OPINION
BOGGS, Circuit Judge.
Ameritech Michigan appeals from the district court’s judgment holding that the Michigan Public Service Commission did not reach a result contrary to federal law when it ordered Ameritech not to impose on BRE special construction charges associated with unbundling or conditioning certain local telephone loops, and found that Ameritech’s attempt to do so constituted unlawful discrimination. Ameritech has not carried its burden of demonstrating that the Commission acted in an arbitrary and capricious manner when it determined that Ameritech recovers or should recover the costs of the special construction work as part of its long-range recurring charges to BRE. We therefore affirm summary judgment in favor of BRE and the Commissioners on this aspect of Ameritech’s complaint. We reverse, however, the district court’s holding that the Commissioners’ finding of discrimination was consistent with federal law because the Commissioners improperly compared Ameritech’s treatment of BRE to its treatment of its retail customers. We remand for entry of an order directing the Commissioners to reconsider their discrimination finding according to the appropriate standard.
I To facilitate competition in the traditionally monopolized local telephone service market, the Telecommunications Act of 1996 (“TCA” or the “Act”), 47 U.S.C. § 151 et seq., imposed on Incumbent Local Exchange Carriers (ILECs), the telephone companies who held the monopoly, “a host of duties intended to facilitate market entry” by Competing Local Exchange Carriers (CLECs), telephone companies seeking to provide competing services. AT&T Corp. v. Iowa Utils. Bd. (Iowa Utilities Board II), 525 U.S. 366, 371-72, 119 S.Ct. 721, 142 L.Ed.2d 835 (1999) (citing 47 U.S.C. §§ 251, 252). The Act specifies three methods of competition: 1) the ILEC must provide to a CLEC that has or builds its own local telephone network, interconnection with the ILEC’s network, see 47 U.S.C. § 251(c)(2); 2) the ILEC must provide access to its own “network elements” on an “unbundled” basis to a CLEC wishing to acquire a network by leasing all or part of the ILEC’s network, see 47 U.S.C. § 251(c)(3); and 3) the ILEC must sell its retail services at wholesale prices to a CLEC planning simply to resell the incumbent’s services at retail prices. See 47 U.S.C. § 251(c)(4). See also GTE South, Inc. v. Morrison, 199 F.3d 733, 737 (4th Cir.1999). The ILEC must provide all of these products and services on a “nondiscriminatory” basis, see 47 U.S.C. § 251(c)(2)(D), (c)(3), (c)(4)(B), which term, “as used throughout section 251, applies to the terms and conditions an [ILEC] imposes on third parties as well as on itself.” In re Implementation of the Local Competition Provisions in the Telecommunications Act of1996, CC Docket No. 96-98,11 FCC Red. 15,499, ¶218 (Aug. 8, 1996) (“First Report and Order ”).
Access to an ILEC’s network facilities comes only through specified procedures for forming “interconnection agreements,” the Congressionally prescribed vehicle for implementing the substantive rights and obligations set forth in the Act. The Act requires ILECs to enter into these agree[583]*583ments with CLECs, see 47 U.S.C. § 251(c)(1); permits the parties to enter into a binding agreement without regard to the standards set forth in § 251(b) and (c), see 47 U.S.C. § 252(a)(1); requires submission of the final agreement to the appropriate state commission for approval, see 47 U.S.C. § 252(e)(1) and (2); gives the state commission authority to interpret and enforce agreements when post-approval disputes arise, see 47 U.S.C. § 252(c), (e)(1) and (2), and Michigan Bell Tel. Co. v. Climax Tel. Co., 202 F.3d 862, 868 (6th Cir.), cert. denied, 531 U.S. 816, 121 S.Ct. 54, 148 L.Ed.2d 22 (2000); and provides for federal court review “in any case in which a State commission makes a determination” under section 252, see 47 U.S.C. § 252(e)(6), and Michigan Bell, 202 F.3d at 866-68.
