Duke Energy Carolinas, LLC v. NTE Carolinas II, LLC

Opinion

PUBLISHED

UNITED STATES COURT OF APPEALS FOR THE FOURTH CIRCUIT

No. 22-2168

DUKE ENERGY CAROLINAS, LLC, Plaintiff - Appellee,

and

DUKE ENERGY CORPORATION; DUKE ENERGY PROGRESS, LLC, Counter-Defendants - Appellees, v.

NTE CAROLINAS II, LLC; NTE CAROLINAS II HOLDINGS, LLC; NTE ENERGY, LLC; NTE SOUTHEAST ELECTRIC COMPANY, LLC; NTE ENERGY SERVICES COMPANY, LLC; CASTILLO INVESTMENT HOLDINGS II, LLC,

Defendants - Appellants.

------------------------------------------ AMERICAN ANTITRUST INSTITUTE, Amicus Supporting Appellants.

CHAMBER OF COMMERCE OF THE UNITED STATES OF AMERICA; NORTH CAROLINA CHAMBER LEGAL INSTITUTE; DR. BENJAMIN ZYCHER; GEOFFREY A. MANNE; PROFESSOR RICHARD A. EPSTEIN; PROFESSOR DONALD J. BOUDREAUX.

Amici Supporting Appellees.

2

Appeal from the United States District Court for the Western District of North Carolina, at Charlotte. Kenneth D. Bell, District Judge (3:19-cv-00515-KDB-DSC)

Argued: May 7, 2024 Decided: August 5, 2024

Before NIEMEYER and THACKER, Circuit Judges, and MOTZ, Senior Circuit Judge.

Vacated and remanded by published opinion. Judge Niemeyer wrote the opinion, in which Judge Thacker and Senior Judge Motz joined.

ARGUED: Derek T. Ho, KELLOGG, HANSEN, TODD, FIGEL & FREDERICK PLLC, Washington, D.C., for Appellants. Morgan L. Ratner, SULLIVAN & CROMWELL LLP, Washington, D.C., for Appellee. ON BRIEF: Matthew J. Wilkins, Caroline A. Schechinger, Jonathan I. Liebman, KELLOGG, HANSEN, TODD, FIGEL & FREDERICK, P.L.L.C., Washington, D.C., for Appellants. Jason D. Evans, TROUTMAN PEPPER HAMILTON SANDERS LLP, Charlotte, North Carolina; Douglas Green, STEPTOE & JOHNSON LLP, Washington, D.C.; Jeffrey B. Wall, Daniel J. Richardson, SULLIVAN & CROMWELL LLP, Washington, D.C., for Appellees. Kathleen W. Bradish, AMERICAN ANTITRUST INSTITUTE, Washington, D.C., for Amicus The American Antitrust Institute. Andrew R. Varcoe, Tyler S. Badgley, UNITED STATES CHAMBER LITIGATION CENTER, Washington, D.C., for Amicus Chamber of Commerce of the United States of America. Michael F. Murray, Mary Walser, PAUL HASTINGS LLP, Washington, D.C., for Amici Chamber of Commerce of the United States of America and North Carolina Chamber Legal Institute. Sean E. Andrussier, WOMBLE BOND DICKINSON (US) LLP, Raleigh, North Carolina, for Amici Dr. Benjamin Zycher, Geoffrey A. Manne, Professor Richard A. Epstein, and Professor Donald J. Boudreaux.

3

NIEMEYER, Circuit Judge:

NTE Carolinas II, LLC (“NTE” 1), a power company based in St. Augustine, Florida, sued Duke Energy Corporation (“Duke” 2), a power company based in Charlotte, North Carolina, alleging that Duke had monopoly power in the wholesale power market in the Carolinas and willfully maintained that power through anticompetitive conduct to exclude NTE from the market, in violation of § 2 of the Sherman Act. See 15 U.S.C. §§ 2, 15. In particular, NTE presented evidence in the district court that Duke devised a plan to ensure that NTE, its only serious competitor, would not have the opportunity to compete for the business of Fayetteville, North Carolina, the only major wholesale customer whose long- term contract with Duke was expiring soon enough to allow NTE to compete for its business.

The district court granted Duke’s motion for summary judgment, in which Duke argued that the conduct that NTE imputed to Duke constituted legitimate competition in seeking to retain Fayetteville’s business and that none of the actions on which NTE relied was unlawful. While the court concluded that there was a question of fact on whether Duke had monopoly power, it also concluded as a matter of law that Duke did not engage in anticompetitive conduct but rather legitimate competition to retain Fayetteville’s business.

NTE will be used as shorthand to refer collectively to NTE Carolinas II, LLC;

1

NTE Carolinas II Holdings, LLC; NTE Energy, LLC; NTE Southeast Electric Co., LLC; NTE Energy Services Co., LLC; and Castillo Investment Holdings II, LLC.

2

Duke will be used as shorthand to refer collectively to Duke Energy Corporation, and its subsidiaries Duke Energy Carolinas, LLC, and Duke Energy Progress, LLC.

4

The record in this case is large, and it contains much evidence related to Duke’s conduct in response to NTE’s competitive efforts. While we recognize that much of Duke’s conduct can be understood to be legitimate competitive conduct, as well explained by very able counsel, we also have found much from which a jury could conclude that Duke’s actions were illegitimate anticompetitive conduct that violated § 2 of the Sherman Act, also as well explained by very able counsel. Because genuine disputes of material fact exist, we vacate the district court’s summary judgment and remand for further proceedings.

We also order that, on remand, the case be assigned to a different judge. In an act of caution, the district judge in this case initially recused himself because of the appearance of one of his former law partners on behalf of Duke. But he was reassigned the case a couple of years later after the “conflict” abated, and he then declined to recuse himself on NTE’s motion, determining that his earlier recusal had not been necessary. We conclude, as most courts have, that once a judge recuses himself from a case, he should remain recused from that case, even though his recusal may not have originally been required.

I

After Duke filed an answer to NTE’s operative antitrust complaint, the parties engaged in extensive discovery, creating a substantial record, which included detailed and complex expert witness reports. Duke then filed a motion for summary judgment based on that record. While the district court concluded that genuine questions of material fact remained on whether Duke had monopoly power, it concluded that NTE failed to show

5

that Duke had engaged in “improper exclusionary conduct harming competition.” Duke Energy Carolinas, LLC v. NTE Carolinas II, LLC, 608 F. Supp. 3d 298, 317 (W.D.N.C. 2022). Identifying five distinct courses of conduct claimed by NTE to have been part of Duke’s anticompetitive scheme, the district court addressed each course independently, found that each was not unlawful by itself, and concluded that “[a]dding up several instances of lawful conduct [could not] total unlawful conduct.” Id. at 319; see also id. at 319–28. The court accordingly granted summary judgment to Duke.

Because one issue raised is whether the record reveals genuine issues of material fact, we find it appropriate to recite the record in some detail.

A

The summary judgment record shows the following:

NTE, as an independent power producer (“IPP”), generates power at power plants, but it does not own transmission lines and therefore cannot, with its own resources, transmit the energy it produces to wholesale customers. Thus, NTE must rely on the transmission networks owned by other energy companies to transmit electricity over power lines to its wholesale customers, typically municipalities. J.A. 4455; FERC, Energy Primer: A Handbook of Energy Market Basics 47 (2020). To facilitate such access, the Federal Energy Regulatory Commission (“FERC”) requires utilities to share their transmission networks with competitors. See generally Promoting Wholesale Competition Through Open Access Non-Discriminatory Transmission Services by Public Utilities, 61 Fed. Reg. 21540 (May 10, 1996).

6

In 2014, NTE began construction of a new combined-cycle natural gas facility in Kings Mountain, North Carolina. See J.A. 4616. To transmit electricity in the Carolinas, however, it needed to use the transmission lines of Duke, a longtime monopolist holding more than 90% of the wholesale power market in the region. J.A. 4481. Duke is a vertically integrated power company, meaning that it owns both power plants and transmission lines and serves both wholesale and retail customers. In accordance with FERC regulations requiring interconnection, Duke and NTE entered into a standard interconnection agreement, with Duke providing NTE access to its transmission network so that NTE could sell power from its Kings Mountain plant. While NTE thus entered Duke’s service area, Duke at first had little concern about its presence as a competitor, inasmuch as NTE’s Kings Mountain plant was a relatively small generator. Duke’s Vice President of Wholesale Power Sales remarked at the time that he “[thought] it [was] very doubtful that the threat [of Duke customers switching to NTE] [was] real.” J.A. 4830.

Duke’s view of NTE changed over the following year, however, as Duke began to realize that NTE was successfully attracting Duke customers. See J.A. 5047 (Duke email stating that NTE winning Winterville was a “total surprise”). In October 2014, Duke executives asked their subordinates to keep them briefed on NTE’s development plans. J.A. 4832–33. In January 2015, Duke’s internal briefing reports showed that “[s]tand-alone combined cycle plants” like NTE’s “offer less expensive energy than Duke Energy system average rates for the foreseeable future, along with lower capacity prices.” J.A. 4881. In short, the combined-cycle plants that NTE was in the business of building were more cost-

7

efficient than Duke’s own plants. See J.A. 4471–72. Duke then recognized that it “[couldn’t] chase the price competition and earn a reasonable return.” J.A. 4888.

When NTE opened its Kings Mountain plant, Duke customers, too, took note of NTE’s new offerings. For example, in October 2015, a representative of a wholesale customer informed Duke that it was signing a Letter of Intent to buy electricity from NTE due to the savings that NTE offered, explaining:

The issue with my towns is cost. We are in an area of the state where wages have stagnated and making ends meet is incredibly difficult for many folks.

Duke’s offer, over the long term, was simply uneconomic.

J.A. 5060. Eventually, Duke lost a total of nine of its customers to NTE selling electricity from its Kings Mountain plant. J.A. 4618. Yet, during that same time, it lost only one customer to a competitor other than NTE. See J.A. 4659.

Despite NTE’s cleaner, more efficient power generators, Duke recognized that it had an advantage by reason of its long-term wholesale power supply contracts with its customers, which spanned 20 years and required several years’ notice of termination. See, e.g., J.A. 2545 (10-year notice period); J.A. 5235 (7-year notice period). These long-term contracts, which were common in the Southeast energy markets, decreased the risk that a Duke wholesale customer would switch to a new entrant provider of electricity like NTE. As a consequence, such contracts limited opportunities for new entrants such as NTE to compete for customers and thus to gain economies of scale. But Duke identified one contract that was set to expire soon enough for Duke to worry about losing a major customer’s business to NTE — the City of Fayetteville, which provided a peak demand load for electricity of approximately 500 MW. See J.A. 2847, 5086.

