CoreCivic Inc v. Governor of New Jersey

• Court: Court of Appeals for the Third Circuit
• Decided: July 22, 2025
• Docket: 23-2598
• Status: Published

Opinion

PRECEDENTIAL

UNITED STATES COURT OF APPEALS FOR THE THIRD CIRCUIT _______________

No. 23-2598 _______________

CORECIVIC, INC.

v.

GOVERNOR OF NEW JERSEY; ATTORNEY GENERAL OF NEW JERSEY, Appellants

_______________

On Appeal from the United States District Court for the District of New Jersey (D.C. No. 3:23-cv-00967) District Judge: Honorable Robert Kirsch _______________

Argued: May 1, 2025

Before: KRAUSE, BIBAS, and AMBRO, Circuit Judges

(Filed: July 22, 2025) Jeremy Feigenbaum [ARGUED] Nathaniel I. Levy Michael L. Zuckerman NEW JERSEY ATTORNEY GENERAL’S OFFICE 25 Market Street Richard J. Hughes Justice Complex P.O. Box 112 Trenton, NJ 08625 Counsel for Appellants

Alex Hemmer ILLINOIS ATTORNEY GENERAL’S OFFICE SOLICITOR GENERAL’S OFFICE 115 S. LaSalle Street, 23rd Floor Chicago, IL 60603 Counsel for Amici States of Illinois, Colorado, Connecti- cut, Delaware, Maine, Maryland, Massachusetts, Michi- gan, Minnesota, Nevada, New York, Oregon, and Washing- ton, and the District of Columbia in Support of Appellants

Farrin R. Anello Molly K.C. Linhorst AMERICAN CIVIL LIBERTIES UNION OF NEW JERSEY P.O. Box 32159 Newark, NJ 07102

2 Counsel for Amici AAPI New Jersey, American Civil Lib- erties Union of New Jersey, American Friends Service Committee, Bayard Rustin Center for Social Justice, Bend the Arc Jewish Action, Deportation & Immigration Re- sponse Equipo, Detention Watch Network, El Pueblo Unido of Atlantic City y Pueblos Cercanos, Faith in New Jersey, First Friends of New Jersey and New York, Latina Civic Action, Latino Action Network, Latino Coalition of New Jersey, LatinoJustice PRLDEF, Law Enforcement Ac- tion Partnership, Lazos America Unida, Make the Road New Jersey, New Jersey Alliance for Immigrant Justice, New Jersey Consortium for Immigrant Children, New Jer- sey Parents Caucus, Inc., New Jersey Policy Perspective, New Labor, Northern New Jersey Sanctuary Coalition, Re- formed Church of Highland Park, Truah Rabbinic Call for Human Rights, Unitarian Univeralist FaithAction New Jersey, Volunteer Lawyers for Justice, and Wind of the Spirit Immigrant Resource Center in Support of Appellants

David N. Cinotti Dominique Kilmartin Brendan M. Walsh PASHMAN STEIN WALDER HAYDEN 21 Main Street Court Plaza South, Suite 200 Hackensack, NJ 07601 Counsel for Amicus Pax Christi USA in Support of Appellant

David J. Goldsmith Thomas A. Kissane

3 Bradley D. Simon [ARGUED] SCHLAM STONE & DOLAN 26 Broadway, 19th Floor New York, NY 10004 Counsel for Appellee

McKaye L. Neumeister [ARGUED] UNITED STATES DEPARTMENT OF JUSTICE CIVIL DIVISION, APPELLATE SECTION 950 Pennsylvania Avenue NW, Room 7231 Washington, DC 20530 Counsel for Amicus United States in Support of Appellee

John M. Miano IMMIGRATION REFORM LAW INSTITUTE 103 Park Avenue, Suite E101 Summit, NJ 07901 Counsel for Amicus Immigration Reform Law Institute in Support of Appellee ______________

OPINION OF THE COURT _______________

BIBAS, Circuit Judge. Just as the federal government cannot control a state, so too a state cannot control the federal government. Each is sover- eign. Each is “protected from incursion by the other.” U.S. Term Limits, Inc. v. Thornton, 514 U.S. 779, 838 (1995) (Ken- nedy, J., concurring). But sometimes their authorities overlap. In such cases, some state rules may legitimately burden the

4 federal government. That is a “normal incident” in a system with dual sovereigns. North Dakota v. United States, 495 U.S. 423, 435 (1990) (plurality) (cleaned up). Sometimes, though, a state goes further, interfering directly with federal policy or “destroy[ing]” it through “hostile legislation.” McCulloch v. Maryland, 17 U.S. (4 Wheat.) 316, 400–01, 430 (1819). And when it crosses that line, it violates the Constitution. New Jersey is on the wrong side of that line. It dislikes some of the federal government’s immigration tools, so it passed a law with the “intent” to forbid new contracts for civil immigration detention. N.J. Stat. Ann. § 30:4-8.15(d). That law interferes with the federal government’s core power to enforce immigration laws. Its construction is admittedly clever: It seeks to sidestep the usual two-prong test that courts use to enforce the “bedrock principle” that states may not regulate their fed- eral counterpart. North Dakota, 495 U.S. at 448 (Scalia, J., concurring in the judgment). Still, we see the law for what “it really is”: a direct regulation on the federal government. McCulloch, 17 U.S. at 431. Because New Jersey’s law violates intergovernmental immunity, we will affirm the District Court’s summary judgment for the contractor. I. NEW JERSEY INTENDED TO BAN IMMIGRATION DETENTION Since 1996, CoreCivic has contracted with the federal gov- ernment to run a private immigration-detention center in Eliz- abeth, New Jersey. CoreCivic planned to renew its federal con- tract in 2023, but New Jersey passed a law (AB 5207) forbid- ding it to do so.

5 Though New Jersey does not want private immigration- detention centers, the government often relies on them. U.S. Immigration and Customs Enforcement (ICE) does not build its own lockups, and it does not operate them alone. Instead, it contracts with private companies or local governments to help run them. See 48 C.F.R. § 3017.204-90; 8 C.F.R. § 235.3(e) (both providing for federal contracting to hold immigrants). This approach gives ICE the flexibility it needs to increase or decrease capacity as the number of deportable aliens fluctuates. Citing its duty to protect human rights and health, New Jer- sey passed AB 5207 with the express “intent … to prevent new, expanded, or renewed agreements to detain people for civil immigration purposes.” N.J. Stat. Ann. § 30:4-8:15(d). The law bans the state, its local governments, and private parties from making, renewing, or extending any contract to detain people for civil immigration violations. § 30:4-8.16(b)(1)–(2). CoreCivic’s detention-center contract fell prey to that ban. So CoreCivic sued New Jersey, claiming that AB 5207 violates the Supremacy Clause because it (1) violates intergovernmental immunity and (2) is preempted by federal law. Soon after, the United States filed a statement of interest in the case. See 28 U.S.C. § 517. That is no surprise. Federal law gives the federal government discretion to find “appropriate places of detention for aliens detained pending removal.” 8 U.S.C. § 1231(g)(1). Exercising this discretion, the government has come to rely on CoreCivic’s detention center as a “mission critical location for [federal government] and ICE operations nationwide.” App. 100 ¶ 8. The center is the only one available in New Jersey “capable of meeting ICE’s requirements,” and its proximity to

6 JFK and Newark Airports makes it “crucial to effect[ing] remov- als from field offices nationwide.” App. 97 ¶ 28, 100 ¶ 9. With- out it, the government would have to take detainees to a center in the middle of Pennsylvania more than 250 miles (and a four- hour drive) away. Driving that far would tie up officers for “at least a full day.” App. 100 ¶ 10. It would also gum up ICE’s flexibility to grow or shrink capacity as the levels and locations of immigration shift. It could even force ICE to release aliens with violent criminal records. So the ban would effectively “cripple [ICE’s] law-enforcement operations in New Jersey and the surrounding region.” App. 99–100 ¶ 7. Based on these facts and concerns, the District Court granted summary judgment for CoreCivic. It thought that AB 5207 “evades easy classification under a particular branch of the Supreme Court’s Supremacy Clause jurisprudence.” App. 18. Still, it saw that the law takes away the federal gov- ernment’s choice of how to detain aliens, a restriction that it held violates intergovernmental immunity and is preempted by federal law. New Jersey now appeals. We review the District Court’s ruling de novo. Aleynikov v. Goldman Sachs Grp., Inc., 765 F.3d 350, 357 & n.2 (3d Cir. 2014). II. THE LAW VIOLATES INTERGOVERNMENTAL IMMUNITY BY DIRECTLY REGULATING THE FEDERAL GOVERNMENT

Our Constitution created a legal system “establishing two orders of government, each with its own direct relationship, its own privity, its own set of mutual rights and obligations to the people who sustain it and are governed by it.” Thornton, 514 U.S. at 838 (Kennedy, J., concurring). But when those orders conflict, the Supremacy Clause makes federal law “the

7 supreme Law of the Land.” U.S. Const. art. VI, cl. 2. States remain sovereign, but they are “subordinate to, and may be controlled by the constitution of the United States.” McCul- loch, 17 U.S. at 427. From this text spring two doctrines: inter- governmental immunity and preemption. First, the immunity doctrine shields the federal government from some state regu- lations. Second, Congress can extend that immunity further by passing a federal law to preempt state laws. North Dakota, 495 U.S. at 439–40 (plurality). CoreCivic raises both Supremacy Clause doctrines, but we need not reach preemption. Even without preemptive legislation, the law violates intergovern- mental immunity. Though New Jersey advocates forcefully and ably for its position, its law directly regulates the federal government. A. Intergovernmental immunity shields the federal government from direct or discriminatory state regulation Because federal law is supreme, “there is a plain repug- nance” in letting states “interfer[e] with or control[ ] the oper- ations of the Federal Government.” McCulloch, 17 U.S. at 431; United States v. Washington, 596 U.S. 832, 838 (2022). As McCulloch recognized, “the very essence of supremacy” empow- ers the federal government to “remove all obstacles to its action within its own sphere … [and] exempt its own operations from [state] influence.” 17 U.S. at 427. To enforce this core principle, known as intergovernmental immunity, modern courts apply a two-pronged test: States can- not “[1] regulate the United States [government] directly or [2] discriminate against” it or its contractors. Washington, 596

8 U.S. at 838 (cleaned up). A state law regulates the United States directly when it “places [either] a prohibition” or man- date on the federal government. Hancock v. Train, 426 U.S. 167, 180 (1976); see also Arizona v. California, 283 U.S. 423, 451 (1931) (“The United States may perform its functions without conforming to the police regulations of a state.”). Meanwhile, a state law discriminates against the federal gov- ernment when it “treats similarly situated state and federal [actors] differently” in a way that cannot be explained by “sig- nificant differences[s]” between the two. Dawson v. Steager, 586 U.S. 171, 177 (2019) (internal quotation marks omitted). A state law that violates either prong is invalid unless Congress has “clearly and unambiguously authorized” it. Washington, 596 U.S. at 840 (cleaned up). Congress has not authorized state regulations like AB 5207. But New Jersey claims that the law survives both prongs. It says the text of the law does not apply to the federal govern- ment directly. And New Jersey says it does not discriminate against the federal government because it ties its own hands in the same way: Its own Department of Corrections cannot hire private companies to house criminal defendants or convicts. But we need not decide whether New Jersey is right that it is a relevant comparator to the federal government or that it in- deed imposes the same restrictions on itself. Either way, this law plainly violates the direct-regulation prong.

