9th Circuit's DEI Grant Ruling Creates a Two-Track System

Thakur v. Trump is easy to misread if the headline is doing too much work. On May 26, 2026, the Ninth Circuit kept roughly $324 million in EPA, NSF, and NEH research grants from being terminated under anti-DEI and environmental-justice directives, but it did not give grant recipients a single clean route back to federal funding.[1] It gave them two tracks. On one track, a grant cancellation based on perceived DEI, DEIA, or environmental-justice viewpoints likely violates the First Amendment. On the other, a claim that is really seeking restoration of grant money may belong in the Court of Federal Claims, not in district court under the Administrative Procedure Act.

That is the core legal point in the Ninth Circuit's DEI grant-cancellation ruling: the constitutional theory became stronger at the same time the remedial path became narrower. A plaintiff who can frame the injury as viewpoint discrimination remains in district court. A plaintiff who wants the government to perform what looks like a grant agreement may be sent to the Court of Federal Claims. And for some non-procurement discretionary grants, that second forum may not be open in the way the channeling theory assumes.

The Two Holdings Do Different Work

The district court had certified two classes that mattered for different reasons. The DEI Termination Class covered recipients whose grants were terminated based on perceived DEI, DEIA, or environmental-justice viewpoints. The Form Termination Class covered recipients whose grants were ended through form letters without a viewpoint-based explanation. Those classes traveled together procedurally, but the Ninth Circuit’s reasoning did not treat them as interchangeable.

For the viewpoint-based terminations, the court saw a familiar constitutional problem: the government was not merely defining the subject of a federal program; it was penalizing a direction of thought within a subject. For the form-letter and restoration claims, the court saw a different problem: the requested relief looked contractual enough to trigger Tucker Act channeling away from district court.

The distinction is not academic. A research center whose award was cancelled because the agency labeled its project “DEI” has a different lawsuit from a recipient whose termination letter says little and whose complaint asks a court to put the dollars back. The same cancellation may generate both theories, but Thakur makes counsel separate them before filing, not after the emergency motion is drafted.

Why “DEI” Was Treated as Viewpoint, Not Just Subject Matter

The most important First Amendment move in Thakur is the court’s treatment of DEI, DEIA, and environmental justice as “inherently directional” viewpoints. That phrasing matters because viewpoint discrimination is not the same thing as content selection. A government grant program may define what it funds; it may prefer one field, one method, or one institutional mission over another. But when it targets a point of view within a field, the constitutional footing changes.

The court’s analysis sits closer to Rosenberger v. Rector than to Rust v. Sullivan. Rosenberger involved a university funding program that could not exclude a student publication because of its religious viewpoint once the program opened a forum for private speech. Rust, by contrast, allowed the government to fund a program with limits tied to the government’s own programmatic message. Thakur treated the grant terminations as punishment of private grantees’ viewpoints, not merely as the government declining to subsidize a program message of its own.[2]

That is why the word “directional” carries so much weight. “Environmental justice” is not just a topic like hydrology, public health, or community planning. It signals a normative orientation about how environmental burdens and benefits are distributed. “DEI” and “DEIA” similarly refer not only to institutional administration but to a set of views about access, exclusion, representation, and remedy. The government may dislike those views. What it may not do, under the Ninth Circuit’s reasoning, is terminate grants because the recipient is perceived to hold or advance them.

NEA v. Finley did not rescue the government’s position. Finley gave the National Endowment for the Arts room to consider “decency and respect” in a competitive arts-funding program, but it did not authorize categorical retaliation against a disfavored viewpoint. Thakur’s point is narrower and stronger: once the termination decision turns on the perceived ideological direction of the recipient’s work, the grant context does not dissolve the First Amendment problem.

This part of the ruling is the reason the decision is more than a procedural skirmish. If an agency can classify projects by politically disfavored terms and cancel them in bulk, the constitutional injury is not limited to delayed reimbursement. The government has changed the price of expressing a viewpoint in federally supported research.

The Funding-Discretion Argument Still Has Teeth, Just Not Everywhere

Grant recipients should not read Thakur as a holding that every DEI-adjacent award is insulated from federal policy change. The government retains broad discretion to design programs, set priorities, decline renewals, and define the scope of what it wants to buy, study, or support. A new administration can shift funding away from certain priorities without automatically committing viewpoint discrimination.

The line Thakur draws is between choosing a program and punishing a view. If an agency prospectively creates a grant program for a different research area, the constitutional question may be modest. If it combs through existing grants for DEI terminology and terminates awards because the funded work is perceived to advance an unwanted viewpoint, the question looks very different.