8

Fayetteville had been a Duke customer for more than 100 years, and, at the time that NTE began its attempt to win Fayetteville’s business, Fayetteville and Duke were operating under a 20-year agreement that they had entered into in 2012 (“2012 Power Supply Agreement”). J.A. 4453, 6719. Under the 2012 Power Supply Agreement, Fayetteville agreed to buy power from Duke until June 30, 2032. But in the Agreement, Fayetteville retained the right to terminate the Agreement as of July 1, 2024, upon notice to Duke given by June 30, 2017 — a deadline that Duke extended to 2019 and later to 2020. J.A. 906– 07, 2845, 3463. Duke not only sold Fayetteville electrical power, but, under a separate contract, it also purchased excess power from Fayetteville that was produced at Fayetteville’s Butler Warner plant. J.A. 2848. That plant was, however, very inefficient, made up of eight gas or oil combustion turbine generating units and one combined-cycle steam turbine unit. See J.A. 3683, 4942. In theory, Duke could sell the excess power generated at the Butler Warner plant to other customers. J.A. 1188; see also Response Br. at 9 (“Duke . . . bought excess power produced by Fayetteville’s power plant, which Duke could potentially sell to other customers”).

In 2016, in light of NTE’s success with its King Mountain plant, Duke’s internal reports observed that “[b]ut for NTE customers,” Duke’s “portfolio [was] stable.” J.A. 5086. Duke’s internal reports additionally projected that Duke’s rates would remain much higher than NTE’s through at least 2025, J.A. 5071, and that Duke’s “[c]ompetitive disadvantage [was] not going away soon,” J.A. 5086. NTE’s competitive advantage prompted Duke to recognize that Fayetteville, which had been providing Duke with $100 million in annual net revenue, was its “[l]argest customer risk.” J.A. 5086. And in the

9

months thereafter, Duke continued to report internally that NTE was its “biggest threat,” J.A. 4700, listing Duke’s relative inefficiency as a major factor for why Duke was, in its words, “not competitive,” J.A. 7273.

NTE did indeed then have plans to build additional power plants in the Carolinas.

But key to its plans for expansion was the rare opportunity — because of the terms of Fayetteville’s agreement with Duke — to compete for Fayetteville’s business. Thus, in 2016, NTE announced its plan to build a second plant, the Reidsville Energy Center, which could serve Fayetteville. That plant would have a capacity of approximately 475 MW and would be a natural gas combined-cycle electric generation plant. J.A. 4466. To bring the Reidsville plant online, however, NTE needed to attract not only wholesale customers who were not already locked into long contracts, but especially a large wholesale customer such as Fayetteville.

Because NTE was an IPP, it also needed to enter into an interconnection agreement with Duke to have access to Duke’s transmission lines for its Reidsville plant. So, in November 2017, again in accordance with FERC regulations, NTE and Duke entered into a standard interconnection agreement for the Reidsville plant (“the Reidsville Interconnection Agreement”). See J.A. 418–504; see also Standardization of Generator Interconnection Agreements and Procedures, Order No. 2003, 104 FERC ¶ 61,103 (2003). Under this pro forma interconnection agreement, NTE agreed to pay Duke $58,917,362 to build the interconnection infrastructure for the Reidsville plant, and once the lines were built, Duke would own those transmission lines and charge NTE to use them. J.A. 465, 491. The contract required NTE to make “security payments” — payments made in

10

advance of incurred costs — for Duke’s construction work on a schedule fixed by the agreement. J.A. 467, 575.

A few months after Duke and NTE entered into the Reidsville Interconnection Agreement, NTE persuaded three of Duke’s wholesale customers to buy power from NTE — McCormick, South Carolina; City of Camden, South Carolina; and Western Carolina University, J.A. 5395, 5989, 6071. These were significant wins for NTE. Camden, for example, had been a Duke customer for more than 70 years. J.A. 5476.

As Duke continued to lose customers to NTE, Duke’s competitive concerns intensified. In 2017, Duke projected its all-in costs to be 30 percent higher than NTE’s costs for its new combined-cycle plants. J.A. 1806. In March 2018, at an all-hands meeting, Duke’s wholesale power segment warned that Duke’s systems were “no longer competitive,” although it observed that the “[p]roblem [was] mitigated by [the] long-term nature of [its] contracts.” J.A. 5388. Duke’s data showed that its average system cost was well above the cost for combined-cycle plants, such as NTE’s. J.A. 5389. In light of these efficiency concerns, Duke internally identified the retention of Fayetteville’s future business as its “biggest upcoming battle.” J.A. 5390.

Later that spring, Duke officials sent internal emails regarding Camden’s decision to move its business to NTE. J.A. 5476. They also discussed a desire to keep Fayetteville from issuing a “Request for Proposal” (“RFP”) — a mechanism that customers use to invite bidders to make proposals in response to specific requests for service — to provide it with options for the supply of wholesale power for the years after it could terminate its agreement with Duke in 2024. J.A. 5476. In those emails, Duke officials recognized that

11

“[o]ther than Fayetteville, [they] [did] not have any other contracts in the Carolinas at risk for several years.” J.A. 5476. The Vice President of Wholesale Power Sales informed other Duke officials that Fayetteville “ha[d] committed to give [Duke] an opportunity to modify [its] contract to meet [Fayetteville’s] needs before they go out for an RFP.” J.A. 5476. A few months later, a Duke report indicated that its “[p]rimary proposal objectives” in its talks with Fayetteville “[were] to avoid RFP and retain [Fayetteville’s] load.” J.A. 7297. To meet these goals, Duke identified an opportunity to make Fayetteville an offer that Duke’s competitors could not match, in part by modifying Duke’s current long-term contract with Fayetteville, the 2012 Power Supply Agreement. It noted, “Formula rate discount prior to 2024 provides value other competitors can’t offer.” J.A. 7297 (emphasis added).

While Duke recognized that it “need[ed] the NTE train to stop,” NTE continued to attract attention with its superior efficiency. J.A. 5733. In December 2018, Stantonsburg’s mayor reported that a “9 percent rate cut will make Christmas a bit brighter for our citizens,” thanks to switching from Duke to NTE. J.A. 5894. That same month, Duke again projected that “the delta” — i.e., the differential — between Duke’s system costs and NTE’s “[was] 25 to 30 percent.” J.A. 4484.

Despite Duke’s concern with NTE’s competition, NTE and Duke continued to operate under the terms of the Reidsville Interconnection Agreement throughout 2018, and by January 2019, NTE had paid Duke $1.6 million in security payments, as required by the schedule set forth in the Agreement. J.A. 575, 2019. In February of that year, however, Duke changed the routine payment practice. It sent NTE an email stating that it was

12

implementing a new payment program, and NTE “should wait and send their payments per the instructions on the invoice, instead of just wiring it in” in accordance with the Agreement’s schedule. J.A. 5891–92. Only days later, Duke officials exchanged emails discussing its hope to “get [Fayetteville] wrapped up and [p]ut it to bed and ruin NTE’s plans” for the Reidsville plant. J.A. 5906. When March 1, 2019, came around — the day that NTE would have owed Duke a security payment under the Reidsville Interconnection Agreement — Duke did not send an invoice. Accordingly, in compliance with Duke’s instructions, NTE did not wire Duke the security payment then due. J.A. 4764–67.

At the same time that Duke failed to send NTE invoices, it again circulated internal reports showing that Duke was unable to compete with NTE on efficiency: “Duke energy rate of ~$60-65/MWh compares unfavorably against IPPs (e.g. NTE) with offers of ~$40- 45/MWh. Duke Energy capacity charges of ~$18/kw-month far exceed NTE’s $8/kw- month.” J.A. 5927. Duke continued to recognize in its internal reporting that, although “[m]ost wholesale contracts continue another 12-15yrs,” “Fayetteville” was the exception. J.A. 5927.

Despite its relative inefficiency, Duke made a highly attractive, multi-faceted offer to Fayetteville, which amounted in the aggregate to a discount of $325 million for Fayetteville and which was unprecedented. J.A. 577, 4586. First, Duke agreed to provide Fayetteville with a $30/kW-year discount on its existing arrangement with Fayetteville from January 2021 to June 2024, which was worth approximately $42 million to Fayetteville. J.A. 4415. It agreed to give Fayetteville this massive discount on the rates set forth in its 2012 Power Supply Agreement with Fayetteville — something no other

13

competitor could offer — but it required as a condition that Fayetteville begin to pay Duke more for its electricity after 2024, indeed more than NTE would have charged Fayetteville. J.A. 577–78, 4414–4416, 6224–27. Duke labeled this two-part discount-now increase-later strategy as “blend-and-extend,” and it later admitted that the strategy would enable Duke to charge customers like Fayetteville “higher prices than offered by the competition” beginning in 2024. J.A. 4663–64. And as the last part of its proposal to Fayetteville, Duke agreed to quadruple the price it paid for the excess power it bought from Fayetteville’s Butler Warner plant — increasing its purchase price from $50/kW-year to $197.82/kW- year. J.A. 4418–19. One Duke employee acknowledged regarding this proposal that Duke would pay more to Fayetteville under this arrangement than it would have to pay at market for the same power. J.A. 4419. Thus, under this multi-faceted proposal, Fayetteville would in the immediate term save money, not because Duke was more efficient, but in part because Duke could offer a large discount on the years during which Fayetteville was already locked into Duke’s services and it could provide Fayetteville with a large profit on Butler Warner purchases.

After receiving Duke’s proposal, Fayetteville considered its options, hiring a consultant group to help it decide whether it should renew its contract with Duke or issue an RFP and potentially buy its electricity elsewhere, such as from NTE. Fayetteville’s consultants compared Duke’s offer to what it projected NTE’s would be. The consultants’ report noted that Duke was able to cut below NTE’s rates by (1) offering a substantial discount on Fayetteville’s current contract and (2) offering to pay far more for energy coming from Fayetteville’s Butler Warner plant. J.A. 6226–27. When listing the benefits

14

of sticking with Duke, the consultants first highlighted that Duke’s offer “provide[d] savings prior to 2024.” J.A. 6233. While NTE would offer a lower price for electricity than would be provided after the potential termination of Fayetteville’s contract with Duke in 2024, NTE’s offer would come with its own risks. The consultants noted disadvantages to doing business with NTE, including an increased potential for “cost volatility” and NTE’s higher credit risk. J.A. 6230.

Subsequently, in May 2019, Fayetteville and Duke signed a Letter of Intent, reflecting Fayetteville’s tentative decision to renew its 2012 Power Supply Agreement with Duke in light of what appeared to be an attractive offer. J.A. 1391–92.