9 B. Some laws directly regulate the federal government functionally, even if not literally AB 5207 falls because it directly regulates the federal gov- ernment. But to get there, we must first sketch out the contours of the direct-regulation prong. New Jersey contends that a law regulates the federal gov- ernment directly only if the law’s text applies to it. Applying that test, New Jersey claims that it has not directly barred the federal government from doing anything. The text of AB 5207 applies only to the state, its municipalities, and private contrac- tors (the sellers of the contracting service), not to the federal government (the buyer). It just so happens that AB 5207 has the exact same effect as a state law that bars the federal gov- ernment from contracting for private immigration services. But New Jersey says this does not matter. The direct-regulation prong does not concern itself with effects, the state says, even if functionally AB 5207 directly restricts federal power or sub- stantially interferes with federal operations. But it is wrong. AB 5207 carries the same sting as a law whose text applies expressly to the federal government. And the direct-regulation prong accommodates that functionalist reading of what is really going on here. Intergovernmental immunity is not a formalist doctrine. In gauging intergovernmental immunity, the Court has long instructed us to “look through form and behind labels to sub- stance.” City of Detroit v. Murray Corp. of Am., 355 U.S. 489, 492 (1958). We must probe the “purpose or self-evident oper- ation of a statute” to see if it is used to evade the limits on im- munity “by indirectly achieving the same result.” Miller v. City

10 of Milwaukee, 272 U.S. 713, 715 (1927). “[W]hat cannot be done directly cannot be done indirectly. The Constitution deals with substance, not shadows.” Students for Fair Admissions, Inc. v. President & Fellows of Harvard Coll., 600 U.S. 181, 230 (2023) (alteration in original) (quoting Cummings v. Mis- souri, 71 U.S. (4 Wall.) 277, 325 (1867)). So the Court has long held that, just as states cannot regu- late the federal government itself, they cannot regulate private parties in a way that severely undercuts a federal function. Osborn v. Bank of U.S., 22 U.S. (9 Wheat.) 738, 786–89, 866– 67 (1824) (noting that states cannot “control” federal opera- tions by regulating its contractors); Crandall v. Nevada, 73 U.S. (6 Wall.) 35, 44–45 (1867) (explaining that the Court has “uniformly denied” state regulations that “affect[ ] the func- tions of the Federal government” or “impede or embarrass the constitutional operations of that government”); Union Pac. R. v. Peniston, 85 U.S. (18 Wall.) 5, 30 (1873) (states may not impose regulations “the direct effect of which shall be to hinder the exercise of any powers which belong to the National gov- ernment”). That anti-interference throughline pervades the caselaw to this day. Smith v. Davis, 323 U.S. 111, 116 (1944) (upholding nondiscriminatory tax on federal contractor’s profits because there was “no basis for assuming that contractors will be any less willing to enter into construction contracts with the United States,” and the tax was not “likely to affect or impair in any way their ability to discharge their duties efficiently”); Graves v. New York ex rel. O’Keefe, 306 U.S. 466, 481 (1939) (holding that states may not “impose a burden on the national govern- ment tantamount to an interference … with the … performance

11 of its functions”); Okla. Tax Comm’n v. Tex. Co., 336 U.S. 342, 364 (1949) (explaining that state regulations on third parties that “actual[ly] interfere[e] [with] or [have] destructive effects upon the performance of obligations to or work for the govern- ment” violate intergovernmental immunity); City of Detroit, 355 U.S. at 495 (upholding non-discriminatory state regulation on federal contractors because there was “no crippling obstruc- tion of any of the Government’s functions [and] no sinister effort to hamstring its power”); United States v. Fresno County, 429 U.S. 452, 463–64, 463 n.11 (1977) (upholding neutral state tax on federal employees because it did not “threaten[ ] to obstruct or burden a federal function,” for instance “by making the Federal Government unable to hire anyone”); United States v. New Mexico, 455 U.S. 720, 735 & n.11 (1977) (upholding neutral tax on federal contractors but noting that “of course … state taxes on contractors are constitutionally invalid if they … substantially interfere with [the federal govern- ment’s] activities”); Davis v. Mich. Dep’t of Treasury, 489 U.S. 803, 814 (1989) (noting that intergovernmental immunity “protect[s] each sovereign’s governmental operations from un- due interference by the other”); Washington, 596 U.S. at 838 (reasoning that states may not “interfer[e] with or control[ ] the operations of the Federal Government”). Still, New Jersey contends that modern intergovernmental- immunity doctrine long ago jettisoned this functional, effects- based test. Our dissenting colleague likewise insists that the many Supreme Court cases embracing a functional view of direct regulation have been “rejected over and over.” Dissent at 12. But they misread the arc of the doctrine. True, there was a time when the Supreme Court stretched the doctrine to bar

12 state regulation of third parties that imposed even an indirect or conjectural financial burden on the federal government. See, e.g., New Mexico, 455 U.S. at 731 (describing mid-nineteenth- century cases striking down neutral state taxes on federal employ- ees, contractors, and private parties operating on federal land). Also true, the Court then corrected course, recognizing that such an “expansive” doctrine was unmoored “both from [its] constitutional foundations … and from the actual workings of our federalism.” Id. (internal quotation marks omitted). So it pared the doctrine back. See, e.g., James v. Dravo Contracting Co., 302 U.S. 134, 161 (1937) (upholding non-discriminatory tax as applied to federal contractor because it did “not interfere in any substantial way with the performance of federal func- tions”); Penn Dairies, Inc. v. Milk Control Comm’n, 318 U.S. 261, 269–70 (1943) (upholding milk price controls as applied to a federal milk supplier because even though the regulation “increase[d] the price which the government must pay for milk,” it “impose[d] no prohibition on the national government”). Yet even as the Court narrowed the doctrine’s applicability to third parties, it never eroded its anti-interference core. Wash- ington, 596 U.S. at 838 (describing intergovernmental immun- ity as “prohibiting States from interfering with or controlling the operations of the Federal Government”). As the doctrine stands today, nondiscriminatory state laws are no longer un- constitutional just because they may remotely affect federal functions. North Dakota, 495 U.S. at 435 (plurality) (summa- rizing how the doctrine has coalesced around this principle). But state laws regulating private parties still violate intergov- ernmental immunity if they “impose a burden on the national

13 government tantamount to an interference … with the … per- formance of its functions.” Graves, 306 U.S. at 481. In short, the modern doctrine distinguishes between laws that merely impose an incidental economic burden on the fed- eral government and those that subvert federal operations. The latter trigger immunity; the former do not. See Taber v. Indian Territory Illuminating Oil Co., 300 U.S. 1, 3–4 (1937) (distin- guishing a “nondiscriminatory” regulation “where there is only a remote, if any, influence upon the exercise of governmental functions” from “one which imposes a direct burden upon the exertion of governmental powers”); Pub. Utils. Comm’n v. United States, 355 U.S. 534, 543–44 (1958) (collecting cases and distinguishing between “nondiscriminatory state taxes on activities of contractors … who do business for the United States, as their impact at most is to increase the costs of the operation” and state laws that “place[ ] a prohibition on the Federal Government” by regulating third-party activity); see also GEO Grp., Inc. v. Newsom, 50 F.4th 745, 755 (9th Cir. 2022) (en banc) (noting this material distinction). Still, the dissent and New Jersey insist that the Supreme Court has since collapsed this distinction between mere bur- dens and substantial subversion. New Jersey, for its part, leans on Washington, which it claims embraced a hyper-formalist version of direct regulation. But Washington was a discrimina- tion case, so it did not consider, and had no occasion to con- sider, the bounds of direct regulation. 596 U.S. at 839. And we presume that the Court does not “overturn, or so dramatically limit, earlier authority sub silentio.” Shalala v. Ill. Council on Long Term Care, Inc., 529 U.S. 1, 18 (2000).