Michigan Bell Telephone Company d/b/a Ameritech Michigan (“Ameritech”) is the ILEC in Michigan. BRE Communications (“BRE”) is one of several CLECs attempting to compete with Ameritech by delivering local telephone service to customers in various locations. Ameritech and BRE executed an Interconnection Agreement,2 which gave BRE leasehold rights to certain elements of Ameritech’s network necessary for BRE to deliver service to consumers itself while maintaining the connection between these elements and the central elements of Ameritech’s network. Section 9.0 of the Interconnection Agreement tracked the terms of 47 U.S.C. § 251(c)(3), providing: “Ameritech shall provide non-discriminatory access to Network Elements on an unbundled basis ... on rates, terms and conditions that are just, reasonable and nondiscriminatory in accordance with the terms and conditions of this Agreement and Sections 251(c)(3) and 252 of the Act.”
One such localized facility, known as a “loop,” comprises three elements3 and links a customer’s home or business to a wire center, thereby providing access to the central Ameritech network and its connection to the worldwide telecommunication network. When a customer changes service from Ameritech to BRE and BRE seeks access to the same loop Ameritech used to serve the customer, in most cases Ameritech simply disconnects the loop from its switch and cross-connects it to BRE’s designated equipment or switch. Ameritech does not charge for this simple service. When two of the three elements of a loop are in place but not connected, or when the third and final element must be added, Ameritech must dispatch a technician to connect the loop in a single running linkage. Ameritech also does not charge CLECs for this relatively simple service. Advances in digital technology have, however, made certain transfers from Ameri-tech to a CLEC such as BRE a labor-intensive and expensive endeavor.
Traditionally, each customer’s communications ran over the customer’s own individual dedicated copper loop. Now, companies can increase efficiency by electronically integrating parts of loops into digital systems in which the middle elements of several customers’ loops meet in a single place and then join together to [584]*584share a single high-speed digital path to the wire center.4 A specific customer whose voice or data transmissions travel over the integrated path cannot easily be reassigned from Ameritech to BRE because only one company at a time can use the integrated path, known as a “bundled loop,” and the bundled loops cannot be “unbundled.”5 If Ameritech will still serve certain customers using that bundled loop after the requested transfer to BRE is made, the customer who wants to use BRE’s service has, essentially, no place to go. When BRE requests access loops running over integrated paths, Ameritech determines whether there exists a “spare” non-integrated loop element that follows the same route as the integrated loop. If so, Ameritech electronically disconnects the specified loop from its integrated element and transfers the loop to the spare non-integrated element at no extra charge. Where no spare single-loop facility exists, Ameritech generally must design and install one to parallel the integrated path in order for a CLEC to provide service to its customer; this is a process known as “disaggregating” or “de-multiplexing.”6 Ameritech seeks to charge CLECs the cost of installing these parallel loop elements.
A related problem arises in two more complex situations. Most of Ameritech’s loops were engineered and built to carry voice traffic, but a CLEC may want to use the loop to carry high-speed data traffic for its customer. Or, if the CLEC has several customers in one area, it may want to use a single high-capacity loop in lieu of multiple individual loops, ie., maintain its own integrated digital loop element. To fill such requests by CLECs like BRE, Ameritech technicians would have to perform special work to upgrade or “condition” the loops for high-speed data or high-capacity traffic. Ameritech seeks to charge CLECs for the cost of “conditioning” loops pursuant their requests.
On July 16, 1998, BRE filed a complaint against Ameritech with the MPSC, contesting Ameritech’s charges for work it performed on 65 loops.7 BRE claimed [585]*585that charging for these instances of work breached the Interconnection Agreement and constituted unlawful “discrimination” under the Agreement and the Act.8 Following the recommendation of its administrative law judge, the MPSC entered an order ruling in BRE’s favor on February 9, 1999. Rejecting Ameritech’s interpretation of the Interconnection Agreement, the Commission determined that the loops at issue were all “available” within the meaning of section 9.4.2 of the Agreement9 and that, therefore, any additional work Ameri-tech undertook to make these loops available to BRE was routine. Since the work was routine, Ameritech was expected to recover the costs associated with de-multi-plexing loops or conditioning loops in its standard tariff of recurring charges to CLECs. The Commission also concluded that Ameritech’s attempt to bill BRE for such standard work when it did not bill its own retail customers for similar work (other than as part of its recurring rates charged pursuant to tariffs) constituted forbidden discrimination. The Commission ordered Ameritech to refund all paid charges and cancel unpaid invoices, cease and desist imposing such charges against BRE, pay reasonable costs and attorney’s fees incurred by BRE in connection with the ease, and pay the state a fine of $170,000.