Meanwhile, NTE was originally required to make another security payment to Duke under the Reidsville Interconnection Agreement on May 1, 2019. J.A. 575. But, as with the March 1 security payment, Duke again did not send NTE an invoice, and NTE, in accordance with Duke’s instructions, did not make the payment. J.A. 4764–67.

Later that month, on May 15, 2019, NTE exercised its contractual right under the Reidsville Interconnection Agreement to suspend work on the construction of the transmission lines facility. J.A. 447, 6239. Such suspensions were common in the industry, and NTE’s suspension was explicitly permitted by the Reidsville Interconnection Agreement, which allowed NTE “to suspend at any time” so long as NTE paid Duke “for all reasonable and necessary costs” incurred “prior to the suspension.” J.A. 447. NTE had decided to suspend the Reidsville Interconnection Agreement to create some flexibility in its development timeline. J.A. 1493.

15

Nonetheless, Duke interpreted NTE’s suspension and nonpayment of the security payments as a breach of the Reidsville Interconnection Agreement, and Duke therefore sent NTE a notice of breach by a letter dated May 22, 2019. J.A. 3865–68. In the letter, Duke also demanded payment of the security amounts for which it had not invoiced NTE. Duke’s letter stated that Duke had in fact sent invoices earlier, but Duke later admitted that that statement was false. J.A. 4764. The day Duke sent its letter, a Duke official discussing NTE’s suspension remarked to another, “breach! breach! punt em!” J.A. 6364.

Important to NTE, when an IPP suspends a standard interconnection agreement, such suspension is not supposed to affect its placement in the queue for interconnection with transmission lines. FERC requires energy providers like Duke to publish information about their transmission capacity to industry participants and the general public through the online Open Access Same-Time Information System (“OASIS”). See Real-Time Information Networks & Standards of Conduct; Notice of Proposed Rulemaking, 60 Fed. Reg. 66182, 66188 (proposed Dec. 21, 1995) (OASIS provides “information between customers and providers regarding available products and desired services”); 18 C.F.R. § 37.2. And an IPP’s listing in an OASIS queue informs the public (including potential customers) that the IPP would be capable of transmitting the electricity it produced. Thus, if an IPP’s interconnection project is listed as “canceled” on an OASIS queue, that signals to the public — including customers and investors — that the IPP is unable to transmit its power.

Duke previously had an arrangement with an independent monitor who was charged with the detection and reporting of anticompetitive conduct in the transmission and

16

interconnection process, including misconduct on the OASIS queue. J.A. 4468–69. But Duke terminated that arrangement in March 2019, leaving it with those responsibilities. J.A. 4468–69.

Duke was aware that NTE’s OASIS queue placement mattered to NTE and that it would be advantageous to Duke if NTE’s Reidsville project was not on track to rise to the top of the interconnection queue. Indeed, soon after NTE provided Duke notice of its suspension, a Duke official asked another, “[D]oes this mean you get to kick NTE Reidsville out of the queue?” J.A. 6246.

During the summer of 2019, Duke and NTE disputed the consequences of NTE’s suspension under the terms of the Reidsville Interconnection Agreement, and in June 2019, Duke sent NTE a formal notice of default. J.A. 3870. NTE disputed that it had breached the agreement or defaulted in any manner, but it offered to pay Duke for the costs that it had incurred on the Reidsville project and for the reasonable and necessary costs caused by the suspension. J.A. 3874–75. Duke then sent a second notice of default and now demanded security payments, stating that it had earlier sent NTE invoices. J.A. 3878–81. Later, however, Duke again admitted that such statement was false. J.A. 4766.

During the same period when Duke was attempting to end its arrangement with NTE, it also was ironing out its proposal to retain Fayetteville’s business. And in a “White Paper” directed to Duke’s Transaction & Risk Committee, the CEO, and the Board of Directors to justify the new Fayetteville proposal, Duke officials reported that “[i]f the environment remains as competitive as it is today . . . when [Duke’s wholesale] customers are ready to negotiate extensions (likely 5 years before expiration), then [Duke] will also

17

have to offer them alternative, more competitive solutions to retain the business.” J.A. 2847. The White Paper reported further that the “structure” of its multi-part offer to Fayetteville — including both the retroactive discount and Duke’s increased purchase price for power from the Butler Warner plant — would allow it “to retain [Fayetteville’s] significant load as well as retain approximately 60% of [Fayetteville’s] contribution towards fixed costs starting in 2024 and beyond.” J.A. 2847. Also in the White Paper, the Duke officials disclosed a plan to shift the cost of the discount it had offered Fayetteville back to its wholesale customers and to its retail customers in years to come. J.A. 2851, 6614. Under the “best case” scenario, officials explained, Duke would cross-subsidize its lower rate starting in June 2024 by raising rates on these other customers, as Duke expected to lose $100 million on the Fayetteville deal. J.A. 2851. That “best case” scenario did not account for the $42 million decrease in total revenues caused by the retroactive discount on the charged rate from January 2021 to June 2024. J.A. 2851; see also J.A. 7003.

On September 6, 2019, Duke unilaterally terminated the Reidsville Interconnection Agreement with NTE. J.A. 3894–97. While the terms of that agreement required Duke to notify FERC prior to any termination, Duke did not do so. J.A. 435. Later that same month, Duke also listed the Reidsville project as “canceled” in its OASIS queue, effectively “punt[ing]” NTE to the end of the line. J.A. 714, 6364. Duke did this at a time when NTE’s lenders were “crystal clear that the[ir] ability to finance the Reidsville project ha[d] everything to do with ensuring that the [Interconnection Agreement] [was] intact.” J.A. 4529.

18

Then, within days, Duke’s Board of Directors passed a resolution approving the new Fayetteville arrangement, considering the retroactive discounts to the 2012 Power Supply Agreement and increased payment on the Butler Warner agreement “collectively.” J.A. 7070. And accordingly, in November 2019, while NTE’s Reidsville project was still listed as “canceled” in Duke’s OASIS queue, Duke and Fayetteville formally executed a new agreement (the “2019 Power Supply Agreement”). J.A. 911. Fayetteville signed the agreement without issuing an RFP to consider other offers, including NTE’s. After execution of the 2019 Power Supply Agreement, Duke submitted it to FERC. J.A. 3518– 27, 3853. But it only submitted that agreement alone, which provided for the discount it was offering Fayetteville as a wholesale customer of electricity, and not its agreement to buy power back from Fayetteville’s Butler Warner plant at a quadrupled rate. J.A. 4414, 4295–96. Duke determined internally that it “[did] not need to file the Butler Warner [power supply agreement] extension at FERC,” and therefore concluded that that portion of its agreement with Fayetteville would not pose a “[f]ederal [r]egulatory [r]isk.” J.A. 6727.

NTE did not seek to intervene in the FERC proceeding initiated by Duke’s filing of the 2019 Power Supply Agreement. See J.A. 2359; see also 18 C.F.R. § 385.206 (permitting “[a]ny person” to “file a complaint seeking [FERC] action against any other person alleged to be in contravention” of a law administered by the agency); 18 C.F.R. § 385.214 (providing a mechanism for a person to intervene as a party in a FERC action); 18 C.F.R. § 385.211 (allowing “[a]ny person” to object to a rate filing with FERC). And

19

on January 22, 2020, FERC approved Duke’s new selling rates, and the 2019 Power Supply Agreement took effect shortly thereafter. J.A. 3480–82.

Without the hope of competing for Fayetteville’s business to justify its Reidsville plant, NTE’s expansion effort lost force. In November 2019, one of NTE’s business partners informed NTE that “Reidsville sounds like a good project . . . but we can’t make any commitment until the interconnection issues are resolved.” J.A. 843. Even so, NTE sought to move forward with the project, and it applied to the North Carolina Utilities Commission for a permit needed to construct the Reidsville plant. See J.A. 7162. Such applications rarely attracted outside involvement and were routinely granted. J.A. 4624. Duke, however, petitioned to intervene before the State Commission to assert that NTE had breached the Reidsville Interconnection Agreement and to suggest that NTE would fail to meet its construction goals “apparently due to a lack of financing and insufficient wholesale customers to justify the need for [Reidsville].” J.A. 516–17, 4293. At that time, however, Duke knew that NTE had won the business of some of Duke’s own customers, such as the City of Camden. J.A. 5476.

In November 2019, NTE petitioned FERC, pursuant to 18 C.F.R. § 385.207(a)(2), for review of Duke’s unilateral termination of the Reidsville Interconnection Agreement, and in May 2020, FERC issued an opinion concluding that Duke’s termination was unlawful. J.A. 562–64. The parties, however, disagree as to the effect of FERC’s order. 3

3

At the parties’ request, FERC “d[id] not address the merits of any breach of contract claim concerning the Reidsville [Interconnection Agreement],” but it granted NTE’s petition “in part, by confirming [FERC’s] exclusive jurisdiction over the

20

When FERC issued its decision, Duke still had the Reidsville project listed as “canceled” in its OASIS queue. Having already secured Fayetteville’s business, Duke corrected the OASIS queue and NTE’s placement on it after FERC’s decision issued. See J.A. 1634. Duke then also sent an invoice to NTE accounting for the actual costs of NTE’s suspension. 4 J.A. 3913–18.

According to NTE, Duke’s actions “destroyed” “the value of the Reidsville project,”

leaving Duke’s customers with no choice but to pay Duke’s higher rates. J.A. 4604, 4754. This litigation followed.

B

The day that Duke unilaterally terminated the Reidsville Interconnection Agreement — September 6, 2019 — it also sued NTE in North Carolina state court for breach of the agreement. J.A. 348. NTE removed the action to federal court and filed a counterclaim

termination of conforming [interconnection agreements], clarifying a transmission provider’s responsibility to file a notice of termination with [FERC] when terminating a conforming [interconnection agreement] over an interconnection customer’s objection, and providing guidance on [electric quarterly reports] and OASIS postings.” J.A. 562. NTE says that “FERC agreed with NTE . . . that Duke’s unilateral termination of NTE’s agreement was unlawful,” Opening Br. at 19, whereas Duke counters that FERC “concluded that its approval was required to terminate, but made clear that it was not addressing the merits of the parties’ contract dispute,” suggesting that FERC “granted NTE’s petition (with Duke’s assent) [merely] in order ‘to remove uncertainty regarding the termination provisions in the’ Interconnection Agreement,” Response Br. at 12–13.

4

The district court stated that “these allegedly due payments were ultimately invoiced to NTE in the Summer of 2019.” Duke Energy Carolinas, 608 F. Supp. 3d at 312 (citing J.A. 3863–68). Our de novo review of the record, however, shows that those summer 2019 invoices were for security payments, and Duke had sent no invoices for actual costs until July 2020. J.A. 3913–18.