That distinction will make administrative records unusually important. Counsel will want to know whether the agency identified statutory or programmatic deficiencies, changed eligibility criteria, invoked budgetary constraints, or instead relied on labels such as DEI, DEIA, equity, inclusion, or environmental justice as proxies for ideological disfavor. Thakur rewards precision in the record because viewpoint discrimination depends on why the government acted, not merely on the fact that money stopped.

The Jurisdictional Holding Is the Part Clients Will Feel First

The procedural half of Thakur is less satisfying and probably more disruptive. The Ninth Circuit concluded that contract-based claims seeking restoration of grant funding could not proceed in district court through the APA because the Tucker Act impliedly channels those claims to the Court of Federal Claims. In ordinary language: if the substance of the lawsuit is that the government must honor a grant obligation and restore money, the plaintiff may not be able to keep the case in district court by styling it as arbitrary-and-capricious review.

That conclusion tracks the Supreme Court’s recent shadow-docket posture. In Department of Education v. California, the Court blocked district-court relief that had halted education grant terminations, emphasizing that claims seeking payment from the government belonged under the Tucker Act framework rather than the APA route in district court.[3] Thakur applies that same pressure to the DEI grant-termination setting.

NIH v. American Public Health Association sharpened the split. There, the Supreme Court allowed $783 million in NIH grant terminations linked to DEI initiatives to proceed, while Justice Barrett’s 4-1-4 alignment distinguished between challenges to general agency guidance and challenges to individual grant terminations.[4] The resulting message is not that all grant challenges are barred from district court. It is that courts are increasingly sorting them by the nature of the claim and the relief requested.

A challenge to a directive, policy, or guidance document may still look like an APA case. A challenge demanding reinstatement of a particular award may look like a contract or money claim. A First Amendment claim may remain in district court even when it arises out of a grant termination. Those distinctions are easy to state after the fact and hard to preserve in a complaint filed under emergency conditions.

The Court of Federal Claims May Not Be the Safe Harbor It Sounds Like

The Tucker Act answer assumes there is somewhere meaningful to go. That is where the remedial problem becomes uncomfortable. The Court of Federal Claims hears money claims against the United States, but Federal Circuit precedent has limited its jurisdiction over certain non-procurement discretionary grants. Rick’s Mushroom Service v. United States and St. Bernard Parish Government v. United States are the cases grant counsel now have to keep close: they stand for the proposition that not every federal assistance arrangement is a contract enforceable in the CFC.[5]

If a grant is treated as a procurement contract, CFC jurisdiction is easier to imagine. If it is a discretionary assistance award, the recipient may face a harder threshold question. The government can argue that district court is unavailable because the Tucker Act precludes APA review, while also arguing in the CFC that the grant is not the kind of contract the CFC can enforce. That sequence is not a merits loss. It is a forum problem masquerading as remedial discipline.

No part of Thakur resolves that gap. The Ninth Circuit could channel contract-flavored claims away from district court; it could not revise Federal Circuit doctrine governing the CFC. For grant recipients, that means the phrase “go to the Court of Federal Claims” may describe the next filing location without answering whether the court will have power to grant relief.

How a Future Recipient Has to Plead the Case

Imagine a nonprofit research institute receives a termination letter after an agency review flags its project as involving equity or environmental justice. The first question is not simply whether the termination was unlawful. It is what injury the lawsuit asks a court to remedy.

• If the injury is viewpoint discrimination, the complaint should identify the government’s classification method, the terms or criteria used, the decisionmakers’ stated reasons, and how the agency targeted a disfavored perspective rather than a neutral program boundary.
• If the injury is nonpayment or cancellation of a specific award obligation, counsel must evaluate whether the claim sounds in contract and whether the Tucker Act pushes it toward the Court of Federal Claims.
• If the award is a non-procurement discretionary grant, counsel must confront the CFC jurisdiction issue before treating Tucker Act channeling as a complete remedial answer.
• If the target is a broader agency directive rather than an individual termination, the APA path may look different from a suit seeking reinstatement of a specific grant.

The trap is that these theories overlap factually. The same termination letter can be evidence of viewpoint discrimination and the operative breach of a grant arrangement. But the complaint cannot assume that one forum can do everything. After Thakur, a pleading that treats constitutional injury, APA invalidity, and contract restoration as a single undifferentiated theory invites a jurisdictional fight before anyone reaches the speech question.