On March 11, 1999, Ameritech filed a complaint against BRE and Commissioners of the MPSC in the district court, seeking to overturn the MPSC’s order. The district court denied the Commissioners’ motion to dismiss for lack of jurisdiction, but it declined to exercise supplemental jurisdiction over “exclusive state law issues.” On January 4, 2000, the district court issued an opinion and order upholding the MPSC’s determinations and entered judgment in favor of the defendants. Ameritech timely appealed.
II
Jurisdiction
Ameritech filed suit pursuant to 47 U.S.C. § 252(e)(6), asserting that the MPSC’s order was contrary to the TCA and involved an erroneous interpretation of the BRE Ameritech Interconnection Agreement.10 Ameritech now appeals from a final order of the district court disposing of all claims, so this court has subject-matter jurisdiction under 28 U.S.C. § 1291. BRE does not contest the jurisdiction of either the district court or this court. The Commissioners, however, have [586]*586maintained that state sovereign immunity deprives federal courts of jurisdiction over them in their official capacities.
In holding that it had jurisdiction over the Michigan commissioners, the district court relied on Michigan Bell Telephone Company v. MFS Intelenet, Inc., 16 F.Supp.2d 817, 824-28 (W.D.Mich.1998), which held that Michigan constructively waived its sovereign immunity, ie., consented to suit, by accepting Congress’s invitation to regulate in the field. On appeal, the Commissioners contend that this was error.
In Climax Telephone, 202 F.3d at 867 n. 2, we declined to address just this question. Instead, we held that state public service commissioners are proper parties to a § 252(e)(6) suit to enjoin their enforcement of an interconnection agreement in purported violation of federal law under Ex parte Young, 209 U.S. 123, 28 S.Ct. 441, 52 L.Ed. 714 (1908). See Climax Telephone, 202 F.3d at 867-68. In the recent case of Verizon Maryland, Inc. v. Public Service Commission of Maryland, - U.S.-,-, 122 S.Ct. 1753, 1760-61, 152 L.Ed.2d 871 (2002), the Supreme Court confirmed this holding. We, therefore, again decline to address the question of whether states waive their sovereign immunity from suit by regulating telecommunications markets pursuant to the TCA. Instead', we reject the Commissioners’ contention that we lack subject matter jurisdiction over them, because the Ex parte Young doctrine provides the necessary jurisdiction in this case.
Standard of Review
As explained above, the Act declares that: “In any case in which a State commission makes a determination under this section [252], any party aggrieved by such determination may bring an action in an appropriate Federal district court to determine whether the ... statement meets the requirements of section 251 and this section [252].” 47 U.S.C. § 252(e)(6). As a result, this court’s review of the MPSC’s order is limited to determining whether the order is inconsistent with sections 251 and 252 of the Act. See Climax Telephone, 202 F.3d at 868. We review the MPSC’s interpretation of the Act de novo and do not accord any deference to its interpretation of the Act. Of course, we consider the FCC’s interpretation of the Act persuasive authority because Congress authorized the FCC to issue rules “to implement the requirements” of § 251, the section relating to duties and terms of interconnection, unbundled access, wholesaling, and other matters. However, the MPSC is not an “agency” within the meaning of the Administrative Procedure Act, see 5 U.S.C. § 701(b)(1), so the standards provided by the APA are not directly applicable. See GTE South, 199 F.3d at 745. And a “state agency’s interpretation of federal statutes is not entitled to the deference afforded a federal agency’s interpretation of its own statutes under Chevron.” Orthopaedic Hosp. v. Belshe, 103 F.3d 1491, 1495 (9th Cir.1997) (referring to Chevron U.S.A., Inc. v. Natural Res. Def. Council, Inc., 467 U.S. 837, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984)).