14 New Jersey and the dissent also point to Penn Dairies for the same point but misread its holding. That case rejected a challenge to Pennsylvania’s neutral price controls on milk because the law “at most … increase[d] the costs of [federal] operation[s]”; it “impose[d] no prohibition” on the federal gov- ernment. Pub. Utils. Comm’n, 355 U.S. at 543–44 (quoting Penn Dairies, 318 U.S. at 270). So Penn Dairies said nothing about the fate of regulations that functionally bar the federal government from doing something. But the Supreme Court has since spoken on that issue. It has explained that the Supremacy Clause does not “bar[ ] all state regulation which may touch the activities of the Federal Government.” Hancock, 426 U.S. at 179 (citing Penn Dairies, 318 U.S. at 269–70). But it does draw a line at those that “place[ ] a prohibition on the Federal Government.” Id. (quot- ing Pub. Utils. Comm’n, 355 U.S. at 544); see McHenry County v. Raoul, 44 F.4th 581, 592 (7th Cir. 2022) (drawing this same distinction). And as we discuss in more detail below, it has applied that rule to hold that certain state regulations on federal contractors can effectively “place[ ] a prohibition on the Federal Government,” thus violating intergovernmental immun- ity. Pub. Utils. Comm’n, 355 U.S. at 544. Without Penn Dairies, New Jersey and the dissent are left to lean on the Supreme Court’s plurality opinion in North Da- kota. They claim that it persuasively rejected an approach to intergovernmental immunity focused on substantial interfer- ence and control. Yet “only the result of North Dakota is bind- ing.” GEO Grp., 50 F.4th at 759. And even taking the plural- ity’s reasoning on its own terms, the case is inapt. North Da- kota addressed whether state reporting and labeling

15 requirements for out-of-state liquor suppliers could be applied to those supplying liquor to a military base. Crucially, those regulations did “not restrict the parties from whom the Gov- ernment may purchase liquor or its ability to engage in com- petitive bidding,” nor did they “require the military to submit to state control or to purchase alcoholic beverage from suppli- ers within the State or prescribed by the State.” 495 U.S. at 441, 443. Instead, those regulations “at worst raise[d] the costs of selling to the military,” thereby indirectly making it slightly “more costly for the Government to do its business.” Id. at 434, 441. True, as our dissenting colleague points out, these descrip- tions of the law’s minimal burden on the federal government come from the plurality’s preemption analysis. But that makes them no less factually accurate. And those facts informed the intergovernmental-immunity question as much as the preemp- tion one. The plurality concluded that, like the price controls at issue in Penn Dairies, North Dakota’s restrictions “regulate[d] federal activity [only] in the sense that they ma[d]e it more costly for the Government to do its business.” Id. at 434. Con- sistent with Penn Dairies, it found that mere economic burden insufficient to trigger intergovernmental immunity. Id. at 437. To be sure, the plurality was also wary of adopting an approach that would invalidate “every state regulation that in any way touched federal activity,” as the dissent points out. Id. at 437 n.8. But it simply did not have occasion to pass on the validity of state laws, like New Jersey’s, that do much more than touch federal activity. So it did not foreclose finding direct regulation when a novel state law, like New Jersey’s, effectively “oper- ate[s]” on or “direct[ly] interfere[s]” with a core function of the

16 federal government. Id. at 437. Plus, North Dakota reiterated that we must take “a functional approach to claims of govern- mental immunity.” Id. at 435. So even after North Dakota, state regulations “on contrac- tors are constitutionally invalid” under the intergovernmental- immunity doctrine “if they … substantially interfere with [the federal government’s] activities.” HMO of N.J., Inc. v. Whit- man, 72 F.3d 1123, 1132 (3d Cir. 1995) (quoting New Mexico, 455 U.S. at 735 n.11). Though the edges of the immunity doc- trine have “evolved,” its anti-interference and anti-control core has stayed solid. Washington, 596 U.S. at 838. C. AB 5207 directly regulates the federal government by banning contracts that only the federal govern- ment can make Applying the functional approach that intergovernmental immunity demands, this law directly regulates the federal gov- ernment. True, its text does not apply to the federal govern- ment. But we can easily see the law for what it really is: a reg- ulation “laid upon the contract of the government.” Dravo Contracting, 302 U.S. at 149. The law prevents the federal government from choosing how and through whom it will carry out a core federal function. It does so by banning private parties from selling immigration detention when “the only entity in the business, so to speak, of [buying private] immi- gration det[ention] is the federal government.” United States v. King County, 122 F.4th 740, 757 (9th Cir. 2024) (striking down a ban on deportation flights as violating both prongs of inter- governmental immunity). Only the federal government has the power to decide whether, how, and why to hold aliens for

17 violating immigration law. It alone has the power to make these contracts in the first place. See DeCanas v. Bica, 424 U.S. 351, 354–55 (1976); see also Arizona v. United States, 567 U.S. 387, 394 (2012) (describing the federal government’s immi- gration power as “broad [and] undoubted”). So this ban is in substance a direct regulation; it destroys the federal govern- ment’s marketplace. Cf. United States v. Town of Windsor, 765 F.2d 16, 19 (2d Cir. 1985) (“Enforcement of the substance of the permit requirement against the contractors would have the same effect as direct enforcement against the Government.”). The Supreme Court has relied on this same rationale to strike down other regulations of federal contractors that in sub- stance regulate the federal government. For instance, it invali- dated a state law that required private common carriers to get state approval before charging the federal government reduced rates. Pub. Utils. Comm’n, 355 U.S. at 535, 544. Technically, the law operated only on contractors and treated the federal government favorably compared to all other buyers. But the Court saw it for what it really was: not just a neutral regulation that affected the federal government but a “prohibition on the Federal Government.” Id. at 544. That was a “clear” Suprem- acy Clause violation. Id. Likewise, the Supreme Court has invalidated state laws that bar federal contractors from working within the state unless they meet certain qualifications “in addition to those that the [Federal] Government has pronounced sufficient.” Johnson v. Maryland, 254 U.S. 51, 57 (1920); see, e.g., Leslie Miller, Inc. v. Arkansas, 352 U.S. 187, 188–90 (1956) (per curiam) (inval- idating a state law that imposed extra licensing requirements on federal defense contractor on both immunity and

18 preemption grounds); see also United States v. Virginia, 139 F.3d 984, 987 (4th Cir. 1998) (barring the application of state “licensing and registration requirements to private investiga- tors working solely for the FBI”). Though such laws techni- cally apply only to the federal contractor, the Court has treated them functionally as bans on the federal government because they restrict its ability to hire whom it chooses. New Jersey’s law is more intrusive than such state licensing requirements and the state law that was struck down in Public Utilities Commission. Those laws just required state approval before a federal contractor could do business with the federal government; New Jersey’s law bans such contracts altogether. So it has the veneer of regulating contractors. But really, it directly regulates the federal government by telling it how to carry out a core function. It is a direct regulation in everything but name. See, e.g., Boeing Co. v. Movassaghi, 768 F.3d 832, 840 (9th Cir. 2014) (holding that a state law that “mandates the ways in which [a federal contractor] renders services that the federal government hired [them] to perform” is an improper direct regulation of the federal government because it effec- tively regulates the “terms of [the] federal contract itself”). New Jersey and the dissent try to distinguish Public Utili- ties Commission and the state-licensing case Leslie Miller as turning on preemption, not intergovernmental immunity. Dis- sent at 12. True, those cases noted “conflicts between federal and state law.” GEO Grp., 50 F.4th at 760. But they also “un- doubtedly drew on principles of intergovernmental immunity.” Id. (noting Public Utilities Commission’s reliance on McCul- loch and other intergovernmental immunity cases); see also United States v. City of Philadelphia, 798 F.2d 81, 89 (3d Cir.

19 1986) (describing Leslie Miller and Public Utilities Commis- sion as “involving questions of governmental immunity”). Prominent constitutional law scholars agree with this reading. See, e.g., Laurence Tribe, American Constitutional Law 393 & nn.11–12 (1978) (listing Public Utilities Commission and Leslie Miller as intergovernmental-immunity cases); Laurence H. Tribe, Intergovernmental Immunities in Litigation, Taxa- tion, and Regulation: Separation of Powers Issues in Contro- versies About Federalism, 89 Harv. L. Rev. 682, 702 & nn.91– 92 (1976) (same); Gerald Gunther, Constitutional Law 309 (10th ed. 1991) (treating Leslie Miller as intergovernmental- immunity case). The Ninth Circuit likewise relied on both cases to strike down a state ban on private detention contracts on intergovern- mental-immunity grounds. GEO Grp., 50 F.4th at 752, 757– 58; cf. McHenry, 44 F.4th at 593 (upholding law in which state refused to detain immigrants on federal government’s behalf— but which left “the federal government … free to … contract with private parties” for detention). We thus align ourselves with our sister circuit in adopting this approach. D. AB 5207 also directly regulates the federal govern- ment by substantially interfering with a core federal function We could stop there. But the law directly regulates the fed- eral government twice over by substantially interfering with its operations. Though the Supreme Court has not had to strike down a state law on these grounds recently, it has noted that some restrictions on federal contractors may violate intergov- ernmental immunity because they “substantially interfere

20 with” or control federal functions. New Mexico, 455 U.S. at 735 & n.11; see Fresno County, 429 U.S. at 463 n.11 (noting that a state tax or regulation that “destroy[s] the federal func- tion” would violate intergovernmental immunity). Indeed, it has implied that courts should treat such laws as regulating the federal government directly. See Goodyear Atomic Corp. v. Miller, 486 U.S. 174, 181 (1988) (explaining that “the federal function must be left free of … state regulation” even when “the federal function is carried out by a private contractor” (cleaned up)); Fresno County, 429 U.S. at 460, 464 (conclud- ing that a non-discriminatory tax on federal employees did not violate intergovernmental immunity because it did not “threaten[ ] to obstruct or burden a federal function” and at most “impose[d] an economic burden”); Taber, 300 U.S. at 3 (noting that the degree of “influence upon the exercise of gov- ernmental functions” is relevant to this inquiry). For instance, in Public Utilities Commission, the Court re- lied on such reasoning to explain why a state law violated in- tergovernmental immunity. The law barred federal officials from exercising their “discretion” to hire shipping contractors without state approval. 355 U.S. at 543. That restriction would have “delay[ed] … shipment[s]”, thus “seriously hamper[ing] or disrupt[ing] the military mission[s]” for which the ship- ments were made. Id. at 545 (internal quotation marks omit- ted). And if every state enacted similar restrictions, they would cripple national policy. Id. at 546. As one military officer tes- tified in that case: “We would find ourselves in an administra- tive morass out of which we would never fight our way, we would never win the war.” Id. (cleaned up).