21

alleging that Duke had monopolized or attempted to monopolize the Carolinas wholesale energy market, in violation of § 2 of the Sherman Act, 15 U.S.C. §§ 2, 15. J.A. 347–416. NTE also alleged that Duke had violated North Carolina’s Unfair and Deceptive Trade Practices Act, N.C. Gen. Stat. § 75-1.1(a); had breached the Reidsville Interconnection Agreement; and was liable for common law unfair competition. J.A. 408–11 The case was assigned to U.S. District Judge Kenneth Bell. When, however, one of his former partners at the Virginia law firm of McGuire Woods LLP appeared as counsel for Duke, Judge Bell recused himself pursuant to a prophylactic policy he had adopted after becoming a judge for cases in which lawyers from his former firm appeared. Consequently, the court clerk reassigned the case to another judge, noting on the docket that “conflict” was the ground for doing so. The other judge presided over the case well into discovery until he also developed a conflict. Then, roughly two years after the case had been assigned to Judge Bell, it was reassigned to him, who had in the interim abandoned his prophylactic recusal policy. NTE filed a motion to disqualify Judge Bell under the principle that “once recused, a judge cannot resume authority over a case, not even to rescind an erroneously entered recusal order or because the conflict originally requiring the recusal has resolved.” J.A. 248. Judge Bell denied NTE’s motion, concluding that his prior recusal “did not reflect a considered ‘recusal.’” J.A. 343. He thus remained on the case.

Following discovery, Duke filed a motion for summary judgment. Duke argued that NTE’s antitrust claims should be dismissed because NTE had failed to present evidence sufficient to show both that Duke held monopoly power in a relevant market and that Duke

22

had engaged in illegal exclusionary conduct. NTE challenged Duke’s arguments based on the sufficiency of evidence it produced to prove its claims. It also argued that disputes of material facts precluded summary judgment, as a reasonable jury could find that Duke had monopoly power and that it had engaged in exclusionary conduct by taking actions that would be irrational but for their tendency to harm NTE. NTE argued that it could demonstrate, among other things, that Duke lost millions of dollars to stop NTE from competing for Fayetteville’s business; that it failed to inform FERC of the full extent of the discount it had offered Fayetteville; that it had unlawfully terminated the Reidsville Interconnection Agreement; that it had falsely published that the Reidsville project was “canceled” in its OASIS queue; and that it made false statements about NTE to the North Carolina Utilities Commission.

In an opinion dated June 24, 2022, the district court granted summary judgment to Duke on NTE’s antitrust and unfair competition counterclaims. See Duke Energy Carolinas, 608 F. Supp. 3d 298. While the court found a triable issue as to whether Duke had or was likely to achieve monopoly power, it concluded that NTE had not demonstrated that Duke had engaged in anticompetitive or exclusionary conduct. In reaching that conclusion, the court divided NTE’s allegations into discrete challenges to assess whether each individually amounted to an antitrust violation, addressing particularly: (1) Duke’s termination of the Reidsville Interconnection Agreement as a “refusal to deal” or “denial of an essential facility,” (2) Duke’s renewal offer to Fayetteville as “predatory pricing” or “fraud on FERC,” (3) Duke’s filing of its breach-of-contract lawsuit against NTE as “sham” litigation, (4) Duke’s erroneous OASIS posting as “defamation,” and (5) Duke’s

23

intervention before the North Carolina Utility Commission as an additional exclusionary act. Id. at 318. The court applied separate tests as relevant to each subject, found each of them lawful, and declined to consider the acts taken as a whole. See id. at 319. The court accordingly held that NTE’s Sherman Act claims failed as a matter of law because “[a]dding up several instances of lawful conduct cannot total unlawful conduct” — “[i]n simple mathematical terms, 0 + 0 = 0.” Id. The court also granted summary judgment to Duke on NTE’s state law claims of unfair competition. Id. at 328–32.

As for the parties’ breach of contract claims against each other, the district court denied summary judgment. 608 F. Supp. 3d at 332–36. The parties, however, then settled those claims.

From the district court’s orders dated December 8, 2021 and June 24, 2022, NTE filed this appeal, challenging the denial of its motion to recuse Judge Bell and the grant of Duke’s motion for summary judgment on NTE’s Sherman Act claim. We address first NTE’s antitrust claim.

II

The summary judgment standard, which is on duty in this case, allows a case to be resolved before and without a trial when there is no genuine dispute of material fact and the moving party is entitled to judgment as a matter of law. Fed. R. Civ. P. 56(a). The court’s role in ruling on such a motion is not to assess the truth of any fact alleged or to weigh facts, as would a jury in finding facts, but only to determine whether facts are disputed and whether the disputed facts are material. See Tekmen v. Reliance Standard

24

Life Ins. Co., 55 F.4th 951, 959 (4th Cir. 2022); Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 249 (1986). In reviewing a summary judgment, we apply the same standard that the district court was required to apply and review the judgment de novo. W.C. English, Inc. v. Rummel, Klepper & Kahl, LLP, 934 F.3d 398, 402–03 (4th Cir. 2019).

In this case, the district court entered summary judgment in favor of Duke on NTE’s antitrust claim, thus requiring us to determine whether material facts relevant and necessary to the judgment were disputed and, if not, whether the undisputed facts, taken in the light most favorable to NTE, entitled Duke to judgment on NTE’s § 2 claim.

Section 2 of the Sherman Act provides in relevant part:

Every person who shall monopolize, . . . or combine or conspire with any other person or persons, to monopolize any part of the trade or commerce among the several States, or with foreign nations, shall be deemed guilty of a felony . . . .

15 U.S.C. § 2. And a plaintiff may bring a civil action when “injured in his business or property by reason of anything forbidden in [§ 2].” Id. § 15(a).

The Supreme Court has held that the purpose of this law is not to protect competitors, but rather to safeguard the competitive process itself, ultimately for the benefit of consumers. See Spectrum Sports, Inc. v. McQuillan, 506 U.S. 447, 458–59 (1993). To that end, it has required that a plaintiff, to be successful on a § 2 claim, must satisfy two essential elements: (1) that the defendant “possess[ed] . . . monopoly power in the relevant market,” United States v. Grinnell Corp., 384 U.S. 563, 570 (1966), and (2) that the defendant willfully acquired or maintained that power through anticompetitive conduct, as

25

opposed to gaining its monopoly status “as a consequence of a superior product, business acumen, or historic accident,” id. at 571.

The first element is not at issue in this appeal. Duke does not challenge the district court’s conclusion that a reasonable jury could find that Duke has or is likely to achieve monopoly power in the relevant market, given Duke’s “durably high market share,” which stands “at or approaching 90%.” Duke Energy Carolinas, 608 F. Supp. 3d at 315–17. But the second element is at issue — whether Duke maintained its power through anticompetitive conduct, i.e., conduct intended to “exclude rivals on some basis other than efficiency.” Aspen Skiing Co. v. Aspen Highlands Skiing Corp., 472 U.S. 585, 605 (1985) (quoting R. Bork, The Antitrust Paradox 138 (1978)).

A monopolist does not violate § 2 by offering a superior product, service, or lower prices, as such conduct is procompetitive and thus increases consumer welfare. Similarly, a monopolist does not violate § 2 even if it attracts customers by a subpar or overly expensive product, as “business acumen” or “historic accident” could explain such fortune. Grinnell Corp., 384 U.S. at 571. Rather, a monopolist violates § 2 when it “use[s] [its] monopoly power ‘to foreclose competition, to gain a competitive advantage, or to destroy a competitor.’” Eastman Kodak Co. v. Image Tech. Servs., Inc., 504 U.S. 451, 482–83 (1992) (quoting United States v. Griffith, 334 U.S. 100, 107 (1948)).

To begin, we address the parties’ disagreement over the proper analysis of Duke’s conduct. NTE alleges that Duke engaged in several, simultaneous courses of conduct that combined to thwart NTE from bringing a more efficient powerplant online and ultimately from competing with Duke in the Carolinas wholesale power market. It argues that the

26

district court erroneously “compartmentalized” the various aspects of Duke’s anticompetitive conduct and asked whether each one, independently, was unlawful. Opening Br. at 4, 25, 27. NTE maintains that, in approaching the record in this manner, the court failed to apply the correct legal standard, which required it to take account of all the conduct holistically and determine its effect on potential competition in the relevant market. It observes that in compartmentalizing Duke’s conduct for analysis, the district court included no discussion of the alleged anticompetitive consequences of Duke’s campaign as a whole — namely, reduced consumer choice, higher prices in the long term, and market foreclosure. Under the correct approach, it claims, the facts presented show that consumers were denied the choice of purchasing wholesale power from someone other than Duke.

Defending the district court’s approach, Duke argues that we must reject NTE’s holistic approach because the Supreme Court has set forth specific tests for various kinds of conduct, such as refusals to deal and predatory pricing, which it argues are involved here, and that NTE flunks each test. Duke maintains that “NTE cannot string together a series of acts — all lawful in themselves under the relevant tests — and claim that the whole is more than the sum of the parts.” Response Br. at 46–47.

In the context of the allegations in this case, we agree with NTE. It is foundational that alleged anticompetitive conduct must be considered as a whole. Section 2 focuses on anticompetitive conduct, not on court-made subcategories of that conduct. To be sure, when anticompetitive conduct is alleged to be typical predatory pricing, refusing to deal, price fixing, or dividing markets, as but examples, the case law has developed tests for

27

analyzing such claims. See Novell, Inc. v. Microsoft Corp., 731 F.3d 1064, 1072 (10th Cir. 2013). In cases where the alleged conduct falls within such well-defined categories, the method relied on by the district court — that 0 + 0 = 0 — is a proper approach. See, e.g., Pac. Bell Tel. Co. v. linkLine Commc’ns, Inc., 555 U.S. 438, 449 (2009) (observing that the presented “price-squeeze claim” focused on “retail prices — where there [was] no predatory pricing — and the terms of dealing — where there [was] no duty to deal” and evaluating the plaintiffs’ claim under those two relevant tests).

But anticompetitive conduct comes in many different forms that cannot always be categorized. See Conwood Co. v. U.S. Tobacco Co., 290 F.3d 768, 783–84 (6th Cir. 2002). Thus, when a court is faced with allegations of a complex or atypical exclusionary campaign, the individual components of which do not fit neatly within pre-established categories, its application of such specific conduct tests would prove too rigid. This is because “the means of illicit exclusion, like the means of legitimate competition, are myriad.” Verizon Commc’ns Inc. v. L. Offs. of Curtis V. Trinko, LLP, 540 U.S. 398, 414 (2004) (cleaned up); see also Viamedia, Inc. v. Comcast Corp., 951 F.3d 429, 453 (7th Cir. 2020) (“The fact that categories of conduct here [refusals to deal and tying] are conceptually related and may overlap should not cause confusion if we stay focused on the underlying inquiry: the conduct ‘must harm the competitive process and thereby harm consumers’” (quoting United States v. Microsoft Corp., 253 F.3d 34, 58 (D.C. Cir. 2001) (en banc))).

