UNITED STATES DISTRICT COURT
FOR THE DISTRICT OF COLUMBIA
NATIONAL DIGITAL INCLUSION
ALLIANCE,
Plaintiff,
v. Civil Action No. 25-3606 (JDB)
DONALD TRUMP et al.,
Defendants.
MEMORANDUM OPINION
The National Digital Inclusion Alliance (NDIA) challenges the President’s dissolution of
the Digital Equity Act’s Competitive Grant Program, part of a multibillion dollar initiative to
bring internet connection and skills to underserved parts of the country. It argues that the
President and federal agencies lack authority to annul programs that are authorized and
appropriated by Congress.
Over the past year, plaintiffs across the country have challenged the President’s
termination of various federally funded programs. These plaintiffs argue that the President lacks
statutory and constitutional authority to disregard the laws that authorize those initiatives and
appropriations that fund them. But obtaining relief is not as simple as showing that the President
acted unlawfully. Plaintiffs bringing funding-related claims face numerous jurisdictional and
threshold requirements that courts must consider before reaching the merits. For many, these
cannot be overcome.
This case is different. The government concedes that federal district court jurisdiction
over this case is proper, and that NDIA has standing to bring its claims. It also agrees, at this
stage, that the President lacks the power to cancel laws passed by Congress based on his bald
disagreement with Congress’s policy determinations. Instead, the government contends that the
Competitive Grant Program is itself unconstitutional because the enabling statute authorizes
funds to be allocated based on the race or ethnic status of the people served by grant awardees.
On that issue, the government is correct. The Digital Equity Act straightforwardly
categorizes members of certain races as “covered” by the Act and targets grant money to
programs that benefit those groups. That is an explicit racial classification, which can only be
upheld if it withstands strict scrutiny. Here, it does not: the offending provision is neither
justified by a compelling governmental interest nor narrowly tailored to meet a permissible goal.
Accordingly, it is unconstitutional.
However, the offending provision is severable from the rest of the statute. NDIA’s claim
to restore the Competitive Grant Program, minus the statutory text authorizing the government to
consider the race of grant beneficiaries, thus survives. And because the government’s only
objection to the Digital Equity Act is this singular unconstitutional provision, it has now
committed to restoring the Competitive Grant Program upon receiving this judicial
determination. Defs.’ Suppl. Br. [ECF No. 56] at 10.
BACKGROUND
A. The Digital Equity Act
Millions of Americans lack reliable, high-speed broadband. As a result, they struggle to
access telemedicine, job listings, and other internet-dependent resources. Over the last decade,
these services have become critical to economic and social life. To be excluded from the
2
internet, therefore, is to be denied the fruits of modern society. And this lack of reliable digital
access is felt disparately across the country. Certain groups, such as lower income and rural
Americans, disproportionately struggle to participate in the online ecosystem.
To address this digital divide, Congress passed the Digital Equity Act. Pub. L. No. 117-58, div. F, tit. III, §§ 60301-60307, 135 Stat. 1209 (Nov. 15, 2021), codified at 47 U.S.C.
§§ 1721-1726. The Act appropriated $2.75 billion to “promote digital inclusion activities, and
spur greater adoption of broadband among covered populations.” 47 U.S.C. § 1724(a)(1). Just
like the national projects of previous generations brought running water, roads, and electricity to
those without them, Congress hoped the Digital Equity Act would do the same for the internet. It
was signed into law on November 15, 2021, as part of the $1.2 trillion Infrastructure Investment
and Jobs Act. Pub. L. No. 117-58, 135 Stat. 429-1467 (Nov. 15, 2021).
To expand broadband access to underserved groups, the Act authorized two grant
programs, the Capacity Grant Program and the Competitive Grant Program. 47 U.S.C. §§ 1723-1724. Each initiative received roughly half of the total funds appropriated by the Act. The
former program allocates money to states, while the latter allocates money to individual
applicants, mostly cities and nonprofit entities. The Competitive Grant Program, at issue in this
case, is administered by the National Telecommunications and Information Administration
(NTIA) and the National Institute of Standards and Technology (NIST), both housed within the
Department of Commerce. NTIA reviews grant applications and recommends determinations to
NIST, which makes the awards and oversees compliance. See Compl. [ECF No. 1] ¶¶ 54-59.
NTIA is authorized to review grant applications based on whether the award will
“increase internet access and the adoption of broadband among covered populations to be served
3
by the applicant . . . .” 47 U.S.C. § 1724(d)(1)(A) (emphasis added). The Act defines “covered
populations” as:
(A) individuals who live in covered households;
(B) aging individuals;
(C) incarcerated individuals, other than individuals who are incarcerated in a Federal
correctional facility;
(D) veterans;
(E) individuals with disabilities;
(F) individuals with a language barrier, including individuals who— (i) are English
learners; and (ii) have low levels of literacy;
(G) individuals who are members of a racial or ethnic minority group; and
(H) individuals who primarily reside in a rural area.
Id. § 1721(8). Once awarded, the use of grant money is restricted to six permissible purposes,
five of which relate to increasing covered populations’ internet access, and one catchall
provision. Id. § 1724(d)(2)(A). The Act, however, does not require that any grantee serve any
particular subset of covered populations, exclusively serve covered populations, or serve covered
populations at the expense of other groups. It also does not require that grantees themselves be
members of covered populations and prohibits “discrimination under any program or activity that
is funded” by the Digital Equity Act. Id. § 1726(a)(1).
B. National Digital Inclusion Alliance
NDIA is a nonprofit dedicated to expanding internet access. Compl. ¶ 22. It advocates
for local, state, and federal policies that increase broadband availability and digital literacy. Id.
¶ 53. NDIA also partners with local communities to design and implement programs that help
their residents acquire computer skills. Id. Its flagship program is the Digital Navigator
initiative, which trains staff and volunteers to help community members use online services. Id.
To expand the Digital Navigator initiative, NDIA applied for funding from the
Competitive Grant Program. Id. ¶ 55. It proposed using Digital Equity Act funds to equip
4
thirteen Digital Navigator partnerships with training and resources to better support their
communities. NDIA projected that grant funding would allow it to serve over 30,000 people in
“covered populations” across America. Id. It also asked for money to support its annual
conference, which features discussions between practitioners, academics, and industry members
on policies related to expanding broadband access. Id. ¶ 56. On January 6, 2025, NTIA awarded
NDIA Competitive Grant Program funding. Id. ¶ 58. Later that month, NIST informed NDIA
that it would receive over $25 million, to be disbursed over a five-year period. Id. ¶ 60.
C. The Government Terminates the Competitive Grant Program
NDIA never received the money it was promised. On January 20 and 21, 2025, the
President signed two executive orders that characterized “diversity, equity, and inclusion” as
“discriminatory” and “illegal” and ordered federal agencies to “terminate all discriminatory and
illegal preferences, mandates, policies, programs, activities, guidance, regulations, enforcement
actions, consent orders, and requirements.” Exec. Order 14151, 90 Fed. Reg. 8339 (2025); Exec.
Order 14173, 90 Fed. Reg. 8633 (2025). Pursuant to that guidance, on April 9, NTIA emailed
NDIA that “any costs associated with diversity, equity, and inclusion conferences, trainings,
and/or professional development [would not be] allowable.” Compl. ¶ 13.
The rest of the Competitive Grant Program was soon to follow. On May 8, 2025, the
President announced that he believed the Digital Equity Act was unconstitutional because it
required the allocation of federal funding on the basis of race. Id. ¶ 66. In the same message, the
President said he was “ending” the initiatives authorized by the Act “immediately.” Id. The next
day, NIST issued a formal letter to NDIA explaining that the Competitive Grant Program was
unconstitutional and that the Program and the grants made under it were terminated. Id. ¶ 67.