21 AB 5207 suffers the same flaw. Federal law gives federal officials discretion to contract for immigration detention. 8 U.S.C. § 1231(g); 48 C.F.R. § 3017.204-90; 8 C.F.R. § 235.3. New Jersey’s law destroys that discretion. By barring all con- tractors from the market, it substantially interferes with federal immigration policy. It would shutter CoreCivic’s “mission crit- ical” detention center, undermining ICE operations “nation- wide.” App. 100 ¶ 8. ICE would have to tie up time and money building and running lockups itself, change its operations, and risk compromising national security. And if every state enacted such bans, they would “destroy the federal function.” Fresno County, 429 U.S. at 463 n.11. In response, New Jersey tries to frame its law’s effect as just a burden, not a ban. The federal government, it stresses, can still buy or lease its own detention centers. But that response fails. Unlike a broad-based tax or workplace-safety rule, this law altogether bans a type of contract, and it does so in a market that exclusively serves a federal power. The Founding genera- tion “surely … did not intend” for federal operations and the exercise of federal powers to “depend upon the discretion of the state governments.” McCulloch, 17 U.S. at 362. Although the reach of intergovernmental immunity has fluctuated over time, that core principle has remained steady from McCulloch to the present day. ***** At bottom, this law is an “effort to hamstring [the federal government’s immigration] power,” making it as hard as pos- sible for it to hold aliens in New Jersey. City of Detroit, 355 U.S. at 495. That is a big step down a slippery slope. If we

22 accepted New Jersey’s logic, consider what else states might be able to do. As New Jersey conceded at oral argument, under its logic, all fifty states could pass laws banning federal con- tractors from building weapons for the federal military. Such bans would cripple national defense. Cf. Osborn, 22 U.S. at 867 (“Can a contractor for supplying a military post with provi- sions, be restrained from making purchases within any State, or from transporting the provisions to the place at which the troops were stationed? or could he be fined or taxed for doing so? We have not yet heard these questions answered in the affirmative.”). Even a patchwork of such state laws “would defeat all the ends of government.” McCulloch, 17 U.S. at 432. States may not do that. Crandall, 73 U.S. at 46. Though “[t]he Framers split the atom of sovereignty,” they also put the Su- premacy Clause at the constitutional nucleus, shielding federal power from disruptive state collision. Thornton, 514 U.S. at 838 (Kennedy, J., concurring). E. New Jersey’s two remaining counterarguments fail Resisting the conclusion that the law directly regulates the federal government, New Jersey replies in two ways. Neither persuades. First, New Jersey and the dissent insist that federal contrac- tors’ immunity is “narrow” and cannot be expanded unless Congress chooses to do so. New Mexico, 455 U.S. at 737. Because Congress could preempt the New Jersey law if it wanted to, they argue that we should stay our hands. We agree that preemption lets Congress “confer immunity from state regula- tion on Government suppliers beyond that conferred by the Constitution alone.” Dissent at 15 (quoting North Dakota, 495

23 U.S. at 439). But a statute preempting New Jersey’s law is “un- necessary” here; the Supremacy Clause itself invalidates state regulations of contractors that “substantially interfere with” the Federal Government’s activities. Whitman, 72 F.3d at 1132. New Jersey’s logic would defang the direct-regulation prong of intergovernmental immunity and collapse intergovernmen- tal immunity into the preemption doctrine. Yet intergovern- mental immunity has independent bite, and courts must apply it to referee “clashing sovereignty.” McCulloch, 17 U.S. at 430. To be sure, CoreCivic is asserting that immunity on the fed- eral government’s behalf. But private parties are protected by intergovernmental immunity when the state law substantially interferes with their ability to carry out their work on behalf of the government. Okla. Tax Comm’n, 336 U.S. at 364. And both the Supreme Court and our sister circuit have let contractors assert this immunity. See Dravo Contracting, 302 U.S. at 149 (considering intergovernmental-immunity challenge brought by federal contractor); Boeing, 768 F.3d at 839–40 (holding that a state law binding a private contractor “directly interferes with the functions of the federal government” and so “violates intergovernmental immunity”). What is more, the federal gov- ernment is here as a friend of the court, agreeing with Core- Civic that New Jersey’s law will hobble federal immigration enforcement. Second, New Jersey protests that a ruling against it would open the door to far more claims of immunity. But our holding is narrow. We address only a state ban on contracting in a mar- ket where the federal government is the only available coun- terparty for services implementing a core federal power. Reg- ulations that merely burden contractors without substantially

24 interfering with the federal government’s operations, or those that impose neutral conditions on contracts rather than bans, may pose different intergovernmental-immunity questions. See, e.g., Nwauzor v. GEO Grp., Inc., 127 F.4th 750, 756–67, 771 (9th Cir. 2025) (rejecting intergovernmental-immunity challenge to applying state minimum-wage law to inmates at a private immigration-detention center). We leave such cases open. ***** “[T]he National Government is, and must be, controlled by the people without collateral interference by the States.” Thornton, 514 U.S. at 841 (Kennedy, J., concurring). Because New Jersey knew that it could not openly bar the federal gov- ernment from contracting to detain immigrants, it instead elim- inated everyone with whom the federal government might con- tract within its borders. It asks us not to notice the federal ele- phant in the room. Yet we can see the law for what it really is, “claiming the authority to dictate the manner in which the fed- eral [immigration] function is carried out.” Goodyear Atomic, 486 U.S. at 181 n.3. Letting states do that would “chang[e] totally the character of” our federal system by “transfer[ring] the supremacy, in fact, to the states.” McCulloch, 17 U.S. at 432. The U.S. Constitution is supreme, and intergovernmental immunity protects that supremacy. New Jersey’s law directly regulates the federal government, so it is unconstitutional as applied to CoreCivic. We will affirm.

25 AMBRO, Circuit Judge, dissenting In 2021, New Jersey enacted AB 5207, N.J. Stat. Ann. §§ 30:4-8.15–8.16. That law prohibits state, local, and private entities from engaging in civil immigration detention within the State. CoreCivic, which operates the lone private immigra- tion-detention facility in New Jersey, and the Federal Government argue that AB 5207 offends the Constitution’s Su- premacy Clause for two reasons. First, it violates intergov- ernmental immunity, which bars states from regulating the Federal Government. And second, it is preempted by various federal statutes that empower the Department of Homeland Se- curity to manage how it detains immigrants. New Jersey’s law no doubt affects the Federal Govern- ment’s civil immigration-detention operations. But neither intergovernmental immunity nor preemption invalidates AB 5207 in my view. Intergovernmental immunity covers only those state laws that either directly regulate or discriminate against the United States. AB 5207 does neither. It applies only to state, local, and private entities. And New Jersey also pro- hibits private general criminal detention, thus imposing the same restriction on itself. CoreCivic’s preemption argument fares no better. The main federal law it invokes—8 U.S.C. § 1231(g)(1)—provides only that the Department of Homeland Security must consider leasing or buying detention facilities before constructing its own. AB 5207 thus obstructs no federal statute. Fortunately, the Constitution provides a solution in net- tlesome federalism cases like this one: Congress can act. If it wants the Federal Government to retain the ability to contract with private detention companies, it may pass legislation say- ing so. Because the majority would instead force courts to make unguided decisions about when states interfere exces- sively with undefined federal interests, I respectfully dissent.

I. BACKGROUND

A. Federal Statutory Background Congress has given certain executive agencies, includ- ing the Department of Homeland Security (DHS) and U.S. Immigration and Customs Enforcement (ICE), significant dis- cretion in managing civil immigration detention. This discretion, mainly codified in the Immigration and Nationality Act (INA), 8 U.S.C. § 1101 et seq., includes deciding how and where to house immigration detainees. For example, Congress has provided that “[t]he [Secretary of Homeland Security] shall arrange for appropriate places of detention for aliens detained pending removal or a decision on removal.” Id. § 1231(g)(1). Before “initiating any project for the construction of any new detention facility,” DHS and ICE “shall consider the availabil- ity for purchase or lease of any existing prison, jail, detention center, or other comparable facility suitable for such use.” Id. § 1231(g)(2). But if existing “Government facilities … or [other] facilities adapted or suitably located for detention are unavailable for rental,” DHS may “expend … amounts neces- sary to acquire land and to acquire, build, remodel, repair, and operate” such facilities. Id. § 1231(g)(1). Congress has also authorized DHS to “make con- tracts … as may be necessary and proper to carry out the Secretary’s responsibilities.” 6 U.S.C. § 112(b)(2). DHS has promulgated regulations permitting ICE to enter contracts with detention facilities to house and detain immigrants as long as those facilities meet certain requirements. See 48 C.F.R. § 3017.204-90; 8 C.F.R. § 235.3(e).

2 ICE primarily houses civil immigration detainees in one of four kinds of facilities: “(1) Service Processing Centers; (2) Contract Detention Facilities; (3) Intergovernmental Ser- vice Agreement facilities; and (4) [spaces provided by] riders on U.S. Marshals Service … or Federal Bureau of Prisons (BOP) contracts.” App. 91. “Service Processing Centers are owned by ICE and staffed by a combination of federal employ- ees (who mainly provide medical care) and contract employees (who provide detention services).” App. 91. Contract Deten- tion Facilities, as the name implies, are “owned by private companies that contract directly with the government and are predominantly staffed by contract employees.” App. 91. Inter- governmental Service Agreement facilities involve agreements between ICE and state or local governments. And riders are interagency agreements between ICE and other federal agen- cies that directly manage their own detention facilities.

B. Immigration Detention in New Jersey and AB 5207 In 2021, ICE had entered into four contracts to house detainees in New Jersey: two intergovernmental service agree- ments with Essex and Hudson Counties; one U.S. Marshals’ agreement involving Bergen County; and one agreement with CoreCivic, which privately owned and operated the Elizabeth Detention Center (EDC). ICE entered into its contract with CoreCivic to operate EDC in 2005 for three years. It has since renewed that contract five times. EDC has the capacity to hold 304 detainees, and in aggregate housed more than 2,000 immi- gration detainees annually in 2022 and 2023. In 2021, New Jersey’s legislature passed and its Gover- nor signed AB 5207, N.J. Stat. Ann. §§ 30:4-8.15–8.16. It prohibits state and local agencies from engaging in civil immi- gration detention, id. § 30:4-8.16(b)(1), and prohibits any “private detention facility” in New Jersey from entering,

3 renewing, or extending any contract to provide private immi- gration detention, id. § 30:4-8.16(b)(2). The law does not affect existing contracts to provide immigration detention. Id. § 30:4-8.16(b)–(c). New Jersey enacted AB 5207 after finding that “[d]etention centers and correctional facilities in New Jer- sey have a history of poor conditions, including inadequate medical and mental health care, use of isolated confinement, and incidents of violence and retaliation against people in de- tention.” Id. § 30:4-8.15(c). After AB 5207 went into effect, Bergen, Essex, and Hudson Counties all announced that they would no longer con- tract with DHS or ICE to detain civil immigration violators. By the end of 2021, EDC became the only facility housing ICE detainees within 60 miles of New York City.

C. Procedural Background CoreCivic challenged AB 5207 on the ground that it violates the Supremacy Clause and sought an injunction against New Jersey’s Governor and Attorney General. The District Court agreed and entered summary judgment for CoreCivic. In its view, AB 5207 violates the Supremacy Clause in two ways: (1) it impermissibly interferes with the immigration-detention functions of the Federal Government, which are protected under intergovernmental immunity, and (2) it is preempted by Congress’s delegation of authority to DHS and ICE to consider leasing existing facilities. New Jersey timely appealed.

