 72 "to manipulate the water flow in response to sensed temperature conditions in the zone." That is literally per-unit pump-based mass-flow control.
• Fan per unit: Newton's fan 22, with FIG. 5 fan-speed control.
• Controller + sensors: Newton's control 38/38′ plus bulbs 40, 42 — and Newton expressly teaches that "automated fan speed control can also be used in embodiments using pumps to control the water flow to the room air conditioning units" (4:35-38), i.e., the controller governs both coolant flow and fan speed — exactly the claim-10 controller.
• AC-unit controller varying coolant temperature delivered to room locations: Newton's timer 19 / three-way valve 18 switching chiller 12 into the loop; more cleanly, Shimizu's variable-speed compressor 21.
• Racks of computer systems, openings, air in/out: Baer (enclosure air inlet/outlet, air-to-liquid HX) + Ex. 1004 DataCool + the specification's own admission.

Motivation to combine (MPEP 2143 rationales A, B, C, D):

1. Two-component invention, each component separately known. Per MPEP 2143, combining two known prior-art systems and employing each "according to its established function" is obvious. Newton uses its pumps for per-unit coolant metering; Baer uses its liquid-cooled enclosures to reject rack heat. Doing both at once is the expected use of each.
2. Simple substitution. Newton itself treats valve 36 and pump 70 as interchangeable alternatives for the same function — the strongest possible evidence that the substitution was within ordinary skill. Claim 8 (method) and claim 10 (system) recite exactly that substitution.
3. Market forces. See Combination 1, rationale 2.

---

Combination 5 — Newton (valve 36) + Shimizu (valves 31/41) (system claim 14 — valves)

Renders obvious: claim 14 — the valve-based system claim, also never challenged at the PTAB.

What the claim requires. Same architecture as claim 10, but with "a plurality of valves configured to meter the flow of cooling fluid through each of said plurality of heat exchanger units positioned along respective cooling fluid lines generally upstream of respective heat exchanger units," with the controller individually controlling mass flow through the valves.

The prior art meets it literally. Newton's valve 36 sits on the individual inlet line 25 feeding heat exchanger 24 — i.e., upstream of the respective heat exchanger unit, on the respective cooling-fluid line ("individual inlet lines 25" taken off main conduit 10). Control 38 sets each valve 36 independently in response to that zone's bulb 40. Shimizu's valves 31 and 41 likewise sit upstream of indoor heat exchangers 34 and 45 on respective refrigerant lines, each independently set to pass "refrigerant of an amount corresponding to the air-conditioning load" sensed at that unit.

Motivation to combine (MPEP 2143 rationales A, E): This is the weakest need for combination — the valve-based architecture is disclosed twice, redundantly, in mechanically analogous positions, in two references from adjacent decades. The motivation is the same limited-options reasoning plus the "known technique to improve a similar device" rationale. Claim 14 is the strongest invalid-over-prior-art candidate in the entire patent on a per-element basis.

---

Combination 6 — Newton + Shimizu + Baer (claim 16, means-plus-function)

Renders obvious: claim 16, subject to the caveat below.

Claim 16 is the means-plus-function system claim. Construing it under pre-AIA § 112, ¶ 6 (the '287 was examined and issued in 2005), each "means" is limited to the corresponding structure disclosed in the specification and equivalents thereof:

Motivation and § 103 treatment: The MPEP § 2143 rationales from Combinations 1–2 apply. The § 112, ¶ 6 framework does not help Valtrus: it narrows the means to the disclosed structures and equivalents, all of which are the ordinary hardware of circulating-fluid temperature control. Means-plus-function claiming is generally harder, not easier, to defend against a structural prior-art mapping, because the corresponding structure is by definition what the art discloses.

⚠️ Caveat: the full text of claim 16 was not available in the source (truncated at "A system for cooling computer systems h…"). Everything above is built on the Summary of the Invention's means-plus-function embodiment and the corresponding structures recited in the Detailed Description. Verify claim 16's granted language before relying on this mapping.

---

6. Combination summary matrix

Cross-cutting observation: claims 1–9 were attacked in the petition only as anticipation (except 5–6), and claims 10, 14, 16 were not attacked at all. Every entry in this matrix is therefore a new invalidity theory that the PTAB never adjudicated and that no estoppel reaches.

---

7. Weaknesses, anticipated rebuttals, and where the case could fail

7.1 "Air conditioning unit" is the pressure point

The petition proposed a construction for "air conditioning unit" (visible in the claim-construction table fragment), which means it is disputed. If Valtrus succeeds in construing it to require a unitary packaged unit (compressor, condenser, evaporator in one cabinet — the classic CRAC), then:

• Newton's central chiller 12 / heater 14 / conduit 10 plant may fall outside it, weakening Grounds 1 and 2 and Combination 1.
• Shimizu's outdoor unit (compressor 21 / outdoor HX 23) plus indoor units is a better fit — which is precisely why Combination 2 (Newton + Shimizu) is the more robust structure than any single-reference attack.
  Recommendation: plead claim 1 both ways — Newton alone, and Newton + Shimizu — so the case survives either construction.