The NIST letter also explained that the decision was final and that there would be no right of
5
administrative appeal. Id. ¶ 68. NDIA nevertheless attempted to administratively appeal the
termination of its award, which was denied. Id. ¶¶ 69, 70.
When the government declines to enforce a statute on the grounds that the law is
unconstitutional, it must transmit a letter to Congress explaining why. 28 U.S.C. § 530D.
Solicitor General D. John Sauer sent that letter on March 4, 2026, explaining that the Capacity
Grant Program and Competitive Grant Program both required the government to consider the
race of the beneficiaries of the grant recipients. Letter from Solicitor General Sauer to Speaker
Mike Johnson, “Classification of Individuals Based on Race in the Digital Equity Act of 2021,”
U.S. Dep’t of Just. (Mar. 4, 2026), https://www.justice.gov/oip/media/1430066/dl?inline
[https://perma.cc/THA4-QKSB]. He reasoned that under the Supreme Court’s recent decision in
Students for Fair Admissions, Inc. v. President & Fellows of Harvard College, 600 U.S. 181
(2023), such consideration of race in awarding government benefits was unconstitutional.
NDIA filed suit in this Court, challenging the “unilateral repeal by the Executive Branch
of the Digital Equity Competitive Grant Program.” Compl. ¶ 1. It brings claims directly under
the Constitution for violations of the separation of powers and the Spending Clause, as well as
under the Administrative Procedure Act (APA). The government has filed a motion to dismiss
for lack of jurisdiction and failure to state a claim. That motion is now ripe.
STANDARD OF REVIEW
“Federal courts are courts of limited jurisdiction.” Kokkonen v. Guardian Life Ins. Co. of
Am., 511 U.S. 375, 377 (1994). The plaintiff bears the burden of establishing jurisdiction, and
because courts have an affirmative obligation to ensure their jurisdictional authority, “the
plaintiff's factual allegations in the complaint . . . will bear closer scrutiny in resolving a 12(b)(1)
motion than in resolving a 12(b)(6) motion for failure to state a claim.” Grand Lodge of
6
Fraternal Ord. of Police v. Ashcroft, 185 F. Supp. 2d 9, 13-14 (D.D.C. 2001) (citation modified).
If a court determines it lacks jurisdiction, the complaint must be dismissed. Arbaugh v. Y&H
Corp., 546 U.S. 500, 514 (2006); Fed. R. Civ. P. 12(h)(3).
A court must also dismiss a complaint that does not contain sufficient factual matter that,
accepted as true, states a plausible claim to relief. See Ashcroft v. Iqbal, 556 U.S. 662, 678
(2009). When considering a motion to dismiss for failure to state a claim, the court will “assume
[the] veracity” of all “well-pleaded factual allegations,” id. at 679, “construe the complaint in
favor of the plaintiff,” and give the plaintiff “the benefit of all inferences that can be derived
from the facts alleged,” Hettinga v. United States, 677 F.3d 471, 476 (D.C. Cir. 2012). A court
does not, however, credit “formulaic recitation[s] of the elements of a cause of action” or other
conclusory assertions. Bell Atl. Corp. v. Twombly, 550 U.S. 544, 555 (2007).
ANALYSIS
The parties have substantially narrowed their dispute. At the outset, the government
contested federal district court jurisdiction, the availability of a cause of action, and whether
NDIA met various requirements to enable judicial review under the APA. But NDIA has since
clarified that it challenges the cancellation of the Competitive Grant Program, not the termination
of its individual award. As a result, the government concedes that most of those threshold
matters fall away. Although the Court has an independent obligation to consider its jurisdiction,
the parties agree that this case belongs in district court and that NDIA may bring its claims under
the APA’s cause of action. The principal remaining issue is whether the Competitive Grant
Program unconstitutionally requires the government to consider race when allocating grant
money.
7
I. Jurisdiction
The Court begins, as it must, with its jurisdiction to hear the case. NDIA’s claims present
two issues: channeling and standing. Channeling is a question of whether federal district court is
the appropriate forum to hear NDIA’s claims, based on Congress’s division of jurisdiction
between the federal district courts and the Court of Federal Claims. Standing ensures that NDIA
has a bona fide stake in the litigation as necessary to satisfy the Constitution’s requirement that
federal courts only hear “cases” and “controversies.” Both requirements are met here.
A. Channeling
Suits against the United States are “available by grace and not by right.” United States v.
Tohono O’odham Nation, 563 U.S. 307, 317 (2011). Accordingly, they are subject to the consent
Congress provides and the conditions that flow from it. The two relevant forms of congressional
consent here are the Administrative Procedure Act and the Tucker Act. The APA waives
sovereign immunity for suits “stating a claim that an agency or an officer or employee thereof
acted or failed to act in an official capacity or under color of legal authority,” 5 U.S.C. § 702,
while the Tucker Act waives sovereign immunity over government contract disputes.
Megapulse, Inc. v. Lewis, 672 F.2d 959, 963 (D.C. Cir. 1982).
But suits brought under the Tucker Act are subject to the exclusive jurisdiction of the
Court of Federal Claims. Crowley Gov’t Servs., Inc. v. GSA, 38 F.4th 1099, 1106 (D.C. Cir.
2022). By negative implication, the Act forbids federal district courts from hearing most federal
contract disputes. Id. This prohibition extends to contract claims against the United States that
might otherwise be brought in district court under a plain reading of the APA. Id. That is
because allowing district court plaintiffs to plead federal contract claims through the APA would
circumvent “the restrictions of the Tucker Act.” Megapulse, 672 F.2d at 967. Accordingly, to
8
vindicate Congress’s forum-allocation decision to channel most federal contract disputes to the
Court of Federal Claims, district courts must ferret out whether a claim is, in its essence, a
contract action, regardless of how it is labeled.
Sorting claims is a two-step process: courts must first consider “the source of the rights
upon which the plaintiff bases its claims.” Id. at 968. When the plaintiff asserts rights that flow
from the contract itself, the claim is likely contractual. But when the source of rights is
constitutional or statutory, the claim likely belongs in district court. Vera Inst. of Just. v. DOJ,
805 F. Supp. 3d 12, 24 (D.D.C. 2025); Crowley, 38 F.4th at 1106-07. Importantly, the mere
presence of a contract in a dispute does not render a claim contractual, because “[c]ontract issues
may arise in various types of cases where the action itself is not founded on a contract.”
Megapulse, 672 F.2d at 968. For instance, an employee may sue the government in district court
for violation of their constitutional rights in connection with their employment, even though the
employer-employee relationship would not exist but-for a contract. See TransOhio Sav. Bank v.
Dir., Office of Thrift Supervision, 967 F.2d 598, 610-11 (D.C. Cir. 1992); see also NIH v. Am.
Pub. Health Ass’n, 145 S. Ct. 2658, 2660-62 (2025) (Barrett, J., concurring).
At step two, courts consider whether “the type of relief sought” is like actions
traditionally brought in contract. Megapulse, 672 F.2d at 968. The telltale form of relief for a
contract action is specific performance. Ingersoll-Rand Co. v. United States, 780 F.2d 74, 80
(D.C. Cir. 1985). Put another way, if the plaintiff’s claim is ultimately about compelling the
government to satisfy the terms of the initial award, it sounds in contract. Id. By contrast,
requests for injunctive or declaratory relief beyond the four corners of the agreement, unavailable
in contract actions, support federal district court jurisdiction. See Megapulse, 672 F.2d at 969.
9
The Supreme Court recently explained how courts should channel claims related to the
termination of individual federal grants versus those concerning the cancellation of grant
programs. In NIH, Justice Barrett’s controlling concurrence sent the plaintiffs’ grant termination
claims to the Court of Federal Claims and directed the program cancellation ones to district
court. 145 S. Ct. at 2661 (Barrett. J., concurring). She reasoned that while the former claims
were essentially contractual disputes about the government’s failure to satisfy a financial
obligation, the latter challenged a broader agency policy more appropriate for APA review in
federal district court. Id.