As for the MPSC’s findings of fact made in the course of exercising its enforcement authority, we join our sister circuits in applying the “arbitrary and capricious” standard. See, e.g., Southwestern Bell Tel. Co. v. Public Util. Comm’n, 208 F.3d 475, 481-82 (5th Cir.2000); US West Communications v. MFS Intelenet, Inc., 193 F.3d 1112, 1117, 1124 n. 15 (9th Cir. 1999).11 This standard
[587]*587is the least demanding form of judicial review of administrative action. When it is possible to offer a reasoned explanation, based on the evidence, for a particular outcome, that outcome is not arbitrary or capricious. Thus, the standard requires that the decision be upheld if it is the result of a deliberate principled reasoning process, and if it is supported by substantial evidence.
Killian, 152 F.3d at 520 (internal quotations and citations omitted).
Ill
Ameritech’s Special Construction Charges
Section 251(c)(3) of the Act imposes on ILECs “[tjhe duty to provide, to any requesting [CLEC], nondiscriminatory access to network elements on an unbundled basis at any technically feasible point on rates, terms, and conditions that are just, reasonable, and nondiscriminatory in accordance with the terms and conditions of the agreement and the requirements of this section and section 252 of this title.” 47 U.S.C. § 251(c)(3). Section 252 directs that determinations by a state commission of the just and reasonable rate for network elements “(A) shall be (i) based on the cost (determined without reference to a rate-of-return or other rate-based proceeding) of providing the ... network element ..., and (ii) nondiscriminatory, and (B) may include a reasonable profit.” 47 U.S.C. § 252(d)(1).
In its First Report and Order, the FCC reviewed these sections of the Act and opined on who should bear the costs of disaggregating bundled loops: “Commen-ters identify a number of other methods for. separating out individual, loops from [integrated digital loop carrier] facilities, including methods that do not require demultiplexing. Again, the costs associated with these mechanisms will be recovered from requesting carriers.” First Report and Order, ¶ 384. As to conditioning loops, the FCC offered that: “[I]f a competitor seeks to provide a digital loop functionality, such as ADSL, and the loop is not currently conditioned to carry digital signals, but it is technically feasible to condition the facility, the [ILEC] must condition the loop to permit the transmission of digital signals.... The requesting carrier would, however, bear the cost of compensating the [ILEC] for such conditioning.” First Report and Order, ¶ 382.12
In reliance on these FCC comments, Ameritech contends that “the hallmark of unbundled access is that the requesting carrier bears the cost and investment risk associated with physical facilities,” as distinguished from a reseller, which does not [588]*588obtain leasehold control over physical assets but simply purchases the use of an incumbent’s overall bundled service at wholesale and resells it to customers at retail. (Since the lines so resold still run to Ameritech over its bundled loops, there is no need to construct parallel loops to accommodate a reseller.) Ameritech argues that the MPSC’s Order contravenes the principles set forth by the FCC and violates the Act, because the rates paid by BRE for unbundled loops are not “based on the cost” of their provision, see 47 U.S.C. § 252(d)(1)(A), insofar as the Order purportedly saddles Ameritech with the cost of the additional provisioning work it must perform.
As BRE concedes, Ameritech correctly characterizes the FCC’s First Report and Order as directing that CLECs requesting access to unbundled loops should bear the cost of conditioning. However, the FCC did not state whether CLECs would pay these costs as special construction charges or, instead, as a built-in part of the overall recurring and non-recurring rates CLECs pay ILECs for access to requested elements of the ILECs’ networks. In essence, all parties agree that, pursuant to §§ 251(c)(3) and 252(d)(1), and consistent with First Report and Order ¶¶ 382 and 384, Ameritech is entitled to recover the costs it incurs in de-multiplexing and conditioning loops at BRE’s request. The question is when or by what means Ameri-tech shall do so. Obviously, Ameritech seeks to recover these costs by means of the special construction charges it assessed against BRE. BRE claims that Am-eritech already recovers these costs as part of the recurring and non-recurring tariffs BRE pays for access to Ameritech’s network.