28

Thus, when a plaintiff alleges that a scheme or course of conduct was anticompetitive, the scheme or conduct must be considered as alleged, not in manufactured subcategories. As Justice Holmes explained,

The constituent elements . . . are enough to give to the scheme a body and, for all that we can say, to accomplish it. Moreover, whatever we may think of them separately, when we take them up as distinct charges, they are alleged sufficiently as elements of the scheme. It is suggested that the several acts charged are lawful, and that intent can make no difference. But they are bound together as the parts of a single plan. The plan may make the parts unlawful.

Swift & Co. v. United States, 196 U.S. 375, 396 (1905). While that case was before the Court under § 1 of the Sherman Act, the principle also applies in the context of § 2. As the Supreme Court expressed in Continental Ore Co. v. Union Carbide & Carbon Corp., 370 U.S. 690 (1962), it is a misapplication of antitrust doctrine for a court to treat a plaintiff’s allegations of anticompetitive conduct “as if they were five completely separate and unrelated lawsuits,” effectively “tightly compartmentalizing the various factual components and wiping the slate clean after scrutiny of each.” Id. at 698–99. Just as the “character and effect of a conspiracy are not to be judged by dismembering it and viewing its separate parts, but only by looking at it as a whole,” id. at 699 (cleaned up), so too must a firm’s exclusionary efforts be considered in their totality, see Grinnell Corp., 384 U.S. at 576 (affirming a judgment against a monopolist whose exclusionary campaign included anticompetitive restrictive agreements, pricing practices, and acquisitions). Indeed, a leading antitrust treatise likewise promotes the view that, in particular, exclusionary conduct alleged under § 2 must be considered holistically:

29

In a monopolization case conduct must always be analyzed “as a whole.” A monopolist bent on preserving its dominant position is likely to engage in repeated and varied exclusionary practices. Each one viewed in isolation might be viewed as de minimis or an error in judgment, but the pattern gives increased plausibility to the claim.

Phillip E. Areeda & Herbert Hovenkamp, Antitrust Law ¶ 310c7 (4th and 5th eds. 2024).

Of course, we recognize that “care must be taken lest . . . illegality be inferred from procompetitive conduct.” Areeda & Hovenkamp, supra, ¶ 310c7. The easier cases are those in which individual practices are each independently unlawful, and so they naturally remain unlawful when considered together. The more challenging cases, however, are those in which the question is “whether two or more practices, while lawful individually, can be aggregated into a series or pattern capable of sustaining a Sherman Act § 2 offense.” Id. While such cases may be uncommon and challenging, they are not categorical impossibilities, for “aggregation is appropriate” when individual acts are all “part of the same scheme to perpetuate dominance or drive the plaintiff from the market.” Id. Thus, while courts must not dismember the individual acts of an exclusionary campaign when those acts are interconnected, they also must take care not to aggregate acts that are procompetitive to produce only a semblance of an exclusionary effect when considered together.

With these principles in hand, we turn to NTE’s claim that it presented sufficient evidence to show that Duke engaged in anticompetitive conduct in the maintenance of its monopoly power in the relevant market based on the combined effect of two main components — Duke’s interference with NTE’s effort to obtain Fayetteville’s business and Duke’s disruption of NTE’s interconnection efforts. While we discuss these components

30

separately because of the complex factual allegations related to each, we recognize that NTE claims them to be part of a singular, coordinated anticompetitive effort. And, ultimately, therefore, we conclude that they must be taken as alleged, considered as part of a single campaign to foreclose competition in the Carolinas wholesale power market.

A

NTE argues that Duke’s conduct in connection with its offer to Fayetteville to supply wholesale power to the municipality after 2024 was irrational and anticompetitive, designed only to exclude NTE from competition.

When NTE announced its plans to construct the Reidsville plant, Duke identified Fayetteville as its “[l]argest customer risk,” noting that otherwise, given the long terms of its prior supplier agreements, its “portfolio [was] stable,” other than the customers it had already lost to NTE in 2019. J.A. 5086. Duke’s risk with respect to Fayetteville existed because its 2012 Power Supply Agreement with Fayetteville allowed Fayetteville to move its future business away from Duke, beginning in 2024. As alleged by NTE, Duke thus engaged in conduct to shore up Fayetteville and to exclude NTE, not for rational business reasons but to exclude competition.

Duke’s take on these events is that Duke simply engaged in healthy competition. It “competed for Fayetteville by lowering its prices.” Response Br. at 55. It argues accordingly that NTE’s challenge to its Fayetteville offer must be viewed through the strict lens of a predatory pricing theory and that NTE has not shown that it could win under such theory. Lowering prices to retain customers is simply old-fashioned competition, and NTE

31

did not show, according to Duke, that (1) Duke’s pricing was “below an appropriate measure of its . . . costs,” Brooke Group Ltd. v. Brown & Williamson Tobacco Corp., 509 U.S. 209, 222 (1993) and (2) there was a “dangerous probability[] of [Duke] recouping its investment in below-cost prices,” id. at 224, which are the elements of a predatory pricing claim.

Agreeing with Duke, the district court applied this analysis to find that Duke’s pricing under its 2019 Power Supply Agreement with Fayetteville was above Duke’s average variable costs, which, the court determined, was the appropriate measure. Duke Energy Carolinas, 608 F. Supp. 3d at 325–26.

NTE argues that its challenge to Duke’s conduct was not simply based on predatory pricing but on a larger scope of anticompetitive conduct, maintaining more broadly that “the structure of Duke’s offer was exclusionary,” a contention that the district court overlooked altogether. Opening Br. at 50. As NTE explains it, “[t]he key structural feature of blend-and-extend was massive discounts and rebates conditioned on a long-term renewal agreement with Duke even though NTE’s rates in the renewal period — the only period NTE could bid for — were lower than Duke’s.” Id. It also notes that the massive discounts included an agreement by Duke to purchase excess power from Fayetteville’s Butler Warner plant at extraordinarily high prices — above market. Thus, NTE argues that the district court should have considered the overall structure of Duke’s multi-faceted renewal offer, which NTE compares more to exclusive dealing or tying, and that the “[r]igid price- cost tests are . . . inapt.” Id. at 51.

32

Because the predatory pricing approach argued by Duke focuses only on the pricing levels in the 2019 Power Supply Agreement, even though the extent of its massive discounts and NTE’s challenge both reach more broadly, we agree with NTE that the predatory pricing analysis cannot fully account for the more comprehensive conditions of Duke’s blend-and-extend strategy and the Butler Warner offering. Duke’s full offer instead was roughly akin to a “package discount,” such as is described by the Third Circuit:

The anticompetitive feature of package discounting is the strong incentive it gives buyers to take increasing amounts or even all of a product in order to take advantage of a discount aggregated across multiple products. In the anticompetitive case, which we presume is in the minority, the defendant rewards the customer for buying its product B rather than the plaintiff’s B, not because defendant’s B is better or even cheaper. Rather, the customer buys the defendant’s B in order to receive a greater discount on A, which the plaintiff does not produce. In that case the rival can compete in B only by giving the customer a price that compensates it for the foregone A discount.

LePage’s Inc. v. 3M, 324 F.3d 141, 155 (3d Cir. 2003) (quoting Areeda & Hovenkamp, supra, ¶ 794). In the case before us, B would refer to the product of electric power that both Duke and NTE sought to provide to Fayetteville after 2024, when Fayetteville could terminate its 2012 Power Supply Agreement with Duke. A would be the pre-2024 product of electric power that Duke alone could sell to Fayetteville under its 2012 Power Supply Agreement. And, we can say, even more inculpatory of Duke’s conduct, that A also includes Duke’s purchase of excess electric power from Fayetteville’s Butler Warner plant, again which NTE could not offer. Due to its superior efficiency, NTE would have offered Fayetteville a better price than Duke on B, but it was unable to offer A. Duke, meanwhile, rewarded Fayetteville for purchasing its more-expensive power beginning in 2024 by bundling that price with the massive retroactive discount on the 2012 Power Supply

33

Agreement and the very attractive terms for purchasing excess power from the Butler Warner plant.

NTE presented evidence that this packaging structure of Duke’s offer was anticompetitive in at least three respects. First, the blend-and-extend strategy hindered a new entrant’s ability to compete on the basis of efficiency with Duke for Fayetteville’s business after 2024. A chart produced in one of NTE’s expert reports illustrates this foreclosure:

Example of Exclusionary Price (Price/Offers in $/kW-month)

Original Above Competitive Offer Incumbent Year Competitive Pricing Starting in 2024 Exclusionary Offer (e.g., Old Duke (e.g. NTE Offer) (e.g. New Duke Contract) Contract)

2021 9.64 9.64 7.14 2022 9.64 9.64 7.14 2023 9.64 9.64 7.14 2024 9.64 6.65 7.37 2025 9.64 6.65 7.37 2026 9.64 6.65 7.37 2027 9.64 6.65 7.37 2028 9.64 6.65 7.37 2029 9.64 6.65 7.37 2030 9.64 6.65 7.37 2031 9.64 6.65 7.37 2032 9.64 6.65 7.37 Average 9.64 7.40 7.31

J.A. 4487. As shown in the chart, the structure of Duke’s offer was such that, even if NTE could offer Fayetteville a better price on power after 2024, it was severely disadvantaged because, as a result of the 2012 Power Supply Agreement, only Duke could provide a

34

discount on pre-2024 prices — discounts made not for the purpose of providing a superior product, but for the purpose of cutting out a more efficient competitor for future years.

Second, NTE presented evidence that Duke designed this strategy with the intent of foreclosing any new entrant from ever competing with it as the incumbent monopolist on the merits. According to Duke, its blend-and-extend strategy enabled Duke to charge customers “higher prices than offered by the competition” once the contract subject to the retroactive discount expired — here, the 2019 Power Supply Agreement. J.A. 4663–64. As Duke’s future prices increased, so too would the attractiveness increase of any retroactive discount Duke later provided under its blend-and-extend strategy. The higher Duke set its prices, the more flexibility it would enjoy to cut those prior prices through a conditional retroactive discount — returning a portion of its monopoly prices to customers without actually competing on efficiency grounds against new entrants. This strategy would permit Duke to perpetually lock out upstart competitors like NTE with well-timed discounts without seriously threatening its bottom line long term. While it is true that Duke’s wholesale rates were subject to FERC approval and that monopoly rents may well fall beyond the scope of what FERC would deem reasonable, see 16 U.S.C. § 824d, a jury could find that Duke’s blend-and-extend strategy was designed to charge consumers up to the limit while impeding market entry by more efficient producers. Looked at in this light, a factfinder could conclude that Duke’s offer did not reflect true price competition, but rather was designed to avoid such competition.