NDIA’s claims belong in district court. The substantive rights it asserts are rooted in the
Constitution and the Digital Equity Act, not in its individual grant award. To understand why,
remember that NDIA asks the Court to reinstate the Competitive Grant Program. See Compl.
¶¶ 22, 76. Its asserted entitlement to that relief is based on the theory that “[n]either the
President nor the Executive Branch has any constitutional power to unilaterally repeal duly
enacted statutes,” Compl. ¶ 73, and that the Digital Equity Act itself requires that the
Competitive Grant Program be administered, Compl. ¶¶ 88, 94. Accordingly, NDIA asserts a
substantive right—an entitlement for the Competitive Grant Program to exist and to be
administered according to the terms of the statute—that is entirely separate from the obligations
that the government owes to NDIA under the terms of its award. See NIH, 145 S. Ct. at 2660.
NDIA satisfies step two of the Megapulse test because it does not request specific
performance or any similar relief. It asks the Court to declare the Competitive Grant Program
lawful and vacate the government’s actions to terminate it. It does not seek reinstatement of its
grant award or an order forcing the government to comply with the terms of that award. Compl.
¶ 22; Opp’n to Mot. to Dismiss [ECF No. 38] (“Opp’n”) at 10 (“NDIA does not challenge the
10
termination of its individual award; it seeks the opportunity to compete.”). Indeed, the requested
relief entirely concerns reversing the program cancellation—which is relief that the Court of
Federal Claims has no power to grant.
NDIA’s claims are the spitting image of the program cancellation claims from NIH.
There, the plaintiffs’ program cancellation claims challenged “internal guidance documents” that
established the agency would “not fund research related to DEI objectives, gender identity, or
COVID-19 [or award] grants to researchers based on race.” NIH, 145 S. Ct. at 2661. Here,
NDIA challenges the government’s attempt to cancel a grant program based on a similar policy
disagreement. See, e.g., Compl. ¶¶ 75, 82, 85. That is a challenge to the government’s
management of the grant program, not to any individual grant termination.
In sum, this Court has jurisdiction over NDIA’s claims. NDIA’s claims are not, at
bottom, contract disputes, so district court jurisdiction is proper.
B. Article III Standing
Article III restricts the jurisdiction of federal courts to “Cases” or “Controversies.” Art.
III, § 2, cl. 1. That limitation gives rise to the requirement that every plaintiff must have standing
to sue. See Diamond Alt. Energy, LLC v. EPA, 145 S. Ct. 2121, 2133 (2025). To demonstrate
standing, a plaintiff must show an (1) injury in fact (2) caused by the defendant that is
(3) redressable by the relief requested. See id. The issue here is whether NDIA has suffered an
injury-in-fact that a favorable decision from this Court could redress. In its reply, the
government concedes that NDIA has standing (at the motion to dismiss stage) based on its
inability to apply for future grant funding. Reply [ECF No. 50] at 18.
That is correct. In Global Health Council v. Trump, the D.C. Circuit held that grantees
“may be harmed by [the] denial of the opportunity to compete for a pool of funds for which they
11
are able and willing to compete.” 153 F.4th 1, 12 (D.C. Cir. 2025). In that case, “the grantees
had asserted that the court could redress their injuries by ordering funds to be made available
because the grantees would be eligible to compete for the funds even if not guaranteed to obtain
them.” Id.; see also Ne. Fla. Chapter, Associated Gen. Contractors of Am. v. City of
Jacksonville, 508 U.S. 656, 668 (1993). Crucially, the court explained that reinstating the grant
program provided redressability irrespective of whether individual grants were restored. Glob.
Health, 153 F.4th at 12.
In terms of injury, this case is on all fours with Global Health. NDIA applied for, and has
received, funding from the Competitive Grant Program. It provides the types of services covered
by those awards. And it has expressed a credible intent to apply for grants from the same
program in the future. The inability to apply for that money is a concrete injury that will
meaningfully change NDIA’s operations going forward. For redressability, NDIA only seeks
prospective relief to reinstate the specific grant program that it intends to apply for. So, as with
Global Health, the relief that NDIA requests is appropriately calibrated to remedy the injury it
has suffered. This Court thus concludes that NDIA has an injury-in-fact that is redressable by a
favorable decision from this Court, as needed to confer Article III standing. The government
agrees.
II. Cause of Action
Getting into federal court is only the first step. Before it can make its substantive
arguments, NDIA must identify a cause of action that gives it a private right to sue. It brings
claims under the separation of powers, the Spending Clause, and the APA. The APA provides a
statutory right to sue federal agencies for unlawful action. But the government contends that
12
NDIA lacks a cause of action for its freestanding constitutional claims (Counts I and II), and so
those should be dismissed.
Adversarial briefing has narrowed this issue down to whether the availability of the
APA’s cause of action, which both parties agree NDIA may employ, displaces NDIA’s ability to
bring claims directly under the Spending Clause and the separation of powers. The government
argues that it does, because equitable jurisdiction is typically unavailable when a party has an
adequate remedy at law. See Reisman v. Caplin, 375 U.S. 440, 443 (1964). For its part, NDIA
does not contest this principle, or the notion that the APA would displace the constitutional
claims if it allowed the Court to grant the same relief as the constitutional claims. NDIA only
argues that the APA does not displace NDIA’s constitutional claims because NDIA requests
declaratory and injunctive relief against the President, which is unavailable under the APA.
Franklin v. Massachusetts, 505 U.S. 788, 796 (1992) (plurality opinion) (“[T]he President is not
an agency within the meaning of the [APA].”). So, the question before the Court is whether the
APA offers an adequate remedy at law for NDIA’s claims, despite that limit.
The answer is yes. Although the APA does not enable judicial review of actions taken by
the President, the Court would similarly refrain from entering injunctive or declaratory relief
against him under an equitable cause of action. That is because a “court generally may not
‘enjoin the President in the performance of his official duties.’” Severino v. Biden, 71 F.4th
1038, 1042 (D.C. Cir. 2023) (quoting Franklin, 505 U.S. at 802). And it is unclear whether
courts have the authority to issue declaratory relief against the President either. 1 Instead, courts
typically enjoin subordinate executive branch officials to enforce compliance with the law. See
1
The D.C. Circuit has said that courts have “never submitted the President to declaratory relief.” Newdow v. Roberts, 603 F.3d 1002, 1013 (D.C. Cir. 2010). Although the D.C. Circuit once issued declaratory relief against the President, see Nat’l Treasury Emps. Union v. Nixon, 492 F.2d 587, 616 (D.C. Cir. 1974), it later questioned whether that case remains good law after Franklin. See Swan v. Clinton, 100 F.3d 973, 978 (D.C. Cir. 1996).
13
Severino, 71 F.4th at 1042-43. NDIA does not argue that enjoining the remaining defendants in
this case, who are subject to the APA, would not constitute effective relief. Accordingly, the APA
provides an adequate remedy at law for NDIA’s claims, and thereby displaces equitable
constitutional claims. 2
III. Claim-Specific APA Arguments
The Court now considers the government’s arguments for dismissal based on the merits
of NDIA’s specific APA claims. NDIA challenges the government’s termination of the Digital
Equity Grant program as contrary to constitutional rights (Count III), contrary to law (Count IV),
in excess of statutory authority (Count V), and arbitrary and capricious (Count VI). The
government argues that NDIA’s “contrary to the Constitution” claim fails for the same reasons as
its separation of powers and Spending Clause claim, and that it complied with various grant
termination regulations.
A. Contrary to the Constitution
In two sentences, the government asserts that NDIA’s APA claim that the termination of
the Competitive Grant Program was contrary to the Constitution for violating the separation of
powers and the Spending Clause should fail because the equitable constitutional claims,
discussed above, fail. Mot. to Dismiss (Mot.) [ECF No. 32-1] at 43. It is mistaken.