Form Letters Are Not the Same as Viewpoint Evidence

The Form Termination Class shows why not every ugly cancellation produces the same constitutional claim. A form letter may be arbitrary, thinly reasoned, or administratively careless. It may disrupt a lab, strand personnel, and waste months of project planning. But unless the record connects the termination to a prohibited viewpoint, the First Amendment theory is harder to prove.

That does not make form-letter terminations harmless. It makes them procedurally dangerous. The more a plaintiff’s case depends on the government’s failure to comply with grant terms or provide adequate individualized reasoning, the more the government can characterize the action as a contract-based dispute over a particular award. That characterization is exactly what moves the case toward Tucker Act preclusion.

This is where clients may hear the least intuitive advice. A bad termination letter is not always the best fact for district court jurisdiction. The best district-court fact is evidence that the government acted because of the recipient’s viewpoint.

The AI Classification Cases Show the Evidentiary Risk

Thakur is not the only case in which viewpoint targeting and blunt classification methods have converged. In ACLS v. McDonald, a Southern District of New York ruling dated May 7, 2026 addressed NEH grant terminations that used ChatGPT-assisted keyword classification. The court described the process as a “textbook example of unconstitutional viewpoint discrimination,” and the reported record included the fact that DOGE staff did not independently review the AI-generated classifications.[6]

The AI point should not be overstated. A keyword tool is not unconstitutional by itself, and the record in one NEH case does not prove that every agency used the same process. The legal problem arises when the tool becomes the government’s mechanism for identifying disfavored viewpoints and converting that identification into termination decisions without meaningful review.

For recipients, that makes discovery and record preservation central. Search terms, classification prompts, spreadsheets, review instructions, and internal routing can matter as much as the final letter. The First Amendment claim becomes stronger when the administrative machinery shows that the agency was not evaluating grant performance or statutory fit, but sorting recipients by perceived ideological direction.

What Thakur Does Not Decide

Thakur does not decide that all DEI-related grants must remain funded. It does not eliminate federal funding discretion. It does not resolve whether every cancelled grant is enforceable in the Court of Federal Claims. It does not settle how the Supreme Court would treat the Ninth Circuit’s First Amendment reasoning if the case returns on certiorari.

It also does not remove the need to distinguish adoption from legality. Agencies may adopt anti-DEI policies; that fact alone does not determine whether a specific termination is unconstitutional. Conversely, a recipient may hold a strong First Amendment theory; that does not guarantee that a court hearing a restoration claim has jurisdiction to order money back into the grant account.

The decision’s importance lies in its separation of those questions. Viewpoint-based cancellation is constitutionally suspect. Contract-based restoration is jurisdictionally constrained. Non-procurement discretionary grants may sit uneasily between the two.

The Ruling Leaves a Strong Claim and an Unfinished Remedy

For universities, nonprofits, and research organizations, Thakur is a real First Amendment victory. The Ninth Circuit’s treatment of DEI, DEIA, and environmental justice as directional viewpoints gives recipients a serious constitutional answer to categorical grant cancellation. Agencies cannot make a viewpoint disappear by calling it a keyword.

But the ruling is not a clean reinstatement story. The next wave of disputes will turn on the forum as much as the First Amendment. A plaintiff may be right that the government targeted a viewpoint and still have to separate that claim from a restoration demand that district court cannot hear. Another plaintiff may be sent to the Court of Federal Claims only to face the argument that the grant is discretionary assistance outside that court’s Tucker Act jurisdiction.

That unresolved remedial gap is where Thakur leaves the law in Q3 2026. It gives DEI-related grantees a stronger constitutional shield than they had before. It also leaves counsel navigating a procedural maze in which the wrong cause of action, the wrong requested relief, or the wrong forum can keep a court from reaching the viewpoint-discrimination question at all.

References

1. Ninth Circuit keeps DEI research grants off Trump's chopping block, Courthouse News Service
2. Cancellation of DEI-Related Grants Likely Violates First Amendment, Ninth Circuit Holds, Volokh Conspiracy / Reason
3. Supreme Court Blocks Use of Administrative Procedure Act to Halt Education Grant Terminations, Holland & Knight
4. Supreme Court allows Trump administration to terminate $783 million in NIH grants linked to DEI initiatives, SCOTUSblog
5. Caught Between Courts: Navigating Grant Termination Claims, Brownstein
6. Court Orders Reinstatement of Grants Terminated By AI-Assisted Review, Feldesman