Third, NTE presented evidence that the structure of the offer was anticompetitive in that it was designed expressly to “shift” the cost of the massive discount “back to retail and

35

wholesale customers.” J.A. 4421. This strategy was similar but not identical to “recoupment” under a traditional predatory pricing framework. In a classic predatory- pricing scheme, the monopolist waits to recoup the losses it incurred by pricing a particular product below cost by raising its prices after the monopolist succeeds at excluding its rival from competing on the same product. Duke’s internal documents tell of a plan instead to raise prices on other of Duke’s wholesale and retail customers to make up for the profit it lost on the Fayetteville deal. See J.A. 6725, 6760, 6980, 7013. Duke’s projected best-case scenario “assumed the difference in fixed costs no longer recovered from [Fayetteville] due to average 190 MW monthly billing demand credit (approximately $40 million) shift back to retail and wholesale customers, with a 1.5-year delay in retail recovery.” J.A. 6725. This evidence is material because a discount forsaking some amount of profits by itself is not anticompetitive — lower prices, of course, enhance consumer welfare — but cross- subsidization can produce anticompetitive consequences, as some customers make up for the discount by paying higher prices. Duke’s own documents, paired with NTE’s experts’ discussions of their anticompetitive effects, leave open a genuine factual dispute as to whether the structure of Duke’s Fayetteville offer was designed to cut out a more efficient competitor at consumers’ expense. See J.A. 4422, 4488–89. And that factual dispute precludes summary judgment.

Duke argues that NTE’s challenge to the structure of Duke’s offer to renegotiate its existing contract with Fayetteville was “forfeited,” as NTE did not present the argument to the district court. Response Br. at 36. It also argues that NTE’s challenge is, in any event,

36

“wrong,” because Brooke Group’s two-prong test must be applied to NTE’s exclusionary- pricing argument. Id. We find both arguments unpersuasive.

First, NTE did make the argument below, at least in substance. Its expert gave his opinion that the structure of Duke’s offer was exclusionary — that Duke’s blend-and- extend strategy was an “exclusionary retroactive discount,” which had two features, a “retroactive discount and a form of deferred rebate pricing.” J.A. 4486. He explained that, “when a monopolist charges above a competitive price, that leaves a margin above competitive costs[, and] [if] the contract also has a substantial prior notice period, that allows the incumbent to offer a discount on the old contract and thereby exclude more efficient competitors.” J.A. 4486. Furthermore, in its summary judgment briefing below, NTE argued that the conditionality of Duke’s offer was exclusionary — “a ‘retroactive discount’ or rebate on an existing contract can be predatory when offered by a firm with market power.” J.A. 4383. And Duke recognized that NTE had argued that its “retroactive discount” was exclusionary in its summary judgment reply brief but argued that “[a]ny such claim must satisfy Brooke Group.” J.A. 7322. The record thus shows that NTE did not forfeit its challenge to the structure of Duke’s Fayetteville bid.

Second, Duke argues again on appeal that NTE’s challenge to its retroactive discount is still precluded by Brooke Group’s holding addressing predatory pricing. Because Brooke Group does not provide a one-size-fits-all analytic framework for assessing exclusionary pricing allegations, we reject Duke’s argument that it provided the only applicable analysis. This is shown, for instance, by the Third Circuit’s decision in LePage’s Inc. v. 3M. The plaintiff in that case, a competitor in the transparent tape market,

37

alleged that 3M had used a range of exclusionary tactics to maintain its monopoly power, including targeted discounts, exclusive dealing arrangements, discriminatory rebates, and promotional allowances. See LePage’s, 324 F.3d at 144–45. Recognizing that the “relevant inquiry is the anticompetitive effect of 3M’s exclusionary practices considered together,” id. at 162, the court rejected 3M’s argument that each aspect of its conduct should be reviewed individually, and it affirmed a § 2 verdict against 3M, observing that the “jury had before it evidence of the full panoply of 3M’s exclusionary conduct, including both the exclusive dealing arrangements and the bundled rebates,” id. at 154. Rather than strictly applying Brooke Group, the Third Circuit thus considered 3M’s interrelated strategies, assessing that 3M’s foreclosure of the market was “caused by exclusive dealing practices [and] was magnified by [its] discount practices.” Id. at 159. It remains true that “[n]othing in any of the Supreme Court’s opinions in the decade[s] since the Brooke Group decision suggest[s] that the opinion overturned decades of Supreme Court precedent that evaluated a monopolist’s liability under § 2 by examining its exclusionary, i.e., predatory, conduct.” Id. at 152; accord ZF Meritor, LLC v. Eaton Corp., 696 F.3d 254, 277 (3d Cir. 2012) (holding that “the rule of reason is the proper framework” for assessing whether conditional rebates on long-term contracts are anticompetitive because “price-cost test cases are inapposite” when “price itself was not the clearly predominant mechanism of exclusion”).

Because we conclude that disputed facts persist regarding whether the structure of Duke’s offer was exclusionary, we need not assess whether the price level of the 2019

38

Power Supply Agreement between Duke and Fayetteville, standing alone, amounted to a violation of § 2 under a strict predatory pricing theory of liability.

But even if we were to focus on a strict predatory pricing theory, a factual dispute would remain as to whether Duke’s pricing was indeed predatory. Duke argues that it “stood to make $60 million profits from its Fayetteville bid, which was plainly the right business decision.” Response Br. at 20. But NTE’s expert calculated that Duke’s offer fell below its average system cost, which in this case converges with its marginal cost. J.A. 4488–89. This could be an appropriate measure in markets with extremely high fixed costs and very low variable costs, as is characteristic of the wholesale power market. 5 See Areeda & Hovenkamp, supra, ¶ 786 (“[A] common characteristic of public utilities is extremely high fixed costs accompanied by very low variable costs. As a result, the average variable cost test . . . may give the public utility defendant too much leeway”). And we further note that Duke’s amici, in arguing that we should not adopt average total cost as the appropriate measure for gauging predation here, fail to account both for the high fixed costs associated with the wholesale electricity market and the fact that the price cut in this case resulted in the exclusion of a more efficient rival. Economics Professors Amici Br. at 13 (arguing that “[i]t would be inconsistent with a consumer-welfare standard to contend that the seller (including an electrical utility) should effectively be forced to stand

5

A fixed cost does not vary with output levels; a variable cost does. Average variable costs are the sum of variable costs divided by output. Average total costs are the sum of fixed and variable costs divided by output. See Areeda & Hovenkamp, supra, ¶¶ 740–41.

39

down from competitor pressure by keeping prices no lower than the average system cost” and reasoning that “[n]aturally, price competition can result in exclusion of less efficient rivals,” which does not harm consumers (emphasis added)). The parties’ experts dispute whether the $60 million that Duke earned from Fayetteville in its renewal contract is properly considered “profit” or is rather a partial recovery of its marginal costs. 6 And that price-cost allocation dispute should be given to the factfinder to resolve. See Greenville Publ’g Co. v. Daily Reflector, Inc., 496 F.2d 391, 397–98 (4th Cir. 1974).

Duke also argues, again relying on the application of a strict predatory pricing analysis, that the “filed-rate” doctrine bars NTE’s challenge to Duke’s Fayetteville renewal offer, because FERC approved the 2019 Power Supply Agreement, deeming the rate “just and reasonable.” 16 U.S.C. § 824d(a). Furthermore, Duke argues, this is especially so here because NTE had the opportunity to intervene in FERC’s review of the 2019 Power Supply Agreement but did not do so. See 18 C.F.R. §§ 385.206, 385.211, 385.214. To be sure, the filed-rate doctrine forecloses a private suit for damages based on a claim that a “rate submitted to, and approved by, [a federal regulator] was the product of an antitrust violation.” Square D Co. v. Niagara Frontier Tariff Bureau, Inc., 476 U.S. 409, 422 (1986) (discussing Keogh v. Chicago & N.W. Ry. Co., 260 U.S. 156 (1922)). Indeed, the idea

6

The district court found that “NTE has acknowledged that Duke’s price at [Fayetteville] contributed $90 million to fixed costs above variable or marginal costs.” Duke Energy Carolinas, 608 F. Supp. 3d at 326. This was a mistake, however. In the deposition of NTE’s expert witness cited by the district court, the expert stated that under the “current contract, not . . . the renewal,” Duke would recover $90 million “towards fixed costs above variable or marginal costs,” and he did not agree that Duke’s revenues under the renewal contract contributed $90 million to fixed costs above variable or marginal costs. J.A. 7426.

40

behind filing rates is that it prevents regulated firms from deviating from their published rates, and thus prevents discriminatory rates. Regulating rates is thus designed to protect customers who pay the filed rates, but not competitors who are not the intended beneficiaries of such a scheme. Recognizing this, the majority of courts of appeals have held that the filed-rate doctrine does not apply to preclude competitor suits. See Cost Mgmt. Servs. v. Washington Nat. Gas Co., 99 F.3d 937, 945–46 (9th Cir. 1996); City of Kirkwood v. Union Elec. Co., 671 F.2d 1173, 1179 (8th Cir. 1982); City of Groton v. Conn. Light & Power Co., 662 F.2d 921, 929 (2d Cir. 1981); Essential Commc’ns Sys., Inc. v. American Tel. & Tel. Co., 610 F.2d 1114, 1121–22 (3d Cir. 1979). Accord Areeda & Hovenkamp, supra, ¶ 247c (“We agree with those decisions refusing to apply Keogh to competitor suits”).

The record in this case demonstrates well why the filed-rate doctrine is ill suited for this competitor suit. For one, the 2019 Power Supply Agreement that Duke filed with FERC informed FERC only of the retroactive discount and the new rate that it agreed to charge Fayetteville upon renewal of the contract. Duke’s filing did not reveal either its cross-subsidization plan or its Butler Warner buy-back agreement, even though its generous pricing on that contract was an integral piece of its Fayetteville offer. Because Duke did not provide FERC with material aspects of its arrangement with Fayetteville, FERC was not invited to scrutinize the structure of Duke’s offer, which is what NTE maintains is the foundation of its exclusionary character. Additionally, FERC also was not asked to consider the exclusionary effects of Duke’s pricing structure; it determined only that Duke’s pricing level as revealed in the 2019 Power Supply Agreement fell within

41

FERC’s zone of reasonableness. The limited scope of FERC’s review of Duke’s offer thus precludes application of the filed-rate doctrine to NTE’s exclusionary pricing allegations.