Ameritech argues that paragraphs 382 and 384 create exceptions to the broader cost-recovery and compensation schemes for interconnection. Therefore, Ameritech concludes, the more specific provisions control over the general and declare the FCC’s intention that ILECs recover from CLECs the cost of provisioning requested network elements independent of recurring and non-recurring tariff schemes. Yet the supposedly “more specific” paragraphs 382 and 384 are not specific at all in addressing the critical question here because, as already stated, nothing in paragraphs 382 and 384 purport to explain when ILECs will recover these costs. Thus, Ameritech’s argument fails.
BRE contends, as do the Commissioners (consistent with their findings in the MPSC’s order), that allowing Ameritech to charge BRE separately for special construction costs gives Ameritech a “double recovery” because BRE already pays for these costs in its recurring monthly fees paid pursuant to Ameritech’s standard loop prices. The MPSC found that “most, if not all, of the special construction charges at issue in this proceeding relate to normal, routine types of costs that are already reflected in the costs and rates determined and approved by the Commission.” 13 In reaching its conclusion, the MPSC relied on testimony of expert witnesses who participated in the Ameritech “cost docket.” One witness, an MPSC staff auditor, stated his conclusion that “these special construction charges costs [sic] are recovered in the monthly recurring rates for unbundled loops.” He relied on Tariff M.P.S.C. No. 20R, Part 2-General Terms and Conditions, Section 5-Construction Charges, Sheet No. 1, which allows Ameritech to impose special con[589]*589struction charges in only three situations, none of which, he concluded, apply to the charges at issue.
We consider the MPSC’s conclusion that BRE already pays for the construction at issue here in its monthly fees to Ameritech a finding of fact. The questions of whether the Act or the interconnection agreement require BRE to make certain payments or to make them at certain times or by certain methods are questions of law. But whether payments already made included or were supposed to include compensation for certain services is a question of fact. Accordingly, the MPSC’s finding that BRE compensates Ameritech for the cost of de-multiplexing and conditioning the loops at issue will be disturbed only if it is not supported by substantial evidence. See Killian, 152 F.3d at 520.
In challenging, as not supported by substantial evidence, the MPSC finding that Ameritech’s Michigan tariffs of recurring charges include payment for de-multi-plexing and conditioning loops, Ameritech raises a new argument on appeal that it failed to make in the district court.
Providing guidance for implementing the Act, the FCC stated, in its First Report and Order, that prices for unbundled network elements should not reflect the cost of the ILEC’s “existing network infrastructures,” and that costs imposed to construct facilities should be spread out over time in monthly recurring charges the ILEC receives from CLECs. See First Report and Order, ¶¶ 683-86. Consistent with this position, the FCC created a long-run cost-recovery methodology called “Total Element Long Run Incremental Cost,” or “TELRIC,” which is based on the cost of a hypothetical “reconstructed” network that would employ new, “most efficient technology.”14 The tariffs of recurring charges at issue in this case are based on the TELRIC methodology.
Ameritech’s new argument is that Michigan’s TELRIC-based tariffs, based as they are on a hypothetical most-efficient method of maintaining network elements, do not contemplate compensating ILECs for the costs at issue in this case, because these costs would never arise in a network of the kind that the TELRIC model assumes exists. That is, no hypothetical, most-efficient network would contain elements that cannot be unbundled, and no hypothetical, most-efficient network for carrying data transmissions would include “load coils” traditionally used to improve the quality of voice communications. As Ameritech points out, BRE’s expert witness and Ameritech’s witness, who prepared the TELRIC cost studies, agreed that the TELRIC-based tariffs by which BRE compensates Ameritech for use of its network assume these non-ideal conditions do not exist. See Appellant’s Reply Brief at 22-24; see also Appellant’s Brief at 28-32.