NDIA cannot bring claims under the separation of powers or the Spending Clause
because Congress, through the APA, has displaced equitable jurisdiction over them. See supra
Part II. But NDIA’s “contrary to Constitution” claim is brought under the statutory cause of
action provided by sections 702 and 706(2)(B) of the APA. Section 702 allows individuals to sue
2
Even if the APA did not supply an adequate remedy, it is unlikely that NDIA’s equitable constitutional claims could proceed. This is not a case where the government has expressly “disclaimed any statutory authority” for its actions, as is necessary for such a claim. Dalton v. Specter, 511 U.S. 462, 473 (1994) (citing Youngstown, 343 U.S. at 585).
14
to challenge unlawful agency action, while section 706(2)(B) explains that such actions include
those that are contrary to the Constitution. The APA thus supplies the missing ingredient from
NDIA’s two constitutional claims—a private right to sue. And indeed, the availability of judicial
review for alleged violations of the Digital Equity Act under the APA is the very reason that
equitable jurisdiction directly under the Constitution for such claims is displaced.
The government only cites a single case to support its argument that dismissal of NDIA’s
constitutional claims merits dismissal of its parallel APA claim. That case reasons that an
“analysis of whether agency action violates the APA because it is contrary to constitutional right
mirrors the analysis of whether the agency action violates the relevant constitutional provision.”
Nat’l Educ. Ass’n v. U.S. Dep’t of Educ., 779 F. Supp. 3d 149, 197 (D.N.H. 2025). The
government misinterprets that decision. The court held that the merits of a plaintiff’s APA
contrary to Constitution claim must reflect whether the underlying substantive constitutional
right was violated. Id.; cf. United States v. District of Columbia, 897 F.2d 1152, 1158 (D.C. Cir.
1990) (finding APA cause of action unnecessary where constitutional claims could be raised
otherwise). For example, an APA contrary to Constitution claim that the government violated a
party’s First Amendment rights would look the same on the merits as a claim brought under the
First Amendment itself. And indeed, in National Education Association, the court found
plaintiffs were likely to succeed on their APA contrary to Constitution claim because that court
had already decided that their underlying constitutional arguments were likely meritorious. 779
F. Supp. 3d at 197.
But whether an agency action violates a party’s substantive constitutional right is separate
from whether that party has a private right of action to sue. In circumstances unlike National
Education Association, where plaintiffs sue under a constitutional provision that does not contain
15
an organic cause of action, APA section 706(2)(B) fills the gap. If the government’s
interpretation were correct, APA section 706(2)(B) would do no work, because it would only
allow parties to sue for constitutional rights that they could already bring to court. Accordingly,
APA section 706(2)(B) is designed precisely for cases like this one, where the plaintiffs allege
violations of a substantive constitutional right but otherwise lack a cause of action through which
to vindicate it. This APA claim thus survives.
B. Compliance with Grant-Termination Laws and Regulations
The government argues that NDIA’s APA contrary to law and excess of statutory
authority claims both fail because it complied with the laws and regulations governing grant
termination when cancelling NDIA’s award. The terminations were procedurally sound, the
argument goes, because the government followed the grant termination provisions in the Office
of Management and Budget’s Uniform Administrative Requirements and in the Digital Equity
Act. See Mot. at 38-41; 2 C.F.R. § 200.340.
Compliance with statutory and regulatory obligations is a defense to an APA claim that
the government acted contrary to or in excess of those requirements. See E. Band of Cherokee
Indians v. U.S. Dep’t of the Interior, 534 F. Supp. 3d 86, 97 (D.D.C. 2021). But those statutes
and regulations must delineate the procedures for the challenged agency action. See Sissel v.
Wormuth, 77 F.4th 941, 947 (D.C. Cir. 2023) (agency action is contrary to law when the agency
fails to “act in accordance with the law governing” the issue in question (emphasis added)).
Compliance with protocols for an entirely separate agency action, by contrast, is not a defense.
Accordingly, the government’s compliance with the legal authorities that govern grant
termination is irrelevant to whether it acted lawfully regarding program cancellation. The
government contends that it complied with two relevant legal authorities: the Office of
16
Management and Budget’s Uniform Administrative Requirements and the grant termination
provisions in the Digital Equity Act. The former is a general regulation, incorporated into the
Competitive Grant Program, that allows for the termination of a grant when “an award no longer
effectuates the program goals or agency priorities.” 2 C.F.R. § 200.340. By its own terms, it
only applies to the termination of individual grant awards. Id. § 200.340(a). The latter is a
provision of the Digital Equity Act that limits the circumstances in which individual grants may
be terminated. Neither of those provisions deal with program cancellation, and the government
does not contend otherwise. Because NDIA’s claims are about program cancellation, compliance
with provisions relating to grant termination is not a basis for dismissal.
IV. Equal Protection
The government contends that all NDIA’s claims should be dismissed because the Digital
Equity Act requires the unconstitutional consideration of race in grantmaking. Because the Act
draws impermissible racial classifications, the argument goes, the government has an obligation
not to enforce its terms. See In re Aiken Cnty., 725 F.3d 255, 261 (D.C. Cir. 2013) (Kavanaugh,
J.) (“[U]ntil a final Court decision in a justiciable case says that a statutory mandate or
prohibition on the Executive Branch is constitutional, the President . . . may decline to follow
that statutory mandate or prohibition if the President concludes that it is unconstitutional.”).
After all, the government cannot be required to carry out an unconstitutional law.
That is half right. The Act authorizes grants to be awarded based on race—which is
impermissible. However, the single unconstitutional condition for allocating grants can be
severed from the remainder of the statute. Accordingly, NDIA’s claim for the reinstatement of
the Competitive Grant Program survives, with the understanding that any relief the Court may
ultimately grant to reinstate the program must exclude the unconstitutional provision.
17
A. Factfinding for the Purpose of Establishing an Administrative Record is
Irrelevant to the Government’s Facial Challenge
At the outset, NDIA contends that the Court should defer resolution of the constitutional
issue to permit discovery into the President’s motives in terminating the program and the
development of an administrative record as to how the Competitive Grant Program was
implemented in deciding grant awards. NDIA asserts that the government’s constitutional
argument is pretext for a policy disagreement that forms the true basis for the cancellation of the
Competitive Grant Program. Further development of the factual record would, it argues, allow
NDIA to prove this. Accordingly, NDIA says, the Court should not consider whether the Digital
Equity Act unconstitutionally discriminates based on race at this stage of the proceedings. But
such factfinding is not necessary for, or relevant to, the resolution of a facial challenge to the text
of the statute.
The government argues that the Digital Equity Act is entirely unconstitutional. If true,
that argument would provide the government an absolute defense against its obligation to enforce
the law. See In re Aiken, 725 F.3d at 261; Smith v. Indiana, 191 U.S. 138, 148 (1903) (“It is but
just to say . . . that the power of a public officer to question the constitutionality of a statute as an
excuse for refusing to enforce it has often been assumed, and sometimes directly decided, to
exist.”). That is because NDIA (and others) cannot receive entitlements that flow from an
unconstitutional statute. A law that conflicts with the Constitution cannot be applied, so this
defense would afford the government a valid basis for dismissal regardless of whether the
President secretly harbored a policy disagreement with the law. Pretext may be a valid reason
for courts to look beyond a purported legitimate justification to ferret out animus, see, e.g., Dep’t
of Com. v. New York, 588 U.S. 752, 780-85 (2019), but it is not a reason to fail to consider
independent grounds upon which a statute is unlawful.