4 II. INTERGOVERNMENTAL IMMUNITY DOES NOT APPLY TO AB 5207 . The Supremacy Clause of the Constitution provides that federal law “shall be the supreme Law of the Land.” U.S. Const. art. VI, cl. 2. “State laws may violate the Supremacy Clause in two ways.” Treasurer of N.J. v. U.S. Dep’t of Treas- ury, 684 F.3d 382, 406 (3d Cir. 2012). First, “under the doctrine of intergovernmental immunity, states may not ‘regu- late the Government directly or discriminate against it.’” Id. (quoting North Dakota v. United States, 495 U.S. 423, 434 (1990) (plurality)). Second, “[u]nder the doctrine of federal preemption, state laws are invalid if they ‘conflict with an af- firmative command of Congress.’” Id. (quoting North Dakota, 495 U.S. at 434). The key difference between intergovernmental immun- ity and preemption is that the former applies irrespective of congressional direction. Immunity flows directly from the Supremacy Clause itself. In a world without federal statutes, intergovernmental immunity would still invalidate offending state regulations. Because intergovernmental immunity is such a powerful constraint on states, basic federalism and separa- tion-of-powers principles limit its application to the clearest state intrusions on federal sovereignty: direct regulations on the Federal Government itself and anti-federal discrimination. Preemption, by contrast, “provides Congress with the power to preempt state legislation if it so intends.” Treasurer of N.J., 684 F.3d at 406 (quoting Roth v. Norfalco LLC, 651 F.3d 367, 374 (3d Cir. 2011)) (internal alteration omitted). In other words, Congress gets to identify through legislation which fed- eral interests are important enough to override otherwise valid state law.

5 My colleagues never consider whether any congres- sional command preempts AB 5207. They instead rest entirely on intergovernmental immunity to strike it down. But using immunity to invalidate a law like AB 5207, which neither applies to nor discriminates against the United States, is like using a hammer to pound in a screw. To make it work, my colleagues create a new and problematic definition of “direct regulation.” I first address what I believe to be their misunderstanding. In my view, preemption, not immunity, is the proper tool for addressing neutral state laws that substan- tially but indirectly burden the Federal Government. Then, once we have the correct test in mind, I believe it becomes clear that AB 5207 satisfies the intergovernmental-immunity analysis.

A. Neutral State Laws that Apply Only to Private Parties Are Not Direct Regulations on the Federal Government. The first question is what constitutes a direct regulation. In the majority’s view, the test is “functional,” Maj. Op. 12, and we must “look through form and behind labels to sub- stance,” id. at 9 (quoting City of Detroit v. Murray Corp. of Am., 355 U.S. 489, 492 (1958)). To my colleagues, AB 5207 is apparently a “functional” direct regulation because it im- poses a “ban” in a market in which the Federal Government is the sole buyer and substantially interferes with a core federal function. I disagree with this argument for three reasons. First, as a matter of both language and law, a functional direct regulation is a contradiction in terms. A regulation is di- rect only when it applies to the object of regulation. A state law that does not apply to the United States or some entity “so closely connected to the [Federal] Government that the two cannot realistically be viewed as separate,” United States v.

6 New Mexico, 455 U.S. 720, 735 (1982), is not a direct regula- tion, and its downstream effects cannot transform it into one. State laws that apply only to private contractors but still affect the Federal Government, even substantially, are indirect regu- lations. Second, the majority’s “functional” direct-regulation test undermines bedrock federalism and separation-of-powers principles. Preemption, not immunity, is the appropriate limi- tation on neutral state laws that do not apply to the Federal Government yet still affect it. That is because preemption is more “accommodating of the full range of each sovereign’s legislative authority and respectful of the primary role of Con- gress in resolving conflicts between the National and State Governments.” North Dakota, 495 U.S. at 435. Congress, after all, has superior institutional capacity to identify which federal interests are so strong that they displace a state’s otherwise valid exercise of its police powers. And third, the majority’s limiting principles—that its rule applies only to regulations that (1) affect markets in which the United States is the sole buyer, (2) impose bans, and (3) in- terfere with core federal functions—are inadministrable.

1. State Laws Must Apply to the Federal Government to Regulate It Directly. The majority’s main argument is that state laws that do not apply to the Federal Government may still count as direct regulations on the United States for immunity purposes be- cause states “cannot regulate private parties in a way that severely undercuts a federal function.” Maj. Op. 11. In my col- leagues’ view, it is a state regulation’s ultimate effect, not its legal application, that makes it direct.

7 But that is not the law, and it has not been for some time. A state law directly regulates the Federal Government only if it applies to the United States or some entity standing in its shoes. Regulations on private contractors that only collaterally affect the Federal Government are indirect. The cases my colleagues cite for the proposition that regulations on private parties count as direct regulations on the United States itself are from the nineteenth century. Id. at 11 (citing Osborn v. Bank of U.S., 22 U.S. (9 Wheat.) 738 (1824); Crandall v. Nevada, 73 U.S. (6 Wall.) 35 (1867); Union Pac. R. v. Peniston, 85 U.S. (18 Wall.) 5 (1873)). Far more recently, however, the Court reined in the excesses of its early immunity doctrine. And it told us so in no uncertain terms. “At one time, [it] struck down many” state regulations on the ground that they “interfered with ‘the constitutional means which have been legislated by the government of the United States to carry into effect its powers.’” North Dakota, 495 U.S. at 434 (quot- ing Dobbins v. Commn’rs of Erie Cnty., 16 Pet. 435, 449 (1842)). But “that view has now been ‘thoroughly repudi- ated.’” Id. (quoting South Carolina v. Baker, 485 U.S. 505, 520 (1988)). The modern cases the majority should rely on tell us that a direct regulation governs “the performance, by federal offic- ers and agencies, of governmental functions.” Penn Dairies, Inc. v. Milk Control Comm’n of Pa., 318 U.S. 261, 269 (1943) (emphasis added). But make no mistake: “[T]hose who con- tract to furnish supplies or render services to the government are not such agencies and do not perform governmental func- tions.” Id. (emphasis added). So when a state “regulation operate[s] against suppliers, not the Government,” then “con- cerns about direct interference with the Federal Govern- ment … are not implicated.” North Dakota, 495 U.S. at 437

8 (citations omitted). After all, if the Federal Government “de- liberately opt[s] for the ‘genius’ of private enterprise in the operation of its” federal functions, it “enjoys the benefits that are derived from private operations, but by the same measure, it must also suffer any reciprocal burdens.” United States v. Pa. Env’t Hearing Bd., 584 F.2d 1273, 1279 (3d Cir. 1978). The majority insists, however, that “even as the Court narrowed the doctrine’s applicability to third parties, it never eroded its anti-interference core.” Maj. Op. 13. I agree that pre- venting state interference with federal functions is the principle animating intergovernmental immunity. But intergovernmen- tal immunity is not responsible for—or even capable of— carrying that burden on its own. When a state law neutrally ap- plies only to private contractors, then it is preemption, not immunity, that takes over as the main anti-interference doc- trine. The very cases the majority cites underscore how ill- suited intergovernmental immunity is to the task of policing state laws that regulate private contractors. For instance, the majority claims that United States v. New Mexico upheld a “neutral tax on federal contractors” be- cause it did not “substantially interfere with [the federal gov- ernment’s] activities.” Maj. Op. 12 (quoting 455 U.S. at 735 n.11) (alterations in original). But New Mexico explains that regulations on federal contractors only rarely count as direct regulations on the Federal Government itself. The United States in that case tried to extend intergovernmental immunity to certain national laboratories by designating them federal agents. 455 U.S. at 737. The Court rejected that effort as a “wooden formalism” that would overextend immunity. Id. When a state law is neutral, intergovernmental “[i]mmunity is appropriate in only one circumstance: when the levy falls on the United States itself, or on an agency or instrumentality so

9 closely connected to the Government that the two cannot real- istically be viewed as separate entities.” Id. at 735. To enjoy immunity, a private contractor “must actually ‘stand in the Government’s shoes.’” Id. at 736 (quoting Murray Corp., 355 U.S. at 503). The majority likewise relies on United States v. Fresno County, claiming that it upheld a “state tax on federal employ- ees because it did not ‘threaten[] to obstruct or burden a federal function.’” Maj. Op. 12 (quoting 429 U.S. 452, 464 (1977)). Not quite. Fresno explains that private parties, even ones serv- ing federal functions, ordinarily do not enjoy intergovern- mental immunity. “The ‘legal incidence’ of the tax in- volved … [fell] neither on the Federal Government nor on federal property.” Fresno, 429 U.S. at 464 (emphasis added). It was “imposed solely on private citizens who work for the Federal Government.” Id. So the tax was invalid “only if it dis- criminate[d] against the Forest Service or other federal em- ployees.” Id. Although the tax “threaten[ed] to interfere with federal laws relating to the functions of the Forest Service,” it did so by “removing an advantage otherwise enjoyed by the Federal Government in the employment market”—not enough to count as a direct regulation. Id. The majority cites footnote 11 of Fresno for, I presume, its discussion of how a state could hypothetically levy a tax on federal employees at a rate so high it would “destroy the federal function performed by” the Federal Government. Id. at 463 n.11. But the Court explained only sentences later that the safe- guard against that kind of tax would not be the bar on direct regulations, but the antidiscrimination principle. To pass con- stitutional muster, a federal-function-destroying tax would need to apply neutrally. But that “danger would never arise” because democratic political pressure would discourage

10 elected officials from “impos[ing]” such a tax “on the income and property interests of all other residents and voters of the [s]tate.” Id. “The political check against abuse of the taxing power found lacking in McCulloch [v. Maryland, 17 U.S. 316 (1819)], where the tax was imposed solely on the Bank of the United States, is present where the State imposes a nondiscrim- inatory tax ….” Id. at 463. The Fresno Court never suggested that such a tax would be unconstitutional on the ground that it somehow directly regulated the United States. To recap: the Supreme Court tells us that a state law di- rectly regulates the United States only when that law applies to the Federal Government or private entities that cannot realisti- cally be viewed as separate from the Federal Government itself. It is not direct when it applies only to private contractors, even those performing federal functions. And a state law that does not apply to the Federal Government but collaterally af- fects it is not a “functional” direct regulation. It is an indirect regulation.