At bottom, while Duke argues that its Fayetteville renewal offer was procompetitive because it lowered prices, that view is myopic in light of NTE’s claims and supporting evidence about the overall structure of Duke’s offer. Because a factual dispute exists concerning whether the structure and price level of Duke’s offer, taken together, had the effect of foreclosing a more efficient rival from competing, we cannot agree with Duke that, as a matter of law, its conduct was procompetitive.

The record is thus sufficient to support a finding that Duke’s blend-and-extend strategy, coupled with its Butler Warner agreement, independently produced anticompetitive effects. Even so, NTE alleged that Duke’s Fayetteville strategy was but one prong of its alleged campaign to keep NTE from building its Reidsville plant. NTE alleged that there was another prong involving Duke’s additional efforts to ensure that NTE’s potential customers and investors did not view NTE as a viable contender in the Carolinas wholesale power market. We now turn to that alleged prong.

B

The second prong of Duke’s campaign, as NTE has alleged, was Duke’s interference with NTE’s effort to connect its Reidsville plant to Duke’s transmission lines, which included disrupting NTE’s placement in Duke’s OASIS queue and interfering with its application to the North Carolina Utilities Commission. Because this aspect of Duke’s

42

conduct somewhat resembles a refusal to deal, that is the framework through which the parties discuss Duke’s conduct.

Typically, firms have no duty to deal with their competitors, although in “limited circumstances . . . a firm’s unilateral refusal to deal with its rivals can give rise to antitrust liability.” linkLine, 555 U.S at 448. Applying the established principles applicable to refusals to deal, Duke notes that NTE needed to show (1) that both NTE and Duke, as competitors, were engaged in a voluntary course of dealing, and (2) that Duke refused to sell its goods or services to NTE on the same terms as it would to others, citing Trinko, 540 U.S. at 408–09. It argues that NTE failed to make that showing.

While NTE agrees that those conditions are sufficient, it argues that they are not necessary to establishing antitrust liability. The “ultimate test” remains whether Duke refused to deal in order to exclude a rival on a basis other than efficiency. Opening Br. at 49.

The leading case concerning refusal to deal with a competitor is Aspen Skiing, where the Supreme Court upheld a jury verdict against a dominant ski resort that had cut a smaller rival out of a joint venture to sell multiarea ski tickets. 472 U.S. at 590–94. The dominant ski resort failed to persuade the jury that its refusal to continue the joint ticket arrangement “was justified by any normal business purpose.” Id. at 608. Instead, the jury credited the smaller rival’s evidence of predation, which included, among other evidence, “statements made by the officers or agents of the company” and “evidence that the conduct was not related to any apparent efficiency.” Id. at 608 n.39 (cleaned up). In those circumstances, the Court identified the dominant firm’s termination of the prior voluntary course of

43

dealing and refusal to sell at a retail price as sufficient conditions for establishing § 2 liability. The main lesson of Aspen Skiing is that “it is fair to characterize [a monopolist’s] behavior as predatory” “[i]f [it] has been attempting to exclude rivals on some basis other than efficiency.” Id. at 605 (cleaned up).

In a subsequent § 2 case arising between parties in a regulated market, the Supreme Court applied the principles of Aspen Skiing. See Trinko, 540 U.S. at 408–11. The Trinko plaintiff was an AT&T customer who claimed that Verizon delayed fulfilling its rivals’ interconnection requests “as part of an anticompetitive scheme” to disincentivize customers from moving to Verizon’s rivals, “thus impeding the [rivals’] ability to enter and compete” in the local telephone services market. Id. at 404. The Telecommunications Act of 1996 required Verizon to share its network with competitors, and the Trinko plaintiff alleged that Verizon breached that duty and that such breach violated § 2 of the Sherman Act. Id. at 401. After holding that the Telecommunications Act did not preclude the plaintiff’s antitrust claim, the Court considered whether the challenged activity “violate[d] pre-existing antitrust standards,” including those articulated in Aspen Skiing. Id. at 407. But the Court distinguished Trinko from Aspen Skiing on two grounds. First, Aspen Skiing involved “[t]he unilateral termination of a voluntary (and thus presumably profitable) course of dealing suggest[ing] a willingness to forsake short-term profits to achieve an anticompetitive end.” Id. at 409. By contrast, the interconnection service Verizon was required to offer its rivals was not profitable to it. See id. Second, “the [Aspen Skiing] defendant’s unwillingness to renew the ticket even if compensated at retail price revealed a distinctly anticompetitive bent.” Id. By contrast, interconnection to Verizon’s own

44

system was not something that Verizon had offered at retail price, and its “reluctance to interconnect at the cost-based rate of compensation,” as provided for in the Telecommunications Act, revealed “nothing about dreams of monopoly.” Id. The Trinko record contained no evidence that Verizon had abandoned a profitable deal for the purpose of undermining competition, and so the Court concluded that Verizon’s behavior did not fall within “a recognized antitrust claim under [the] Court’s existing refusal-to-deal precedents.” Id. at 410. Accordingly, the Court concluded that the plaintiff’s complaint failed to state a claim under the Sherman Act. Id. at 416.

An important distinction between Aspen Skiing and Trinko is that Trinko — like the case now before us — involved a regulated market. Even so, the Trinko Court did not adopt a rule that unlawful refusals to deal were impossible in regulated markets. Instead, it instructed that “the existence of a regulatory structure designed to deter and remedy anticompetitive harm” was an important “factor” in an antitrust analysis because the existence of such a regime can lessen the need for antitrust enforcement by courts. 540 U.S. at 412.

Trinko’s statement that the presence of regulatory oversight is only an important “factor” in any antitrust analysis confirmed prior Supreme Court precedents that recognized that § 2 liability can arise in regulated markets. See, e.g., New York v. FERC, 535 U.S. 1, 9 n.6 (2002). Indeed, in Otter Tail Power Co. v. United States, where the refusal to deal involved the wholesale power market, the Court rejected Otter Tail’s contention that “by reason of the Federal Power Act it [was] not subject to antitrust regulation with respect to its refusal to deal.” 410 U.S. 366, 372 (1973). The Court

45

affirmed a district court judgment against Otter Tail for unlawfully refusing either to sell or transmit wholesale power to newly established municipal power distribution systems. Id. at 368–71. The municipalities had purchased Otter Tail’s distribution services before building their own competing systems, at which point Otter Tail refused to sell them wholesale power. Id. at 371. Although Otter Tail had no prior course of dealing with the municipality distribution systems, the Supreme Court recognized that Otter Tail’s “refusals to sell at wholesale or to wheel were solely to prevent municipal power systems from eroding its monopolistic position.” Id. at 378. At the time, the Federal Power Commission — FERC’s predecessor — “ha[d] the authority to compel involuntary interconnections,” but “[o]nly if a power company refuse[d] to interconnect voluntarily.” Id. at 373. Indeed, the Commission had compelled interconnection for one of the municipal power systems in Otter Tail, id. at 371, and yet the Court still found that Otter Tail’s refusal to deal was unlawful. Otter Tail thus made clear that “[a]ctivities which come under the jurisdiction of a regulatory agency nevertheless may be subject to scrutiny under the antitrust laws.” Id. at 372.

In this case, if a jury were to resolve all factual disputes in NTE’s favor, it could reach the conclusion that Duke, like the defendant in Aspen Skiing, “[forsook] short-term profits to achieve an anticompetitive end” by unilaterally terminating the Reidsville Interconnection Agreement. Trinko, 540 U.S. at 409. A reasonable jury could find that Duke put NTE in the position of appearing to have breached its agreement with Duke, thereby giving Duke a justification to upset NTE’s placement in Duke’s OASIS queue. NTE’s apparent breach and lack of interconnection ability, in turn, would ensure that

46

Fayetteville would not issue an RFP to NTE, thereby ultimately stalling it from bringing its Reidsville plant online.

In this formulation of facts, FERC’s regulatory oversight would not foreclose antitrust liability. It is true that, as in Trinko, Duke only offered NTE interconnection services because it was compelled to do so by statute, and therefore the parties’ dealing was not voluntary. But the Reidsville Interconnection Agreement was profitable to Duke, a fact Duke does not dispute. Once NTE paid Duke $59 million to build the transmission infrastructure, Duke would own the infrastructure, and it would then provide transmission services at FERC-approved rates. See 18 C.F.R. § 2.22. Thus, in terminating that relationship, Duke forewent a profitable arrangement. Trinko emphasized that foregoing a “presumably profitable” course of dealing was one reason why the refusal to deal in Aspen Skiing, unlike that in Trinko itself, was anticompetitive. 540 U.S. at 409. Duke, however, observes that it “stood to gain ‘short-term profits’ and ‘infrastructure upgrades’ from the Interconnection Agreement with NTE . . . only if NTE actually paid its bills.” Response Br. at 31. Yet, a reasonable jury could find that Duke actually instructed NTE not to pay those bills and ultimately walked NTE into an apparent breach of the Reidsville Interconnection Agreement. Moreover, it could find that Duke sacrificed short-term profits as part of its concerted efforts to keep NTE from expanding its footprint in the Carolinas.

The record includes evidence from which a jury could find that Duke sought out an opportunity to terminate its agreement with NTE in order to keep NTE from bringing the Reidsville plant online and to avoid having to compete with NTE on the merits because Duke believed it was at a “competitive disadvantage” efficiency-wise, which was “not

47

going away soon.” J.A. 5086. Indeed, some evidence indicated that Duke was eager for NTE to “breach!” so that Duke could “punt em!” from the queue. J.A. 6364. Although the district court did not consider such evidence in its opinion, “the record in this case comfortably supports an inference that the monopolist made a deliberate effort to discourage its customers from doing business with its smaller rival,” Aspen Skiing, 472 U.S. at 610, as Duke questioned whether it could “kick Reidsville out of the queue,” J.A. 6246, even when doing so would require Duke itself to terminate a profitable deal.

In short, we conclude that a reasonable jury could find that Duke’s unilateral termination of the Reidsville Interconnection Agreement and attendant disruption of NTE’s place in the OASIS queue was anticompetitive conduct.