18
NDIA also argues that the Court should allow discovery into the administrative record of
how the Department of Commerce awarded grants. Pls.’s Supp. Mem. [ECF No. 55] at 1. It
believes that such discovery would demonstrate that, in practice, the Department did not racially
discriminate when awarding funds. The government counters that the Department of
Commerce’s post-enactment agency practice is irrelevant to the question of whether the Digital
Equity Act is facially unconstitutional. Here too, the government is correct.
The facial validity of an explicit “racial classification” is a “legal question,” not a factual
one. Lynch v. Donnelly, 465 U.S. 668, 694 (1984) (O’Connor, J., concurring) (citation
modified); see also Beach Commc’ns, Inc. v. FCC, 959 F.2d 975, 977 (D.C. Cir. 1992) (noting
that an equal protection challenge raises a “purely legal” issue). A party lodging a facial
challenge need not delve into any “specific fact situation” created by implementation. Rust v.
Sullivan, 500 U.S. 173, 195 (1991). Instead, that party must show that the “institution that
ma[de] the racial distinction”—here, Congress—created a statutory scheme that irreconcilably
conflicts with a constitutional guarantee. Louisiana v. Callais, 146 S. Ct. 1131, 1153 (2026).
NDIA’s confusion arises from its misunderstanding of the differences between various
types of equal protection claims. Some of the cases NDIA cites address the need for an
administrative record in an as-applied challenge. See, e.g., Pl.’s Supp. Mem. at 5 (citing Nat’l
Urb. League v. Trump, 783 F. Supp. 3d 61, 90 (D.D.C. 2025) (remarking that factfinding was
necessary for an as-applied First Amendment challenge). But in such a challenge, the plaintiff’s
claim is that a law cannot be leveraged against them, based on their circumstances. Factfinding
is inherently necessary to review such a claim, because the court must determine whether the
circumstances in question are protected by the Constitution. NDIA also confuses facial
challenges with a discriminatory-purpose challenge to a facially-neutral law. See Arlington
19
Heights v. Metro. Hous. Dev. Corp., 429 U.S. 252, 256 (1977). To bring such a challenge, the
plaintiff must build a record that demonstrates that the legislature was motivated by a hidden
animus. But no such record is necessary in a case like this, where the racial classification
appears on the face of the statute and the government argues that the law is unconstitutional in all
applications. See Shaw v. Reno, 509 U.S. 630, 642 (1993) (“[N]o inquiry into legislative
purpose is necessary when the racial classification appears on the face of the statute.”). 3
And finally, there is no basis to conclude that the record will be meaningfully developed
through discovery. Although this case is at the motion to dismiss stage, the government does not
contest NDIA’s factual arguments about how grants are awarded in practice, which must be taken
as true. Cf. Brown v. District of Columbia, 390 F. Supp. 3d 114, 126-27 (D.D.C. 2019)
(declining to evaluate a constitutional claim at the motion to dismiss stage where doing so would
require the court to “ignore, or contradict, [the] complaint’s factual allegations”); see also Roman
Cath. Diocese of Brooklyn v. Cuomo, 592 U.S. 14, 17-19 (2020) (finding restrictions on public
gathering likely fail strict scrutiny at the temporary restraining order posture). 4 The Court asked
the parties to file supplemental memoranda on whether further factual development was
necessary. Notably, in that brief, NDIA did not suggest that any further development of the
congressional record is necessary to enable this Court to apply strict scrutiny, or that any of the
3
NDIA points out, correctly, that the government cannot succeed on its facial challenge unless it shows that “no set of circumstances exists under which the Act would be valid.” United States v. Salerno, 481 U.S. 739, 745 (1987). To be sure, that is a high burden. But here, Salerno stands for the opposite of what NDIA wants. The Supreme Court’s high standard for a facial challenge requires courts to consider whether any judicially cognizable circumstances exist where the law would be valid—not whether the Executive Branch’s implementation of a law thus far has produced any particular results. In Salerno itself, the Supreme Court considered the constitutionality of the Bail Reform Act based on the sweep of the law, not the government’s record in applying it. Id. at 745-55.
4
Louisiana v. Callais, 146 S. Ct. 1131 (2026), which NDIA cites to suggest that discovery is necessary prior to applying strict scrutiny, revolved around proving whether Louisiana had engaged in unjustified race-based redistricting. Here, the race-based preference is apparent from the statutory terms themselves and there is no dispute about how grants are awarded in practice.
20
relevant statutory terms are unclear. 5 In such circumstances, courts are fully competent to
address constitutional arguments, including in cases that NDIA cites for the opposite proposition.
See Jeffery v. City of New York, 113 F.4th 176, 188 (2d Cir. 2024). For example, in Jeffrey, the
Second Circuit subjected New York City’s COVID-19 curfew law to strict scrutiny based on the
text of the law, “the pleadings and judicially noticeable facts.” Id. at 194. It is thus appropriate
for this Court to address the constitutionality of the Digital Equity Act now, before any
discovery.
B. The Digital Equity Act Unlawfully Authorizes the Disbursement of Grants Based
on Race
The Fifth Amendment prohibits the federal government from depriving any person of
life, liberty, or property, without the due process of law. U.S. Const. amend. V. That guarantee
encompasses the protections of the Fourteenth Amendment’s Equal Protection Clause and
applies them to the federal government. See Bolling v. Sharpe, 347 U.S. 497, 498-99 (1954). In
so doing, the Fifth Amendment prohibits the federal government from engaging in “all invidious
racial discriminations.” Loving v. Virginia, 388 U.S. 1, 8 (1967). That is because legal
distinctions based on race are “odious to a free people whose institutions are founded upon the
doctrine of equality.” Id. at 11 (quotation omitted).
Federal laws that classify based on race are thus subject to strict scrutiny. Adarand
Constructors, Inc. v. Pena, 515 U.S. 200, 224 (1995). Strict scrutiny is “the most exacting” form
5
In the Second Amendment context, for example, courts may be required to consider facts about whether a firearm named in a statute presents similar dangers to firearms the Supreme Court has held may be prohibited, such as machine guns and sawed-off shotguns. See, e.g., Bianchi v. Brown, 111 F.4th 438, 453 (4th Cir. 2024), cert. denied sub nom. Snope v. Brown, 145 S. Ct. 1534 (2025); Barnett v. Raoul, No. 24-3060, 2026 WL 1982951, at *6 (7th Cir. July 9, 2026); cf. Wolford v. Lopez, No. 24-1046, 2026 WL 1825723, at *17 (U.S. June 25, 2026) (Barrett, J., concurring) (explaining that the constitutionality of a firearm regulation may turn on an examination of whether the regulation targets a “specific, heightened risk” of firearm misuse). In such circumstances, factfinding to clarify the meaning of ambiguous terms in the law itself may be necessary for judicial review, which would otherwise be based on speculation. But here, the government argues that the Digital Equity Act is an explicit, facial race classification, and neither party argues that the Act contains any undefined terms. Such a challenge is appropriately resolved based on the statute itself.
21
of judicial review. Id. at 218 (quotation omitted). To satisfy it, the law must be necessary to
“serve a compelling governmental interest.” Id. at 235 (quotation omitted). And Congress must
also “clearly articulate[] the need and basis” for the racial classification. Id. at 229 (quotation
omitted). That requirement is rarely met—the Supreme Court has only identified two objectives
sufficiently persuasive to permit racial sorting: remediating specific instances of past unlawful
discrimination and avoiding imminent violence in prisons. 6 Students for Fair Admissions, 600
U.S. at 207.