2. The Majority’s Effects-Based Test Forces Courts to Make Judgment Calls Better Left to Congress and Cannibalizes Preemption. To be sure, New Jersey has not discovered a loophole in the Supremacy Clause. States do not have free rein to “un- dercut[] a federal function” through neutral indirect regulation. Maj. Op. 11. Preemption—a doctrine that courts apply far more often than intergovernmental immunity—fills that gap. Yet the majority tries to jam neutral indirect regulations that burden the Federal Government into the intergovernmental- immunity framework rather than analyzing them through preemption.

11 The Supreme Court’s decision in North Dakota, which considered how to treat nondiscriminatory state laws that indi- rectly burden the Federal Government, illustrates the majority’s error. Justice Stevens, writing for the plurality, con- cluded that “[a] state regulation is invalid only if it regulates the United States directly or discriminates against the Federal Government or those with whom it deals.” 495 U.S. at 435. “Whatever burdens are imposed on the Federal Government by a neutral state law regulating its suppliers ‘are but normal inci- dents of the organization within the same territory of two gov- ernments.’” Id. (quoting Helvering v. Gerhardt, 304 U.S. 405, 422 (1938)). “Claims to any further degree of immunity must be resolved under principles of congressional pre-emption.” Id. (citing Penn Dairies, 318 U.S. at 271). Justice Brennan, by contrast, would have looked to the regulation’s effect on the Federal Government’s operations: “[C]ontrary to the plurality’s view, … those dealing with the Federal Government enjoy immunity from state control not only when a state law discriminates but also when a state law actually and substantially interferes with specific federal pro- grams.” Id. at 451–52 (Brennan, J., concurring in the judgment in part and dissenting in part). Like my colleagues in the ma- jority, Justice Brennan would have extended federal immunity to nondiscriminatory state regulations that “substantially ob- struct[] … affirmative federal policies.” Id. at 452. In the decades leading up to North Dakota, Justice Brennan’s view had been rejected over and over, making it clear that Justice Stevens, even in a non-binding plurality opin- ion, had the better of the debate. The majority’s holding here, just like Justice Brennan’s theory on which it relies, is incom- patible with bedrock principles of federalism. The Court in Penn Dairies held outright that “[s]ince the Constitution has

12 left Congress free to set aside local … regulation of govern- ment contractors which burden the national government, [there is] no basis for implying from the Constitution alone a re- striction upon such regulations which Congress has not seen fit to impose ….” 318 U.S. at 271. The Court similarly held in New Mexico that a state tax on national laboratories did not trigger freestanding constitutional immunity, regardless of its effect on federal functions, because “[s]uch complex problems are ones which Congress is best qualified to resolve.” New Mexico, 455 U.S. at 744 (quoting United States v. City of De- troit, 355 U.S. 466, 474 (1958)) (alteration in original). The plurality in North Dakota in my view correctly syn- thesized these cases to explain that intergovernmental immunity does not extend to neutral and indirect regulations on the United States. This approach is more “accommodating of the full range of each sovereign’s legislative authority and respectful of the primary role of Congress in resolving conflicts between the National and State Governments.” North Dakota, 495 U.S. at 435. My colleagues claim that Penn Dairies is distinguisha- ble because the state law there merely “burdened the federal government incidentally”; it “‘impose[d] no prohibition’ on the federal government.” Maj. Op 15 (quoting Pub. Utils. Comm’n v. United States, 355 U.S. 534, 543–44 (1958)) (alterations in original). In their view, “Penn Dairies said nothing about the fate of regulations that functionally bar the federal government from doing something.” Id. This argument fails for two rea- sons. First, it tries to draw a legally administrable distinction between a burden and a ban. As I explain below, even the United States conceded at oral argument that it could not define when a state law crosses that line. We were told that we will

13 know one when we see one. See infra at 19–21. The majority appears stumped too because it also fails to supply any frame- work. But second, and more important, my colleagues are simply wrong about what Penn Dairies says. I grant them that the Court did not use the phrase “regulations that functionally bar the federal government from doing something.” Maj. Op. 15. But Penn Dairies did say that “governmental immun- ity from state taxation and regulation” does not extend “beyond the national government itself” and the “governmental func- tions performed by its officers and agents.” Penn Dairies, 318 U.S. at 270 (emphasis added). Today, the majority cries foul over the same kind of state regulations that Penn Dairies dis- regarded, those that “inevitably impose[] some burdens on the national government of the same kind as those imposed on cit- izens of the United States within the state’s borders.” Id. at 271. But “those burdens, save as Congress may act to remove them, are to be regarded as the normal incidents of the operation within the same territory of a dual system of government, and … no immunity of the national government from such bur- dens is to be implied from the Constitution which established the system.” Id. I do not know how to read this language as saying anything but that state regulations on private parties that burden the Federal Government are for Congress to manage. The majority nonetheless insists that the Court has “spo- ken” on this issue in Public Utilities Commission, which supposedly held that state regulations on federal contractors that “effectively ‘place[] a prohibition on the Federal Govern- ment’” violate intergovernmental immunity. Maj. Op. 15 (quoting Pub. Utils. Comm’n, 355 U.S. at 544) (emphasis added). One problem: Public Utilities Commission dealt with preemption, not immunity. And you need not take my word for

14 it. The Court tells us that it “put to one side ‘cases where, ab- sent a conflicting federal regulation, a State seeks to impose safety or other requirements on a contractor who does business for the United States.’” North Dakota, 495 U.S. at 435 n.7 (quoting Pub. Utils. Comm’n, 355 U.S. at 543) (emphasis added). It “invalidated the state law because there was a clear conflict between the state policy of regulation of negotiated rates and the federal policy, expressed in statute and regula- tion, of negotiated rates.” Id. (emphases added). The same was true of another case the majority relies on, Leslie Miller, Inc. v. Arkansas, 352 U.S. 187 (1956): “the state licensing law came into direct conflict with ‘the action which Congress … ha[d] taken to insure the reliability of persons and companies con- tracting with the Federal Government.” North Dakota, 495 U.S. at 435 n.7 (quoting 352 U.S. at 190) (emphases added). That, it bears repeating, is about preemption. The majority also believes that I misread the plurality opinion in North Dakota. In my colleagues’ view, that case is distinguishable because the state regulations there did “not re- strict the parties from whom the Government may purchase liquor or its ability to engage in competitive bidding,” nor did they “require the military to submit to state control or to pur- chase alcoholic beverage[s] from suppliers within the State or prescribed by the State.” Maj. Op. 16 (quoting North Dakota, 495 U.S. at 441, 443). Without a hint of irony, the majority recites the North Dakota Court’s discussion of preemption, not immunity. The plurality there begins the discussion the majority quotes by not- ing that “[t]he conclusion that the labeling regulation does not violate the intergovernmental immunity doctrine does not end the inquiry into whether the regulation impermissibly inter- feres with federal activities.” 495 U.S. at 439. Why? Because

15 “Congress has the power,” by preemption, “to confer immunity from state regulation on Government suppliers beyond that conferred by the Constitution alone.” Id. For the reasons the majority now parrots, the Court concluded that Congress had “not … spoken with sufficient clarity to pre-empt North Da- kota’s attempt to protect its liquor distribution system.” Id. at 440 (emphasis added). In fact, when the North Dakota plurality did discuss im- munity, it rejected the very premise of my colleagues’ holding: that courts, without congressional direction, can distinguish be- tween incidental economic burdens and excessive interference with federal interests. North Dakota’s labeling requirement stopped five out of six alcohol suppliers from selling to an in- state military base. Id. at 437 n.8. The sixth raised its prices by as much as $20.50 per case. Id. That seems awfully close to a “regulation[] that functionally bar[s] the federal government from doing something.” Maj. Op. 15. Yet even though the state regulation functionally barred a United States military base from purchasing alcohol, the plurality “decline[d] to embark on an approach that would either result in the invalidation or the trial, by some undisclosed standard, of every state regula- tion that in any way touched federal activity.” North Dakota, 495 U.S. at 437 n.8. It did not, as my colleagues seem to be- lieve, contemplate an exception for when courts think they know better. The majority takes an opinion that stands for the opposite proposition, quotes it selectively, and holds it out as supporting their novel test. I would admire the alchemy if I were not so alarmed by the product. Last, the majority claims that Justice Stevens’s ap- proach in North Dakota would “collapse intergovernmental immunity into the preemption doctrine.” Maj. Op. 24. Alt- hough this argument is hard to follow, I understand my

16 colleagues to mean that a more tailored conception of direct regulation would make that prong of the analysis a dead letter, leaving everything to preemption. I disagree. Justice Stevens’s test would not “defang the direct-regulation prong,” as the ma- jority claims. Id. It would invalidate all state laws that apply to the United States—for example, those that would prohibit any party from printing money or detaining immigrants in New Jer- sey. It is the majority’s view, not mine, that would merge preemption and immunity. Indeed, the majority’s test would seem to make it easier to invoke freestanding constitutional im- munity than preemption, which does not permit this kind of “freewheeling judicial inquiry into whether a state statute is in tension with federal objectives.” Chamber of Com. v. Whiting, 563 U.S. 582, 607 (2011) (internal quotation marks omitted).

3. The Majority’s Limiting Principles Are Unworkable. Perhaps recognizing the gravity of the task it has set for courts, the majority tries to characterize today’s decision as a ticket for one ride only. On its telling, “[w]e address only a state ban on contracting in a market where the federal govern- ment is the only available counterparty for services implement- ing a core federal power.” Maj. Op. 24–25. I count three sup- posed limiting principles: that the majority’s expanded direct- regulation test applies only to state laws that (1) regulate mar- kets in which the United States is the sole buyer, (2) impose bans, and (3) limit core federal functions. But on closer inspec- tion, these limitations provide little direction. Each is unworkable, forcing judges to draw their own politically sen- sitive conclusions. When faced with these deficiencies, the majority has nothing to say. Unfortunately, they leave it to a future court to sort out the mess.