As mentioned above, Ameritech raised this argument for the first time on appeal. Ameritech’s brief on the merits of its claim in the district court nowhere mentions TELRIC. Nor does it mention the tariffs [590]*590of recurring charges or the MPSC’s factual finding that BRE compensates (or would compensate, if Ameritech had filed more comprehensive cost studies) Ameritech for the special construction costs at issue here via these TELRIC-based tariffs. Ameri-tech apparently knew that it could raise such an argument. Its brief repeatedly cited the §§ 251(c)(3) and 252(d)(1) requirements that Ameritech be compensated for the cost of providing CLECs access to its loops on an unbundled basis; however, its argument relied on the proposition that paragraphs 382 and 384 of the First Report and Order required payment of the construction costs via special charges rather than recurring tariff-based charges.
This court does not ordinarily address new arguments raised for the first time on appeal. See Elkins v. Richardson-Merrett, Inc., 8 F.3d 1068, 1072 (6th Cir.1993). In explaining why, we have written:
“Propounding new arguments on appeal in attempting to prompt us to reverse the trial court-arguments never considered by the trial court-is not only somewhat devious, it undermines important judicial values.... In order to preserve the integrity of the appellate structure, we should not be considered a ‘second shot’ forum, a forum where secondary, back-up theories may be minted for the first time.”
Isaak v. Trumbull S & L Co., 169 F.3d 390, 396 n. 3 (6th Cir.1999) (quoting Estate of Quirk v. Commissioner of Internal Revenue, 928 F.2d 751, 758 (6th Cir.1991)). Therefore, we will not consider Ameri-tech’s contention that the TELRIC-based tariffs do not in fact compensate Ameri-tech for these costs because they are based on an assumption that these costs will not arise. However, in declining to consider this argument, we note that if the TELRIC-based tariffs for one reason or another in fact undercompensate Ameri-tech for these unusual charges, Ameritech can-as both BRE and the Commissioners agree-seek modification of its cost studies and tariffs to reflect these expenses.
The MPSC’s Finding of Discrimination
The MPSC found that Ameritech’s imposition of special construction charges on BRE constituted discrimination, in violation of the agreement and Michigan law, which contains discrimination provisions that parallel the Act. The MPSC first observed that Ameritech must treat CLECs the same as it treats itself, then concluded, “[t]he event that precipitates a finding of discrimination is Ameritech Michigan’s determination that under certain circumstances it can require BRE to pay special construction charges in connection with the provisioning of an unbundled loop when, under identical circumstances, it routinely foregoes the collection of such charges from its own customers to whom it is provisioning unbundled loops.” The MPSC apparently relied on BRE’s witness, who identified specific instances in which Ameritech assessed special construction charges on BRE for conditioning certain loops while it did not impose those charges when it (Ameritech) directly served the same retail customers, instead charging only the standard line-establishment charge of $42.
In offering additional support for the MPSC’s finding, BRE draws our attention to Ameritech’s tariff of retail charges. The tariff, BRE claims, shows that Ameri-tech does not recover from its customers the cost of conditioning and de-multiplex-ing loops when doing so is necessary to serve its customers by means of special construction charges. It recovers these costs over time in its recurring service charges or as part of its standard line-establishment fee. As an example of cir[591]*591cumstances in which Ameritech cannot, under the tariff filed with the MPSC, impose on retail customers a special construction charge, BRE cites cases where the loop was not “connected through” to the wire center or had a defective segment preventing it from forming a contiguous circuit.
In their brief to this court, the Commissioners contend that Ameritech’s interpretation of “nondiscriminatory” “suggests that any type of discrimination is appropriate as long as Ameritech discriminates equally among all CLECs.... If Ameri-tech’s view ... were adopted, Ameritech would be free to treat all CLECs in an abusive, discriminatory, and anti-competitive manner as long as it gave similar treatment to all CLECs, without regard to how it treats itself or its end-user customers.” MPSC’s Brief at 38 (emphasis added).