Second, the law must be “narrowly tailored to further that interest.” Adarand, 515 U.S at
235. In practice, “narrowly tailored” means “necessary,” Fisher v. Univ. of Tex. at Austin, 570
U.S. 297, 311-12 (2013), because “[r]acial classifications are simply too pernicious to permit any
but the most exact connection between justification and classification,” Gratz v. Bollinger, 539
U.S. 244, 270 (2003) (quotation omitted). Any asserted racial categories must bear a tight
connection to the means they employ and the goals they pursue without being overinclusive or
underinclusive. See Students for Fair Admissions, 600 U.S. at 215-17 (finding racial
classifications used in college admissions both overinclusive—for grouping together students of
many backgrounds—and underinclusive—for not providing categories that apply to certain
students). The government must also have considered whether race-neutral means could have
achieved the same objectives and must have designed the program to be limited such that it “will
not last longer than the discriminatory effects it is designed to eliminate.” Adarand, 515 U.S. at
237-38. (quotation omitted).
6
Prior to Students for Fair Admissions, the Supreme Court had identified academic diversity in higher education as a compelling governmental interest. See Regents of Univ. of Calif. v. Bakke, 438 U.S. 265, 311-12, 314 (1978). But Students for Fair Admissions overturned that assessment and held that diversity does not qualify as a compelling governmental interest. 600 U.S. at 214-15.
22
Strict scrutiny applies equally to laws that ostensibly benefit the target racial group.
Adarand, 515 U.S. at 223. It also “does not change merely because the challenged classification
operates against a group that historically has not been subject to governmental discrimination.”
Wygant v. Jackson Bd. of Educ., 476 U.S. 267, 273 (1986); see also Yick Wo v. Hopkins, 118
U.S. 356, 368-69 (1886) (holding that the Equal Protection Clause applies “without regard to any
differences of race” and is “universal in [its] application”). This principle, known as equal
protection “consistency,” Adarand, 515 U.S. at 224, flows from the notion that racial
“distinctions impinge upon personal rights,” and so entitle individuals to “a judicial
determination that the burden he is asked to bear on that basis” satisfies strict scrutiny. Bakke,
438 U.S. at 299.
Recall that the Digital Equity Act established the Competitive Grant Program to “increase
internet access and the adoption of broadband among covered populations . . . .” 47 U.S.C.
§ 1724(d)(1)(A). The Act defines “covered populations” by a list of eight groups, including
“individuals who are members of a racial or ethnic minority group.” Id. § 1721(8). In
determining whether to issue an award, the Assistant Secretary of NTIA must consider whether
the grantee’s proposal will increase broadband adoption among covered populations. Id.
§ 1724(d)(1). Funding is further contingent on the grantees’ performing one of five specified
activities for the benefit of covered populations—such as “providing hardware and software . . .
to covered populations”—or satisfying a residual clause. Id. § 1724(d)(2); cf. Fischer v. United
States, 603 U.S. 480, 487 (2024) (explaining that residual clauses should be read to be consistent
with the preceding terms).
The government argues that the Digital Equity Act impermissibly classifies based on
race. In its view, the statute sorts individuals into “covered populations” based on their racial
23
status and requires that the government prefer benefitting those groups when awarding funds.
The government further argues that such race-based preferences fail strict scrutiny because
NDIA does not identify a compelling interest to justify racial classifications and the statutory
scheme is not narrowly tailored.
The government is correct. The provisions in the Digital Equity Act that authorize the
consideration of racial status constitute an explicit racial classification. Those provisions
straightforwardly authorize NTIA to consider the grant beneficiary’s racial group status when
awarding grants. See Rothe Dev., Inc. v. DOD, 836 F.3d 57, 62, 63-64 (D.C. Cir. 2016)
(explaining that a regulation that designated certain racial groups for funding preferences was a
racial classification). After all, the Act categorizes members of certain races—“racial . . .
minority group[s]”—as “covered populations” and instructs the government to award grants
based on whether the application, if approved, would improve broadband access among those
groups. The Act also conditions grantees’ eligibility on performing a limited set of activities,
including activities for the specific (albeit not exclusive) benefit of covered populations. And the
fact that the law focuses on the race of the grant beneficiary as opposed to the grantee, or exists
for the benefit of those groups, changes nothing. See Adarand, 515 U.S. at 205, 213 (federal
incentive for contractors to hire subcontractors based on race was a racial classification). At this
stage, what matters is that the Digital Equity Act empowers the government to consider race
when allocating federal money.
NDIA resists this conclusion by claiming that the Digital Equity Act is a “race-conscious
policy,” not a “racial classification.” Opp’n at 33. In support, it cites this Circuit’s decision in
Rothe, which held that “[p]olicymakers may act with an awareness of race . . . without thereby
subjecting the resultant policies to the rigors of strict constitutional scrutiny.” 836 F.3d at 72. Of
24
course, Rothe predates the Supreme Court’s decision in Students for Fair Admissions. But Rothe
cuts against NDIA’s position anyway. In that case, the circuit held that a law that provided grants
to “socially and economically disadvantaged” individuals was not a racial classification because
that requirement used “race-neutral terms.” Id. at 61-62. But that court also concluded that the
agency’s implementing regulation that designated individuals of minority racial groups as
presumptively included within the scope of the statute did constitute a racial classification. Id.
at 62. The lesson from Rothe is that legislators may write laws that address broad-based social
inequities with the understanding that those benefits may largely flow to disadvantaged racial
groups without triggering strict scrutiny, but if the law explicitly targets certain racial groups,
strict scrutiny applies. See Callais, 146 S. Ct. at 1146 (explaining that in equal protection
challenges, “if race play[s] a role in a decision made by a government actor, strict scrutiny
applie[s]”). And the Digital Equity Act is more like the implementing regulation than the statute
in Rothe because it explicitly directs the government to consider racial status in awarding grants.
The Act does not escape scrutiny by authorizing the consideration of criteria aside from
race. Because grants must benefit “covered populations” generally, it is entirely possible for
individual awards to be made without race being dispositive—for example, to organizations that
primarily benefit rural populations. 7 But in Students for Fair Admissions, the Supreme Court
rejected the notion that “race is not a negative factor” in college admissions simply “because it
does not impact many [individual] decisions.” 600 U.S. at 219. The Court held that heightened
review is appropriate when laws allow government actors to “intentionally allocate preference”
7
NDIA also argues that the Digital Equity Act does not classify on the basis of race because it merely authorizes, as opposed to requires, the government to consider the race of grant beneficiaries. Opp’n at 35. That is incorrect. Although the statute gives the government discretion in weighing grant benefits to various covered populations, it requires that covered population status be considered. See 47 U.S.C. § 1724(d)(1) (providing that NTIA “shall” consider the impact of grants on covered populations “to the extent practicable”). And “even when an agency has significant discretion in deciding how much weight to accord each statutory factor, that does not mean it is free to ignore any individual factor entirely.” Carlson v. Postal Regul. Comm’n, 938 F.3d 337, 344 (D.C. Cir. 2019) (citation modified).
25
between racial groups. Id. at 220. Even when the ultimate impact on decisionmaking may be
minor, the Court concluded that racial classifications reflect impermissible racial stereotypes and
erroneously suggest that it is appropriate for the government to treat people differently because
of their skin color. Id. The Digital Equity Act unquestionably allows NTIA to privilege grant
applications that purport to benefit certain racial groups.
NDIA also argues that the Act does not contain a racial classification because it requires
the government to consider the population-level demographics of grant beneficiaries when
allocating money, as opposed to making racial classifications on a person-by-person basis. To
NDIA, “benefits to the individual are not allocated based on race” because even grants that
primarily support racial minorities may benefit non-minorities who access the same programs.
Tr. [ECF No. 56-1] 36:23-37:6. But NDIA concedes that the “covered population” framework
enables the government to “consider and include” certain racial groups as target beneficiaries
(albeit not specifically excluding others). Tr. 36:16-22. If so, the provision must be a racial
classification. NDIA offers no principled reason as to why the Equal Protection Clause would
allow the government to explicitly use race to steer federal funding at the population-level.