17 i. Markets in which the United States is the sole buyer. The first limitation the majority identifies is that AB 5207 applies to a market in which the United States is the only buyer. This seems persuasive at first blush. After all, if a state purports to regulate a market in which the United States is the sole buyer, then we may have a strong indication that the state is targeting the Federal Government indirectly. The ma- jority stumbles, however, in characterizing that kind of state law as a direct regulation. The immunity test already accounts for state laws that single out the Federal Government: they are unconstitutional because they are discriminatory. By calling its single-market-participant claim a direct-regulation argument rather than a discrimination one, the majority capitalizes on the intuition that New Jersey is discriminating, but without identi- fying someone treated better. A state, however, “does not discriminate against the Federal Government and those with whom it deals unless it treats someone else better than it treats them.” United States v. Washington, 460 U.S. 536, 544–45 (1983). And as New Jersey has repeatedly explained, it takes its own medicine by separately banning private general crimi- nal detention. The majority’s argument also forces us to grapple with complex questions about how to define the relevant market. My colleagues claim that New Jersey regulates private immi- gration detention, a market from which only the United States buys. But that is true only if you ignore New Jersey’s separate prohibition on general private detention. For the majority’s test to make sense, my colleagues must accept one of two proposi- tions. First, AB 5207 would have been constitutional if it had instead been passed in a single statute banning all private de- tention in New Jersey. It is unlawful here because it was passed

18 separately, even though the substantive effect is the same. Or second, even a statute like the one I described would be partly unconstitutional because every neutral statute that applies gen- erally to contractual relationships could be recharacterized as a partial direct regulation on the United States. Neither makes sense. Even the United States agreed that the framing endorsed by the majority is a discrimination argument. When asked whether it matters if we define the market as private immigra- tion detention or private detention generally, the Government demurred, stating that it does not “understand that framing to be very important when … talking about the regulation prong.” Tr. 72:3–5. When pressed on whether the market definition “bear[s] on how much we look through form to substance,” the United States responded that “it’s definitely relevant, I think, for the discrimination prong.” Tr. 72:13–14, 22–23. If a state law targets a market in which the United States is the sole buyer but without applying to the United States, it may be unconstitutional. But if it is, that is because that law discriminates against, not because it directly regulates, the Fed- eral Government. ii. Bans. The majority’s second limitation is that its test applies only to bans, not burdens or taxes. But the difference between a ban and a burden is one of degree, not kind. At oral argument, the United States essentially conceded that this distinction is unstable and declined to provide a framework for how to dis- tinguish a ban from a burden. When asked how we “know when something goes from incidental to impermissible, func- tionally or otherwise,” the Government admitted that “that will be a difficult question in some circumstances for courts to con- sider.” Tr. 73:14–17. When pressed on whether the difficulty

19 of this question is what motivated the plurality in North Dakota to reject a substantial-burden test, the United States “agree[d] there will be difficult line-drawing problems.” Tr. 69:20–21. These cases will present “very, very fact-specific inquiries that [will be] impossible to pre-determine in advance.” Tr. 73:21–23. Rather than supplying a framework, the Govern- ment told us that we will know a ban when we see one: “[W]hatever difficult line-drawing there might be in other cases, this is a clear case where the Supremacy Clause does not permit that kind of obstruction.” Tr. 84:20–22. It cannot be that clear, however, because I see a limitation of discretion, not a ban. See Tr. 86:13–16 (United States acknowledging that “[t]he effects on the [F]ederal [G]overnment and the prohibi- tion to which [it] keep[s] referring” can also be “framed” as “just a curtailing of discretion”). An example illustrates the point. Suppose New Jersey did not stop private entities from entering into immigration- detention contracts but uniformly taxed them for each immi- grant they detain. Is that a ban or a burden? If the tax is low, the majority would probably characterize it as an incidental burden that does not directly regulate the United States. But what if it were so high that doing business in New Jersey be- came impossible for private immigration-detention companies? Then perhaps the majority would say it is effec- tively a ban, and thus a “functional” direct regulation. But at what point does the tax cross the line? The majority not only fails to tell us—it does not even explain how we would go about figuring it out. The plurality in North Dakota anticipated this problem. It acknowledged that “Justice Brennan’s test contains no stand- ard by which ‘burdensomeness’ may be measured.” 495 U.S. at 437 n.8. Determining when a regulation goes from burden to

20 ban implicates thorny questions about federal interests that courts are ill-equipped to resolve without congressional guid- ance. To keep courts out of that mire, the plurality chose to “rely upon our traditional standard of ‘burden’—that specified by Congress and, in its absence, that which exceeds the burden imposed on other comparably situated citizens of the State.” Id. I would follow Justice Stevens and keep courts out of this thicket. iii. Core federal functions. The majority’s final limitation is that immigration de- tention implicates core federal functions. On its telling, the Supreme Court has “implied that courts should treat such laws as regulating the federal government directly.” Maj. Op. 21. But the cases the majority cites do not say that.1 Worse yet, my colleagues fail to mention the cases that hold the opposite. Penn Dairies considered and rejected the argument that a neutral, indirect state law that substantially affects important federal operations nevertheless triggers immunity. Pennsylva- nia’s price control on milk passed constitutional muster because it “impose[d] no prohibition on the national govern- ment or its officers,” only private sellers. Penn Dairies, 318 U.S. at 270. Sure, “[b]y the exercise of control over the seller,

1 As I explain above, New Mexico and Fresno County reject the majority’s test. See supra at 9–10. The other case the majority cites for this proposition is Goodyear Atomic Corp. v. Miller, 486 U.S. 174, 181 (1988). It should come as no surprise that this case is also distinguishable. Ohio tried to apply its work- ers’ compensation safety requirements to “a federally owned facility performing a federal function.” Id. Of course, that state law was a classic direct regulation because it applied by its terms to the Federal Government.

21 the regulation impose[d] or may [have] impose[d] an increased economic burden on the government,” forcing the United States to “procure a supply from without the state.” Id. “But in this burden,” absent congressional action to forbid it, Pennsyl- vania’s law was “no different or greater [an] impairment of federal authority” than a “state regulation of the operations of a trucking company in performing its contracts with the gov- ernment to transport workers employed on a Public Works Administration project,” or “local building regulations applied to a contractor engaged in constructing a postoffice building for the government”—state laws the Court had blessed. Id. (first citing United States v. Baltimore & A.R. Co., 308 U.S. 525 (1939) (mem.) (per curiam); and then citing James Stewart & Co. v. Sadrakula, 309 U.S. 94 (1940)). “The trend” of the Court’s decisions, even then, was “not to extend governmental immunity from state … regulation beyond the national govern- ment itself and governmental functions performed by its officers and agents.” Id. Perhaps my colleagues in the majority believe private immigration detention is more important than transporting workers employed for federal projects or building post offices. Indeed, maybe everyone can agree, at a high level of abstrac- tion, that “immigration” is a core federal function. But we inevitably must draw lines within the immigration domain be- tween core and non-core functions. And that exercise is fraught with political judgments. Is supplying the food and bedding for immigration-detention facilities core? What about deciding the wages paid to workers at those facilities? What about their san- itation standards? Or the taxes they must pay? These questions now matter because the majority has given constitutional sig- nificance to the answers.

22 My colleagues would have us make our own unguided decisions about whether private ownership of immigration de- tention facilities falls on the right side of the line. But Congress, not courts, is supposed to determine which federal functions are sufficiently important that they displace other- wise valid state law. And we properly analyze those cases under the rubric of preemption; we do not contort ourselves to explain how they are somehow direct regulations on the Fed- eral Government.

B. Applying the Correct Test: AB 5207 Does Not Violate Intergovernmental Immunity. Above, I explained the distinction between a direct reg- ulation and a neutral, indirect regulation that nevertheless substantially affects the United States. With the right test in mind, the conclusion is easy: AB 5207, whatever its effects, does not violate intergovernmental immunity.

1. AB 5207 does not directly regulate the Federal Government. CoreCivic argues that AB 5207 is a direct regulation on the United States because it ultimately affects the Federal Gov- ernment’s ability to contract with private parties. Even though the statute applies only to private entities, CoreCivic, like the majority, insists that we should look past form to substance, and focus on the regulation’s functional effect rather than its legal application. But as noted above, a regulation is not direct just because it affects the Federal Government, even signifi- cantly. It must regulate “the performance, by federal officers and agencies, of governmental functions.” Penn Dairies, 318 U.S. at 269 (emphasis added).

23 The Seventh Circuit’s recent decision in McHenry County v. Raoul, 44 F.4th 581 (7th Cir. 2022), is instructive. Illinois prohibited local and municipal governments from en- tering into contracts with the United States to detain immi- grants. Several Illinois counties sued, arguing that the law vio- lated intergovernmental immunity by, among other things, directly regulating the United States. After all, if a county can- not contract with ICE, that means ICE cannot contract with that county. A unanimous panel of the Seventh Circuit dismissed that argument in short order. In its view, “the Illinois Act … impose[d] no direct regulation on any federal official or agency.” Id. at 593. It reached that conclusion even though “a consequence of the Act—the intended consequence of the Act—[wa]s that the federal government will not be able to use cooperative agreements to house immigration detainees in Illi- nois State or county facilities.” Id. (emphasis in original). Even when a state law’s purpose and effect are to burden the Federal Government, that state law “does not directly regulate the [F]ederal [G]overnment” if it merely “appl[ies] non-discrimi- natory regulations to private entities or local governments … that contract with the [G]overnment.” Id. at 593 n.6.

2. AB 5207 does not discriminate against the Federal Government or those with whom it deals. The discrimination prong of intergovernmental immun- ity prohibits states from singling out the Federal Government or its contractors “for less favorable ‘treatment’” or regulating them “unfavorably on some basis related to their governmental ‘status.’” United States v. Washington, 596 U.S. 832, 839 (2022) (first quoting Washington, 460 U.S. at 546; and then quoting North Dakota, 495 U.S. at 438 (plurality)). The

24 District Court did not reach this prong of the analysis. Core- Civic and the United States nevertheless argue that we can affirm on this ground because AB 5207 discriminates against the Federal Government and its contractors in two ways. The challengers first argue that AB 5207 is discrimina- tory because only the United States engages in civil immigration detention. In their view, a restriction on entities that engage in civil immigration detention necessarily targets the Federal Government. This is the majority’s argument that New Jersey targets a market in which the United States is the sole buyer. See supra at 18–19. As noted already, and as the parties all agree, this argument sounds in discrimination. But it fails when analyzed as such because nobody identifies a com- parator treated better. “Differential treatment is critical to a discrimination-based intergovernmental immunity claim.” Ra- oul, 44 F.4th at 594. “The mere fact that [AB 5207] touches on an exclusively federal sphere is not enough to establish dis- crimination.” Id. CoreCivic and the United States also argue that AB 5207 is discriminatory because New Jersey allows private entities to detain certain categories of state prisoners. They note that New Jersey’s Department of Corrections has limited authority to “authorize the confinement of eligible inmates in private facilities.” N.J. Stat. Ann. §§ 30:4-91.9–10. The United States also claims that New Jersey permits counties to “‘con- fin[e] inmates who are in need of and receiving rehabilitative and similar services in private facilities’ operated by for-profit entities.” U.S. Br. 17 (quoting Essex Cnty. Corr. Officers PBA Loc. No. 382 v. Cnty. of Essex, 106 A.3d 1238, 1248 (N.J. Su- per. Ct. App. Div. 2014)). The challengers mischaracterize those laws. Sec- tions 30:4-91.9–10 permit the Department of Corrections to

25 transfer a limited number of low-security prisoners to “non- profit” “residential center[s]”—also known as “halfway houses”—to finish their sentences. App. 110. Another statute, § 30:4-27.2, allows specialized psychiatric facilities to provide involuntary mental-health services. And in Essex County, the New Jersey Superior Court acknowledged that state law ordi- narily forbids private companies from performing the “core governmental function” of “confining … inmates.” 106 A.3d at 1249. Such facilities are permissible only when they serve the “purposes of providing substance abuse, rehabilitative, and similar services to inmates,” and not merely as “alternative jail facilities” for “incarceration.” Id. at 1250. Private immigrant-detention facilities are unlike half- way houses, psychiatric prisons, and special-purpose rehabilitative facilities. CoreCivic does not claim to provide those services. Immigrant detention is more like general pri- vate criminal detention, which New Jersey also prohibits.