7.2 Baer's § 102(a) / § 102(b) defect and swear-behind exposure

Carried forward from the earlier Prior Art section: the petition's grounds table labels Baer § 102(b), while its own text argues § 102(a). Baer published 2001-11-22, inside the one-year window before the 2002-08-02 priority date. § 102(a) art is defeated by a pre-AIA § 1.131 affidavit of prior invention. Since the '287 is a divisional of an application filed exactly on the priority date, and the inventors were HP Labs researchers with an internal disclosure-docketing pipeline, a swear-behind attempt is a realistic risk.
Mitigation: use Baer only as ancillary art in a § 103 combination (Combinations 1, 4, 6) rather than as the sole anticipating reference, and buttress the computer-room context with Ex. 1004 (DataCool) and the specification's own admission of DATACOOL — both of which are admissions that cannot be sworn behind.

7.3 Claims 5–6 ("sum") remain the soft spot

Discussed at § 5, Combination 3. Do not lead with them.

7.4 ⚠️ Pre-AIA § 103(c) — the commonly-owned-art trap

This is the single most likely structural defect in any combination that reaches for the incorporated applications or for Ex. 1006 ("Patel," U.S. Pub. 2003/0067745 A1).

Pre-AIA § 103(c)(1) provides that subject matter that qualifies as prior art only under § 102(e), (f), or (g) shall not preclude patentability under § 103 where, at the time the invention was made, the subject matter and the claimed invention were commonly owned or subject to an obligation of assignment to the same person. The three applications the '287 incorporates by reference — Ser. Nos. 09/970,707 (2001-10-05), 10/157,892 (2002-05-31), 10/122,010 (2002-04-24) — are all HP/HPE applications, commonly owned with the '287. Two of them filed in 2002 could not have published before the 2002-08-02 priority date, so they would qualify only under § 102(e), and § 103(c) disqualifies them from an obviousness combination. The same analysis likely applies to Ex. 1006, a "Patel" publication — which may explain why the petition used that exhibit for something other than a combination ground.
Two consequences:

1. A defendant cannot build a § 103 combination out of the incorporated-by-reference applications. The prior art section's caveat to that effect is correct and worth repeating.
2. Those applications remain usable as (i) evidence of the state of the art and what a POSITA knew, and (ii) — because the '287 specification expressly relies on them for enablement of the "anticipated sensing" and load-balancing features — written-description / enablement ammunition under § 112, ¶ 1, which is a separate attack worth preserving.
   ⚠️ I have not verified the ownership, filing dates, or publication date of Ex. 1006. Confirm before relying on either the § 103(c) disqualification or its use as art.

7.5 Secondary considerations (Graham factor 4)

On the current record I found no evidence of:

• unexpected results — the specification claims none, and attributes no surprising property to per-unit coolant metering or to temperature summing;
• long-felt but unsolved need — the '287's own Background shows the need was identified and addressed by the prior art (it acknowledges DATACOOL and the incorporated HP applications);
• industry praise, copying, or licensing due to the merits;
• commercial success with nexus — the '287 was never commercialized by a party in privity with the patentee; the accused products are the defendants' own; and portfolio licensing revenue from an assertion entity carries a well-recognized nexus problem. Enforcement activity and PTAB petitions are not secondary considerations.

Valtrus's most plausible secondary-consideration argument — that a jury/Board should credit the patent because it was valuable enough to be asserted broadly — is legally irrelevant to § 103.

7.6 The presumption of validity and the clear-and-convincing standard

Because IPR2025-00667 was not instituted, no claim was canceled, and in district court the '287 carries the § 282 presumption. Every combination above must be proved by clear and convincing evidence (i4i). That standard is more forgiving to Valtrus than the PTAB's preponderance — so a defendant should expect to need more than the petition's record: live expert testimony, corroborating contemporaneous literature, and, ideally, the testimony of the named inventors or the HP/HPE engineers of record about what was known in 2002.

---

8. Bottom line

Yes — the claims of the '287 patent are vulnerable to § 103, and the vulnerability is broader than anything the PTAB ever saw. The patent's asserted point of novelty — per-unit mass-flow control of cooling fluid in a multi-heat-exchanger system — is squarely, and in some places redundantly, disclosed:

1. Newton (U.S. 3,384,155, 1968) discloses per-unit coolant metering on individual branch lines driven by per-unit sensed zone temperature, and discloses it twice (valve 36; pump 70), together with a fan-speed control loop and an aggregate-demand central-plant control. This single 1968 patent reaches claim 1[f], claim 7, claim 8, claim 9, claim 14, and the pump architecture of claim 10.
2. Shimizu (U.S. 5,317,907, 1994) independently discloses the same per-unit metered-fluid architecture in a vapor-compression system, plus variable compressor output — which cleanly supplies claim 2.
3. Baer (U.S. Pub. 2001/0042616 A1, 2001) supplies the data-center/enclosure framing and the rack heat-density market pressure; it is a Liebert Corporation application, and the '287 specification itself admits the Liebert DATACOOL unit.
4. No combination faces a teaching away, an inoperability objection, or a non-analogous-art objection, and the motivation is supplied by the same market forces the patent's own Background recites.