Relying on Trinko, Duke nonetheless argues that FERC’s regulatory purview over its interconnection agreements forecloses NTE’s ability to complain of an antitrust refusal to deal. But Trinko has a more limited effect. As discussed above, in enforcing the Telecommunications Act of 1996, the Federal Communications Commission’s role was “to eliminate the monopolies.” Trinko, 540 U.S. at 415 (cleaned up). The Trinko Court thus observed that the Commission’s “complex regime for monitoring and enforcement” made it an “effective steward of the antitrust function.” Id. at 401, 413. In contrast here, however, FERC’s enforcement of the Federal Power Act, while aimed to promote competition in the market, is not designed specifically to eradicate existing monopolies. And, in any event, the timing of the events in this case shows that FERC’s ability to play any antitrust-enforcement function was limited. Duke unilaterally terminated the Reidsville Interconnection Agreement in September 2019 and promptly thereafter listed

48

NTE’s status in its OASIS queue as “canceled.” J.A. 714. Listing NTE’s status as “canceled” signaled to potential customers and investors that the Reidsville project would not move forward. And once NTE was apparently out of the picture, Duke and Fayetteville finalized the 2019 Power Supply Agreement. NTE did petition FERC for review of Duke’s unilateral termination, but by the time FERC issued an opinion agreeing with NTE, the damage had already been done — NTE had lost the opportunity to compete on efficiency grounds for an anchor customer, Fayetteville, and Fayetteville then became locked into a new contract with Duke for years into the future. NTE’s recourse at that point could not lie with FERC but with the courts. In short, FERC’s regulatory oversight, even as exercised, did not diminish the likelihood of significant antitrust harm because FERC was not timely presented with the full range of alleged anticompetitive conduct.

Regarding NTE’s claim that Duke interfered with its efforts to interconnect with Duke’s transmission line, Duke asserts that it had good reasons for terminating the Reidsville Interconnection Agreement and canceling NTE’s position in its OASIS queue. Of course, if that were a true fact, Duke would not be liable for violating § 2 under a refusal- to-deal theory. But accepting Duke’s business justifications as fact at this stage of the litigation would require resolving factual disputes in favor of Duke, in violation of the applicable standards for summary judgment. While Duke maintains that it “stopped dealing with NTE only when NTE stopped paying,” Response Br. at 27, NTE presented evidence that it stopped paying on the Reidsville Interconnection Agreement only at Duke’s request.

49

At bottom, the facts of record support a potential finding that Duke timed its unilateral termination of the Reidsville Interconnection Agreement to achieve anticompetitive ends. And we need not determine, as a matter of law, whether, if those facts are believed, such conduct in isolation amounted to a § 2 violation under a refusal-to- deal theory of liability. Rather, we recognize NTE’s claim that this conduct was but a part of a larger scheme. As NTE has shown, the interconnection dispute occurred during the very same time that Duke designed its retroactive rebate for Fayetteville to keep it from issuing an RFP. On NTE’s telling of the facts, the two prongs were executed simultaneously and to the same effect. As a synthetic consequence, NTE could not compete to secure an anchor customer for its Reidsville plant, thus depriving it of any practical ability to bring the more efficient plant to market. Such foreclosure to competition is precisely what § 2 seeks to proscribe.

C

In discussing the two prongs of Duke’s conduct, we have pointed to material disputed facts — both in NTE’s claims and Duke’s responses — which are sufficient to preclude summary judgment at this stage of the proceedings. Upon resolution of those disputed facts, a jury might well conclude that Duke’s conduct was simply good, old- fashioned competition, which, in the end, favors the consumers of electric power in the relevant market. On the other hand, the factfinder might just as well conclude that Duke saw a more efficient competitor in NTE and acted, through a broad range of anticompetitive conduct in various contexts, to eliminate that competition, to the detriment of consumers.

50

The facts supporting the parties’ conflicting positions, we conclude, are fairly disputed and therefore require a trial to resolve.

And there are numerous more particular examples. For instance, as discussed at length, Duke asserts that it “competed for Fayetteville by lowering its prices,” Response Br. at 55 (citing J.A. 4188–89), but NTE asserts that “Duke offered such a massive discount that NTE could not win the Fayetteville contract even though it produced power more efficiently,” Opening Br. at 53, and that Duke planned to recoup its lowered prices after pushing NTE out of the market. And as we have also discussed, Duke asserts that it terminated the Reidsville Interconnection Agreement only after NTE breached and that it simply tried to recover the money NTE owed, Response Br. at 55 (citing J.A. 4180–81), but NTE asserts that Duke unilaterally terminated the Reidsville Interconnection Agreement for a pretextual reason — Duke instructed NTE not to pay until it received an invoice, held off on sending invoices, demanded payments that NTE did not yet owe, and then terminated at the time when NTE needed the contract intact in order to compete for Fayetteville’s business, Opening Br. at 18 (citing J.A. 3894–97). Duke asserts that it reported the status of the Reidsville project as required by FERC regulations and promptly updated its terminology from “canceled” or “terminated” to “suspended” when it was informed that it had erroneously categorized NTE’s project status as terminated in its OASIS queue, Response Br. at 55 (citing J.A. 4190–91), but NTE asserts that Duke’s motivation for misidentifying the suspension as a termination was, in Duke’s own words, to “kick NTE Reidsville out of the queue” in order to “stop” the “NTE train,” Opening Br. at 13 (quoting J.A. 5733); id. at 49 (quoting J.A. 6246). Duke asserts that it innocently

51

intervened in state regulatory proceedings when NTE sought a permit for its Reidsville plant and that NTE “failed to disclose . . . that NTE . . . defaulted under the [Interconnection Agreement],” Response Br. at 55 (citing J.A. 517), but NTE asserts that “Duke falsely asserted to the [state agency] (among other things) that NTE lacked customers, even though the City of Camden had already notified Duke that it was moving its business to NTE,” Opening Br. at 20 (citing J.A. 515–19). Duke asserts that it “brought a meritorious contract claim” to recover the money NTE owed, Response Br. at 55 (citing J.A. 4189–90, 4207), but NTE asserts that the Duke’s suit for breach of contract was a “sham” and that it settled with Duke on that claim without admitting liability, Reply Br. at 5 (citing J.A. 4207).

Alongside those disputes, NTE has presented evidence that Duke’s conduct was deliberate in that Duke consciously sought to exclude NTE from the relevant market because it was as a more efficient rival and Duke’s efforts produced that effect. See Grinnell Corp., 384 U.S. at 571. NTE presented evidence addressing the mens rea of Duke’s conduct, suggesting that it amounted to anticompetitive malice. That evidence bolsters our conclusion that the case is trial worthy, as we have recognized that “summary procedures should be used sparingly in complex antitrust litigation where motive and intent play leading roles.” Dickson v. Microsoft Corp., 309 F.3d 193, 212 (4th Cir. 2002) (quoting Poller v. Columbia Broadcasting Sys., 368 U.S. 464, 473 (1962)).

In these circumstances, we conclude that many genuine disputes of material fact persist in this case, and accordingly we vacate the district court’s summary judgment and remand for further proceedings.

52

III

Finally, NTE contends that if we vacate the district court’s order, we should remand the case to a different district judge, as its motion to recuse Judge Bell under 28 U.S.C. § 455(a) should have been granted. Indeed, NTE argues that Judge Bell’s refusal to recuse himself serves as an independent ground to vacate the summary judgment.

After this case was assigned to Judge Bell, a partner from his former law firm appeared in the case in December 2019 on behalf of Duke. At the time, Judge Bell had in place a standing prophylactic policy, adopted when he ascended the bench, to recuse himself from cases involving lawyers from his former firm, and in accordance with that policy, the clerk of court reassigned the case to another judge, recording the reason on the docket as “conflict.”

Almost two years later, in October 2021, the judge to whom the case had been reassigned developed a conflict himself, and the case was then reassigned to Judge Bell who had by then abandoned his initial prophylactic policy. Nonetheless, NTE filed a motion that Judge Bell recuse himself, arguing that once a judge has recused himself in a case, he should not later return to that case, regardless of whether the original recusal was necessary or whether the original conflict had been resolved. Judge Bell denied the motion, concluding that his prior withdrawal “did not reflect a considered ‘recusal,’” J.A. 343, and NTE has appealed that order.

Judicial partiality or bias is a fundamental and structural procedural error. See Neder v. United States, 527 U.S. 1, 8 (1999). And accordingly, Congress has provided that even if a judge’s “impartiality might reasonably be questioned,” he should recuse himself from

53

the proceeding. 28 U.S.C. § 455(a). Indeed, even the appearance of partiality requires recusal.

Judge Bell, when newly appointed, responded with appropriate sensitivity to the rule by recusing himself from cases in which lawyers from his former firm appeared. But such a prophylactic rule needed to serve only a limited, albeit sufficient, amount of time to create a clearly perceived distance from his former firm.

No one contends that Judge Bell acted inappropriately when he disqualified himself initially, nor does anyone contend that Judge Bell acted inappropriately when he abandoned the initial prophylactic recusal policy. Rather, the question presented here is whether a judge, once he recuses himself from a case, can return to the same case later if circumstances have changed such that he no longer perceives himself to have a conflict or an appearance of one.

For good reasons, especially for the appearance of impartiality, we have held that once a judge is recused, the judge is “out of service insofar as that case is concerned” and that he “should take no action which would possibly affect the outcome of [the] case.” Arnold v. Eastern Air Lines, Inc., 712 F.2d 899, 904–05 (4th Cir. 1983) (en banc) (discussing whether a majority of “active judges” had voted to hear a case en banc). Such a brightline rule can be applied with ease and promotes the goal of ensuring public confidence in the impartiality of the judicial process. It also accords with the practices adopted by several other jurisdictions, which have implemented a “once recused, always recused” rule. See United States v. O’Keefe, 128 F.3d 885, 891 (5th Cir. 1997) (holding that a judge who recused himself after granting the defendant a new trial should not have

54

ruled on a motion for reconsideration); El Fenix de P.R. v. M/Y JOHANNY, 36 F.3d 136, 141 (1st Cir. 1994) (holding that a judge who has recused himself cannot reconsider the order of recusal); Moody v. Simmons, 858 F.2d 137, 143 (3d Cir. 1988) (holding that a judge should not have continued to enter non-ministerial orders after announcing his intention to disqualify himself because his daughter worked for one of the parties).

In this case, because Judge Bell had previously recused himself for a potential conflict, prudence instructs that he should not have reentered the case, even after what the docket identified as a “conflict” had been resolved. See, e.g., O’Keefe, 128 F.3d at 891 (“Once a judge recuses himself from a case, the judge may take no action . . . even when recusal is improvidently decided”); El Fenix, 36 F.3d at 141–42 (observing there was “no authority for” the proposition that “an improvident recusal order may be revisited by the recused judge”). That rule serves the judicial process well, and we adhere to it.

But, to be clear, we do not find that Judge Bell acted inappropriately in recusing himself in the first instance, nor do we find that he did not allow sufficient time to pass before abandoning his initial policy. Rather, we simply apply our precedent that eliminates gray areas, public confusion, and any question about the integrity of the judicial process.

* * *

For the reasons given, we vacate the district court’s June 24, 2022 summary judgment order and remand to a different district judge for further proceedings.

IT IS SO ORDERED.