Quite the opposite—“laws that explicitly distinguish between individuals on racial grounds”
receive strict scrutiny, even when those distinctions are aggregated and the statutory benefit
flows to those individuals indirectly, such as in the redistricting context. See Shaw, 509 U.S. at
642 (redistricting context). For example, states could not escape equal protection scrutiny by
allowing officials in charge of agricultural grants to prefer applications from majority-white
municipalities, a population-level distinction, merely because the municipalities do not formally
exclude minorities. There, as here, equal protection forbids the explicit use of race data to
26
structure political benefits, even though no individual is awarded personal favor based on their
race. Accordingly, this Court will apply strict scrutiny.
The challenged provision fails strict scrutiny at the outset. Congress identified no
governmental interest that specifically justifies racial classifications either in the statute itself or
in the legislative record. See Wisc. Legislature v. Wisc. Elec. Comm’n, 595 U.S. 398, 404
(2022) (racial classifications must be justified “at the time of imposition”). NDIA gestures
towards unrelated legislative hearings and interviews to suggest that Congress considered the
digital divide to be a civil rights issue, Opp’n at 37, but none concern the Digital Equity Act or
endorse the idea that a racial classification is necessary as part of a remedial scheme. Certainly,
there is nothing in the Digital Equity Act like laws where Congress has affirmatively justified the
need for classifications. For instance, in the Surface Transportation Reauthorization Act, passed
in the same legislative package as the Digital Equity Act, Congress “reviewed testimony and
documentation of race and gender discrimination” in surface transportation markets and
determined that “race- and gender-neutral efforts alone are insufficient to address the problem,”
thereby warranting preferential treatment for certain groups. Pub. L. No. 117-58, div. A, tit. I,
§ 11101(e), 135 Stat. 443, 448-49 (2021). The lack of such a finding here is fatal, because
Congress must “clearly articulate” the need and basis for a racial classification at its inception.
Adarand, 515 U.S. at 229. 8
At most, the Act includes a statement of purpose that lack of broadband “exacerbates
existing wealth and income gaps, especially those experienced by covered populations.” 47
U.S.C. § 1722. NDIA nominally contends that the Act “remedies specific instances of past
8
NDIA briefly argues that the COVID-19 pandemic made assembling a legislative record difficult. Opp’n at 38. That may be true. But the Supreme Court has specifically rejected the notion that the pandemic reduced the government’s burden under strict scrutiny to articulate and demonstrate a compelling government interest and employ the least restrictive method of achieving that goal. See Roman Cath. Diocese, 592 U.S. at 19-20 (2020) (“But even in a pandemic, the Constitution cannot be put away and forgotten.”).
27
discrimination,” but its argument is that the law is justified by facially neutral events like the
COVID-19 pandemic, which disparately impacted certain communities which lack internet
access. Opp’n at 37. Its stance boils down to the idea that the “digital divide disproportionately
affects communities of color.” Id. (citation modified). 9 Addressing that gap is a laudable goal,
but the Supreme Court has admonished that ameliorating general societal inequalities—as
opposed to specific instances of past discrimination—“does not constitute a compelling interest
that justifies race-based state action.” Students for Fair Admissions, 600 U.S. at 226. Otherwise,
Congress could deploy racial classifications when confronted with any situation of an uneven
resource distribution. Shaw v. Hunt, 517 U.S. 899, 909-10 (1996) (“[A]n effort to alleviate the
effects of societal discrimination is not a compelling interest.”). Accordingly, Congress has not
supplied a compelling governmental interest for the racial classification provision of the Digital
Equity Act.
Nor is the provision narrowly tailored. It “is almost impossible to assess whether the
[racial criterion] is narrowly tailored to remedy prior discrimination since it is not linked to
identified discrimination in any way.” City of Richmond v. J.A. Croson Co., 488 U.S. 469, 507
(1989). As discussed above, nothing in the statute, its legislative history, or the record points to
specific instances of past discrimination that the Digital Equity Act purports to remedy.
To the extent that a legislative intent to address specific instances of past discrimination
in broadband access is ascertainable, the classification in question is ill fit for that purpose. The
Act allows the government to provide a blanket advantage for certain racial groups, regardless of
whether the applicant actually suffers from lack of broadband access. That is substantially
9
Amicus Benton Institute, in support of NDIA, similarly argues that the purpose of the Act is to address systemic, population-level disparities in internet access, as opposed to specific instances of past discrimination. Benton Br. [ECF No. 42-1] at 8-10.
28
overinclusive, because it affords certain beneficiaries “an absolute preference over other citizens
based solely on their race,” many of whom may not have “suffered the effects of prior
discrimination.” Id. at 508. Nor is there any indication that Congress considered whether the
Competitive Grant Program could have achieved similar outcomes using only race-neutral
selection criteria. Race-neutral policies may achieve some of the same outcomes as race
conscious policies because universal barriers can be felt more acutely by members of
disadvantaged groups. Id. at 509-10. This Court will not speculate as to the impact of such an
approach here, but there is no evidence that Congress engaged in the “good faith consideration of
workable race-neutral alternatives” required to satisfy strict scrutiny. Grutter, 539 U.S. at 339.
The Digital Equity Act contains a provision that authorizes the use of race to award
federal funds. That provision is neither justified by a compelling governmental interest nor
narrowly tailored to advance any such interest. Accordingly, under current Supreme Court case
law, particularly Students for Fair Admissions, it fails strict scrutiny and is unconstitutional.
C. The Provision is Severable
The Court has found that the Digital Equity Act is unconstitutional insofar as it authorizes
the allocation of federal money based on a racial classification. But that does not mean NDIA’s
complaint must be dismissed. Dismissal is only warranted if the identified infirmity renders the
complaint incapable of “stat[ing] a claim upon which relief can be granted.” Fed. R. Civ.
P. 12(b)(6). NDIA’s claim—that the government is obligated not to terminate the Competitive
Grant Program—survives because the unconstitutional aspect of the Digital Equity Act is
severable from the remainder of the statute. If the unconstitutional component can be severed,
then declaratory or injunctive relief, should NDIA prevail, could restore the Competitive Grant
29
Program minus any unconstitutional racial classification within the Digital Equity Act.
Accordingly, dismissal would be inappropriate.
A law that conflicts with the Constitution is invalid. U.S. Const. art. VI, cl. 2. But the
Constitution only displaces laws to the extent that the two actually collide—and by necessary
implication, statutory provisions that are harmonious with the Constitution survive. See Barr v.
Am. Ass’n of Pol. Consultants, Inc., 591 U.S. 610, 625 (2020); William Baude, Severability First
Principles, 109 Va. L. Rev. 1, 8 (2023). Accordingly, courts that find a statutory provision
unconstitutional must strike down only the invalid aspects of the law and leave the remainder
intact where possible. Nat. Res. Def. Council v. Wheeler, 955 F.3d 68, 81 (D.C. Cir. 2020). 10
Statutes are presumed to be severable because, even without an explicit severability
clause, “[t]he unconstitutionality of a part of an Act does not necessarily defeat or affect the
validity of its remaining provisions.” Champlin Refining Co. v. Corp. Comm’n of Okla., 286
U.S. 210, 234 (1932). And severability manifests respect for Congress by preserving the fruits of
the political process where possible. Am. Ass’n of Pol. Consultants, 591 U.S. at 626-27. It also
prevents litigants from using a discrete constitutional infirmity to take down an entire law. Id.