The problem with this statement, Ameri-tech contends, is that the Commissioners impermissibly added the condition that ILECs cannot discriminate between CLECs and the ILEC’s retail customers. Contrary to the Commissioners’ assertion, Ameritech readily concedes that the term “nondiscriminatory” means that it cannot discriminate among CLECs or between CLECs and itself in the provision of unbundled network elements.15 In fact, Am-eritech contends that its policy of incurring the cost of disaggregating (or aggregating) and conditioning loops used by its customers, then recovering that cost through recurring retail tariff charges, while billing BRE for similar' work .so that BRE bears the cost and can pass it on to its customer as it sees fit, is nondiscriminatory because it treats BRE exactly the same as it treats Ameritech. In short, whichever carrier actually provides the service bears the cost of multiplexing and conditioning and passes it on to its customer.
Ameritech correctly explains the fault in the MPSC’s reasoning in its brief to this court:
Unlike BRE, retail customers do not lease unbundled loops at cost-based rates. Instead, they buy something entirely different — namely, bundled ‘basic local exchange’ service — and they pay a different price (retail) for it.... While retail service is an end-to-end telecommunications service that includes switching (of local and long-distance calls), features (like touchtone and call waiting), and enhanced services (like directory assistance), an unbundled loop is simply a [592]*592path from the customer’s premises to an end office switch.
Appellant’s Brief at 34. Ameritech points out that if BRE wants to be treated like retail customers, BRE can pay Ameritech wholesale rates according to a scheme based on retail rates and then resell such service to customers it signs up, pursuant to §§ 251(c)(4) and 252(d)(3), which govern resale of communications services. In contrast, § 252(d)(1)(A) creates a separate pricing structure for unbundled network elements based on the “cost” Ameritech incurs to provide the network element. As Ameritech aptly explains, “the absence of special charges on the retail side is neither surprising nor sinister,” because retail customers do not lease pieces of the network but instead buy services provided by Am-eritech over its own existing network.
BRE argues that Ameritech “completely ignores” the FCC’s further elaborations on nondiscrimination. The First Report and Order states:
[T]o enable new entrants, including small entities, to share the economies of scale, scope, and density within the [ILECs’] networks, we conclude that [ILECs] must provide carriers purchasing access to unbundled network elements with the pre-ordering, ordering, provisioning, maintenance and repair, and billing functions of the [ILECs’] operations support systems. Moreover, the incumbent must provide access to these functions under the same terms and conditions that they provide these services to themselves or their customers.
First Report and Order ¶ 316 (emphasis added).
Reliance on this statement, however, is misplaced, because the construction charges at issue here do not arise in the normal course of serving customers. Am-eritech does not provide customers access to disaggregating or conditioning functions but performs those tasks for itself when it deems such alterations to existing loops beneficial in order to serve its customers better. In contrast, Ameritech must provide these services to BRE essentially on demand when Ameritech’s network does not serve BRE’s interest in competing with Ameritech. And BRE will, when the service is complete, obtain leasehold interests in the conditioned loops, while customers obtain no such interest; Ameritech retains the right to alter the elements of loops serving customers at its discretion. Another FCC Order, issued after the First Report and Order, confirms this analysis. See In re Application of Ameritech Michigan Pursuant to Section 271 of the Communications Act of 193b, CC Docket No. 97-137, 12 FCC Red. 20,543, ¶ 141 (Aug. 19,1997) (observing that “the ordering and provisioning of unbundled network elements” have “no retail analogue”).16
[593]*593The Act does not forbid Ameritech from discriminating between a CLEC requesting unbundled network elements and Am-eriteeh’s own retail customers, even if Am-eritech decides that it can best serve its customer by building the same network modifications as the CLEC happened to request. The MPSC’s order that Ameri-tech’s attempt to impose on BRE the special construction charges at issue here is based on a flawed interpretation of the nondiscrimination requirement of federal law. Accordingly, the district court’s judgment must be reversed and remanded with instructions to enjoin the enforcement of the MPSC’s order insofar as it relies on this flawed interpretation of the nondiscrimination requirement. Specifically, $40,000 of the $170,000 fine was based on a finding of discrimination, and assessment of that portion of the fine must be enjoined.
IV
For the reasons set forth above, the judgment of the district court is AFFIRMED IN PART, REVERSED IN PART, and REMANDED WITH INSTRUCTIONS to enter an order enjoining enforcement of a portion of the MPSC’s order.