* * * I recognize that ICE wants private immigration-deten- tion facilities in New Jersey. AB 5207 will require ICE to spend more money either to buy or lease existing facilities or to build new ones. And I understand the majority’s discomfort with allowing New Jersey to affect federal immigration deten- tion. In fact, I share that discomfort. But our duty as intermediate appellate court judges is to follow the doctrine as it exists, even when it leads to results that make us uncomfort- able. My colleagues disfigure a clear rule—states cannot directly regulate the Federal Government—to alleviate their discomfort. I cannot follow them.

26 III. NO FEDERAL LAW PREEMPTS AB 5207 . The majority does not address preemption. But as I stated at the outset, I believe its substantial-interference argu- ment goes to preemption, not immunity. The preemption doctrine “provides Congress with the power to preempt state legislation if it so intends.” Treasurer of N.J., 684 F.3d at 406 (brackets and internal quotation marks omitted). “There are three types of preemption: express preemption … , field preemption[,] and conflict preemption.” Id. The United States and CoreCivic invoke only conflict and field preemption. Alt- hough it is a closer call, both arguments fail.

A. AB 5207 Is Not Conflict Preempted. Conflict preemption is a kind of implied preemption that displaces a state law when that law “stands as an obstacle to the accomplishment and execution of the full purposes and objectives of Congress.”2 Arizona v. United States, 567 U.S. 387, 399 (2012) (internal quotation marks omitted). A conflict- preemption argument, “like all preemption arguments, must be grounded in the text and structure of the statute at issue.” Kan- sas v. Garcia, 589 U.S. 191, 208 (2020) (internal quotation marks omitted). “Our ultimate task in any pre-emption case is to determine whether state regulation is consistent with the structure and purpose of the [federal] statute as a whole.” Gade v. Nat’l Solid Wastes Mgmt., 505 U.S. 88, 98 (1992) (internal quotation marks omitted). As I previewed above, this is not “a

2 Conflict preemption can also occur if “compliance with both federal and state regulations is a physical impossibility,” Ari- zona v. United States, 567 U.S. 387, 399 (2012) (internal quotation marks omitted), but no one argues that is the case here.

27 freewheeling judicial inquiry into whether a state statute is in tension with federal objectives.” Whiting, 563 U.S. at 607 (in- ternal quotation marks omitted). Rather, “a litigant must point specifically to … a federal statute that does the displacing or conflicts with state law.” Va. Uranium, Inc. v. Warren, 587 U.S. 761, 767 (2019) (internal quotation marks omitted). Under the INA, Congress requires ICE to detain noncit- izens, e.g., 8 U.S.C. § 1231(a)(2), and granted DHS discretion in deciding how to do so, see id. § 1231(g)(1). The Secretary of Homeland Security must “arrange for appropriate places of detention for aliens detained pending removal or a decision on removal.” Id. § 1231(g)(1). The INA also contemplates the possibility that spaces may be “unavailable for rental.” Id. When that is so, the Secretary “may expend … amounts neces- sary to acquire land and to acquire, build, remodel, repair, and operate facilities … necessary for detention.” Id. (emphasis added). Before “initiating any project for the construction of any new detention facility,” the INA directs that ICE “shall consider the availability for purchase or lease of any existing prison, jail, detention center, or other comparable facility suit- able for such use.” Id. § 1231(g)(2) (emphasis added). Regulations implementing the INA provide that ICE may “en- ter into contracts of up to fifteen years’ duration for detention or incarceration space or facilities, including related services.” 48 C.F.R. § 3017.204–90. AB 5207 prohibits New Jersey, local government agen- cies, and “privately owned or operated [detention] facilit[ies]” from “enter[ing] into, renew[ing], or extend[ing] any immigra- tion detention agreement,” defined as any “contract, agreement, intergovernmental service agreement, or memoran- dum of understanding that authorizes the State, local government agency, or private detention facility to house or

28 detain individuals for civil immigration violations.” N.J. Stat. Ann. § 30:4‑8.16. As the United States puts it, this “eliminates the possibility of choice,” requiring the Federal Government to own and operate its own facilities. U.S. Br. 6. Section 1231 does not evince a clear congressional in- tent that ICE be permitted to use private immigration detention. Nor does it contemplate that private facilities must be a part of the detention scheme. And it does not impose a scheme on pri- vate and public detention facilities or indicate the way it wishes the system to work. Rather, it leaves the Secretary with discre- tion to consider and choose from what is available. “[A] high threshold must be met if a state law is to be preempted for con- flicting with the purposes of a federal Act,” Whiting, 563 U.S. at 607 (internal quotation marks omitted), and this case does not meet the threshold set by our precedent. For example, this case is distinguishable from Leslie Miller and Public Utilities Commission. In Leslie Miller, Con- gress instructed that the United States “shall” award the relevant contract to the bidder whose bid “w[ould] be most ad- vantageous to the Government, price and other factors considered.” 352 U.S. at 188 (internal quotation marks omit- ted). Arkansas law imposed on contractors additional requirements for obtaining a license to perform such activity. Id. The Court concluded that this was enough to create a con- flict because it “frustrate[d] the expressed federal policy of selecting the lowest responsible bidder.” Id. at 190. In Public Utilities Commission, Congress had adopted a “comprehensive policy governing procurement” that explained how “the head of an agency” would “negotiate such a purchase or contract.” 355 U.S. at 540–41. Federal law mandated select- ing the least costly means. Id. at 542. Under California law, that discretion could “be exercised and reduced rates used”

29 only if California approved. Id. at 543. The “conflict between the federal policy of negotiated rates and the state policy of regulation of negotiated rates” was “clear.” Id. at 544. In those cases and elsewhere, the “principal indication that Congress in- tended to pre-empt state law” was its use of “shall.” Gade, 505 U.S. at 99. But nothing in the INA requires the Secretary to con- tract with privately operated detention facilities. Even if ICE were to lease a privately owned building, it could operate that building itself. I agree with the Seventh Circuit that the statu- tory language “demonstrates at most a general preference to use existing facilities when they are available.” Raoul, 44 F.4th at 591 (emphasis added).

B. AB 5207 Is Not Field Preempted Either. State law is also preempted “when federal law occupies a field of regulation so comprehensively that it has left no room for supplementary state legislation.” Murphy v. NCAA, 584 U.S. 453, 479 (2018) (internal quotation marks omitted). We can infer congressional intent to displace state law altogether from a framework of regulation “so pervasive that Congress left no room for the States to supplement it or where there is a federal interest so dominant that the federal system will be as- sumed to preclude enforcement of state laws on the same subject.” Arizona, 567 U.S. at 399 (internal quotation marks and alterations omitted). Federal law does not so wholly and exclusively occupy the field of private immigration detention. Certainly “[t]he Government of the United States has broad, undoubted power over the subject of immigration and the status of aliens.” Id. at 394, 400 (applying the preemption analysis to the “field of alien registration”). For example, “[p]olicies pertaining to the

30 entry of aliens and their right to remain here are … entrusted exclusively to Congress.” Galvan v. Press, 347 U.S. 522, 531 (1954). And “[t]he authority to control immigration—to admit or exclude aliens—is vested solely in the Federal Govern- ment.” Truax v. Raich, 239 U.S. 33, 42 (1915). But just because a state law touches on immigration law does not mean field preemption applies. See Ocean Cnty. Bd. of Comm’nrs v. Att’y Gen. of N.J., 8 F.4th 178 (3d Cir. 2021) (finding no preemption of a law barring local law enforcement from assisting federal immigration authorities in certain ways); Lozano v. City of Hazleton, 724 F.3d 297, 304 (3d Cir. 2013) (explaining that in Whiting “the Supreme Court upheld an Arizona statute that al- lowed state courts to suspend or revoke the business licenses of employers who knowingly or intentionally employ unau- thorized aliens”). CoreCivic hardly develops this argument, but it seems to suggest that the field of “immigration detention” is off lim- its. AB 5207, however, says nothing about who can be detained, why, for how long, or even where. It regulates only whether private entities can operate immigration detention fa- cilities in the State. This is not the stuff field preemption is made of.

IV. CONCLUSION AB 5207 applies only to private, local, and state parties, not the United States. And New Jersey does not single out im- migration detention—it prohibits itself from using private general criminal detention too. The majority nonetheless in- sists that AB 5207 violates intergovernmental immunity be- cause it is “functionally” a direct regulation on the United States. But as a matter of both doctrine and plain language, state laws that do not apply to the Federal Government are not

31 direct regulations—they are indirect because it is their down- stream consequences that burden the United States. Case law, basic separation-of-powers principles, and common sense all tell us that Congress, not courts, should decide when those spillover effects are so substantial that they violate the Suprem- acy Clause. And the proper doctrinal method for determining whether Congress intends to displace a neutral indirect state law is preemption, not intergovernmental immunity. Because the majority would have judges rather than legislators make politically sensitive judgments about when a neutral state law that does not apply to the United States interferes too much with federal interests, I respectfully dissent.