Courts determine whether a statutory provision is severable by considering whether the
remainder is capable of functioning independently and whether the absence of the offending
10
Nothing about the atypical posture of this case changes the Court’s obligation to conduct a severability analysis. Ordinarily, challenges to the constitutionality of laws are raised by the plaintiff, seeking to enjoin an allegedly unlawful statutory provision. But here, the government raises its constitutional argument as, essentially, a defense against its obligation to enforce the Digital Equity Act. Neither party asserts that this posture should alter the Court’s obligation to consider severing the unconstitutional aspects of the Act rather than setting it aside altogether. Despite the unusual posture of this case, then, the Court is bound by the same obligation to set aside legislation only to the extent it conflicts with the Constitution. See, e.g., Zivotofsky ex rel. Zivotofsky v. Kerry, 576 U.S. 1, 29-32 (2015) (invalidating a single provision of the Foreign Relations Authorization Act that the President declined to enforce based on constitutional concerns). Moreover, it is not clear why the unconstitutional race criterion allows the government to suspend the entire Digital Equity Act in the first instance, prior to securing judicial review. If the Constitution requires courts to consider the degree to which an unconstitutional provision infects the remainder of the law, and to fashion relief accordingly, the same obligation likely applies to the executive branch.
30
provisions substantially undermines Congress’s design. Id. (citing Seila Law v. CFPB, 591 U.S.
197, 235 (2020)). Here, both considerations strongly support severability.
As an initial matter, the Digital Equity Act is perfectly capable of functioning
independently without the racial classification. The Act defines a set of eight “covered
populations,” 47 U.S.C. § 1721(8), for whom it directs funds to be allocated, see id.
§§ 1724(a)(1), (d)(1)(A)(i), (d)(2)(A). Only one of those eight is racial or ethnic status. The
others are low income Americans, “aging individuals,” “incarcerated individuals,” “veterans,”
“individuals with disabilities,” “individuals with a language barrier,” and individuals who live in
rural areas. 47 U.S.C. § 1721(8). Even without the racial criterion, grant money can be allocated
among the remaining groups. The Act requires consideration of covered population status
generally and does not mandate that money go to any particular subgroup. That means that the
Competitive Grant Program can be administered—consistent with the language of the Digital
Equity Act—exactly as it was before, minus consideration of racial or ethnic status.
Nor does severability undermine the Digital Equity Act’s purpose. The Act was passed to
ensure that “all people of the United States” have “the information technology capacity that is
needed for full participation in the society and economy.” 47 U.S.C. §§ 1721(10), 1722(3). And
congressional proponents of the Act foresaw that funding would support “communities that have
not yet secured the skills, technologies, and support needed to take advantage of broadband
connection,” not just particular racial groups. 167 Cong. Rec. S5684 (daily ed. Aug. 3, 2021)
(statement of Sen. Collins). The Act also includes a separate provision that prohibits
discrimination on the basis of race within any funded program. See 47 U.S.C. § 1726. This
suggests that the Act, at its core, does not require unconstitutional race discrimination in order to
31
carry out its design. Common sense dictates that Congress would prefer to have a version of the
Digital Equity Act without the racial criterion than to have no law at all.
Precedent also supports severability here. In equal protection challenges, “the Court
typically severs the discriminatory exception or classification, and thereby extends the relevant
statutory benefits or burdens to those previously exempted, rather than nullifying the benefits or
burdens for all.” Am. Ass’n of Pol. Consultants, 591 U.S. at 632. Nullification is particularly
inappropriate for federal programs that allocate funding, because it “would impose hardship on
beneficiaries whom Congress plainly meant to protect.” Califano v. Westcott, 443 U.S. 76, 90
(1979). That is certainly the case here, where nullification would deprive all Americans who
lack reliable broadband of access to funds. And, based on the briefing, the government has no
apparent objection to distributing these grants aside from the provision that authorizes allocations
based on race or ethnicity. Defs.’ Suppl. Br. at 10; see also U.S. Dep’t of Just., Letter from
Solicitor Geneal Sauer to Speaker Johnson, “Classification of Individuals Based on Race in the
Digital Equity Act of 2021” (Mar. 4, 2026), www.justice.gov/oip/media/1430066/dl?inline.
(setting out race criterion as sole objection to carrying out the Act).
To summarize, the provision of the Digital Equity Act that authorizes the consideration of
race or ethnicity in grantmaking is severable from the remainder of the statute. That means
NDIA may still claim that it is entitled to apply for benefits that flow from the Act, minus
consideration of the provision of the Act authorizing funding based on race or ethnicity.
V. APA Reimbursement Claim
NDIA also brings a claim challenging the government’s policy—articulated in an April 9,
2021, email—prohibiting grantees from requesting reimbursements for costs incurred from
conferences, training, and professional development programs associated with diversity, equity,
32
and inclusion efforts. NDIA challenges that policy on a prospective basis, because it will
“continue to apply to any Competitive Grant Program recipient.” Pl.’s Suppl. Br. at 10. It brings
this claim separate from its challenge to the termination of the Competitive Grant Program. For
its part, the government argues that NDIA lacks standing to bring the claim. Reply at 24.
In any event, courts have an “independent obligation” to ensure “that jurisdiction is
proper.” Inst. Shareholder Servs., Inc. v. SEC, 142 F.4th 757, 763 (D.C. Cir. 2025) (quotation
omitted). And in this Court’s view, the issue is more easily conceptualized as one of ripeness,
although standing and ripeness overlap. See Trump v. New York, 592 U.S. 125, 131 (2020)
(explaining that standing and ripeness are “[t]wo related doctrines of justiciability—each
originating in the case-or-controversy requirement of Article III”); cf. Indus. Energy Consumers
of Am. v. FERC, 125 F.4th 1156, 1163-65 (D.C. Cir. 2025) (Henderson, J., concurring) (arguing
that constitutional ripeness has been subsumed under imminent injury in fact component of
standing). The ripeness doctrine requires that courts consider whether an issue is “dependent on
contingent future events that may not occur as anticipated, or indeed may not occur at all.’”
Trump, 592 U.S. at 131 (quotation omitted). A claim may be unripe, for example, if it relies on a
party’s speculation about how the Executive Branch will implement a general statement of
policy. Id.
This issue is not presently ripe. Even setting aside this decision, it is not clear how the
government intended to apply its new policy—for example, what kinds of programming it would
deem impermissibly associated with diversity, equity, and inclusion issues. Now, the
government plans to imminently reinstate the Competitive Grant Program and must fashion new
guidelines for grant applicants on a prospective basis. It is thus even less clear whether this (or
any) policies that governed the previous iteration of the Competitive Grant Program will remain,
33
and in what form. By necessary implication, it is premature to say how the policy articulated in
the April 9 email will apply to future applicants or even if it will apply at all. The Court cannot
review the reimbursement policy articulated in the April 9 email, which applied to the nowterminated Competitive Grant Program, on speculation about how the policy will be applied or
even that the same policy will apply to the renewed program. And of course, the policy has not
yet been applied to any future application by NDIA.
Ripeness has practical benefits as well. With its premature claims dismissed, NDIA may
wait to litigate on the basis of the actual terms of the renewed Competitive Grant Program that
they will be subject to. And it will benefit the government by allowing it a chance to clarify
whether the April 9 email is an accurate description of a continuing agency policy and how it
would be implemented.
Because this claim is unripe, the Court lacks jurisdiction and it will be dismissed.
CONCLUSION
NDIA survives dismissal. Its claims challenging the termination of the Competitive
Grant Program belong in federal court because they are rooted in the Digital Equity Act and the
Constitution, not its individual grant award. NDIA has standing to sue based on its interest in
competing for Competitive Grant Program funding, and the APA provides a private right of
action to vindicate those rights. Finally, although a provision of the Digital Equity Act is
unconstitutional, that infirmity does not require the Court to disregard the still-valid remainder.
What next? NDIA will have the opportunity to press its APA claims challenging the
cancellation of the Competitive Grant Program. But the government has committed to
reinstating the Competitive Grant Program upon a judicial determination that the race criterion is
unconstitutional and severable. When it does so, the Court must consider whether some or all of
34
NDIA’s claims are moot. Until then, this case will continue along the ordinary track of civil
litigation.
A separate order will issue on this day.
/s/
JOHN D. BATES
United States District Judge
Dated: July 15, 2026
35