Highest-value lines, ranked:

1. Claim 14 — upstream per-unit valves, disclosed redundantly by Newton and Shimizu (very high confidence).
2. Claim 10 — per-unit pumps; Newton FIG. 6 literally replaces each unit's valve with a pump, and Newton expressly teaches fan-speed control in that pump-based embodiment (high confidence).
3. Claims 1, 7–9 — Newton alone, or Newton + Shimizu as the fallback against an adverse "air conditioning unit" construction (high confidence).
4. Claims 3–4 — Newton's zone/water temperature comparison and hot/cold switching; Shimizu's preset-temperature comparison and compressor modulation (high confidence).
5. Claims 5–6 — aggregate-demand teaching in Newton and total-load teaching in Shimizu (moderate confidence).
6. Claim 16 — means-plus-function mapping is straightforward once the granted text is confirmed (moderate confidence pending verification).

What will decide it: the construction of "air conditioning unit" and of "individually manipulating a mass flow rate of the cooling fluid supplied to each of the plurality of heat exchanger units." Both are being briefed in the E.D. Tex. claim-construction track (P.R. 4-3 statement 2025-08-11; P.R. 4-5(d) chart 2025-10-01). Notably, per the earlier PTAB section, Valtrus's own P.R. 4-3 appendix cites the defeated IPR petition and Dr. Abraham's declaration as extrinsic evidence — a two-edged sword, because that same declaration contains the claim-by-claim prior-art mapping reproduced above.

---

9. Verification items I could not close

1. The full text of claim 16 (truncated in the source at "A system for cooling computer systems h…"). Claim 16's § 103 mapping is therefore provisional.
2. The full "Grounds" and "Motivation to Combine" sections of the IPR2025-00667 petition. I retrieved the exhibit list, the challenged-claims listing, the claim-by-claim mapping excerpts, and the claims 5–6 obviousness argument — but not the complete grounds section. Confirm whether any combination ground beyond "Newton + POSITA knowledge" was pleaded.
3. The 2025-07-31 discretionary-denial paper — needed to confirm no merits reasoning exists (bearing on estoppel and on the "settled expectations" narrative).
4. Ex. 1013 — the earlier PTAB section cited "Ex. 1013" for the 2001 HPE/Vertiv-predecessor DATACOOL joint development; the exhibit list I retrieved ends at Ex. 1012. Resolve the discrepancy.
5. The use made of Ex. 1005 (Stahl, U.S. 6,556,624), Ex. 1006 (Patel, U.S. Pub. 2003/0067745 A1), Ex. 1009 (Feeney, U.S. 5,467,609), and Ex. 1010 (Arima, U.S. 6,006,528) — not visible in the retrievable petition excerpts.
6. Ownership, filing date, and publication date of Ex. 1006 (Patel, U.S. Pub. 2003/0067745 A1) — determines whether pre-AIA § 103(c) disqualifies it from an obviousness combination.
7. The patent's front-page examiner-cited reference list. Still unretrieved (the Google Patents text omits the citation table). If Newton, Shimizu, or Baer was cited during prosecution, the file history may contain applicant arguments distinguishing them — high-value for claim construction. Route: USPTO PatentCenter / IFW (look for the PTO-892).
8. The office action(s) that issued the '287 claims. The claimed advance was added at the divisional stage relative to parent 10/210,040; identifying what the examiner allowed over — and what the applicant argued — is the fastest way to find the intended claim scope for 1[f].

Citation sources: https://patents.google.com/patent/US6854287/en (authoritative full text) · https://ptacts.uspto.gov/ptacts/public-informations/petitions/1557496/download-documents (IPR2025-00667 exhibit list, challenged-claims listing, claim mapping, motivation) · https://patentimages.storage.googleapis.com/5a/44/75/96ff3858e5be82/US3384155.pdf (Newton, U.S. 3,384,155) · https://ipverse.greyb.com/ptab-web/cases/case-details/IPR2025-00667 (proceeding status)

Cross-reference note (no contradiction): this analysis is consistent with the earlier Litigation, PTAB, and Prior Art sections as to the identity of the three applied references and the grounds table. It adds the reference numbers (U.S. 3,384,155; U.S. 5,317,907), the expert's full name (John P. Abraham, Ph.D.), the four additional exhibit references, and the six § 103 combinations — none of which was pleaded at the PTAB. It flags one discrepancy (Ex. 1013 vs. an exhibit list ending at Ex. 1012) and one legal trap (§ 103(c) commonly-owned art) that the earlier sections did not reach.