Trump’s Power Stops at Money, Votes, and Fear

The easiest mistake in covering Trump’s second term is to treat every setback as proof that the system is working, or every victory as proof that the system has collapsed. I think both readings miss the actual machinery of power.

The useful question is not whether President Trump is “being checked.” He is. The question is whether those checks change what the administration can do, or merely slow the route it takes to get there. On that standard, the answer is uneven but clear: courts still matter, Congress matters when money or Senate math is at stake, civil-service law is now a contested speed bump rather than a wall, and the pardon power is almost entirely a one-person lever. The deepest constraint is not institutional virtue. It is a harder thing: whether a rival institution controls something Trump needs, and whether Republicans fear the political cost of giving it to him.

Start with the courts, because they are the most visible check and the easiest to overstate. An injunction is a court order telling someone to do something or, more often in these cases, to stop doing something. During the first 100 days of Trump’s second administration, the Congressional Research Service identified 25 cases in which federal courts issued nationwide injunctions against federal actions, including disputes over birthright citizenship, an Office of Management and Budget funding pause, Treasury payment-system access, National Institutes of Health grant rates, foreign aid, federal employment, and agency shutdown efforts according to CRS¹. That is not theater. When a judge blocks implementation, an agency cannot simply proceed as if nothing happened.

But this is also where the story gets less comforting for Trump’s opponents. In June 2025, the Supreme Court’s decision in Trump v. CASA limited lower courts’ ability to issue what lawyers call universal or nationwide injunctions, meaning orders that protect nonparties as well as the plaintiffs in court. CRS summarized the ruling this way: federal courts may not issue injunctions broader than needed to give complete relief to the parties before them, though class actions, state suits, Administrative Procedure Act cases, and other forms of broad relief remain available under the Court’s framework². That turns litigation from a big red stop sign into a more technical system of gates. Some gates still close. Others just force challengers to queue up correctly.

The June 2026 SNAP case shows the difference. A federal judge in Boston granted a preliminary injunction for 20 Democratic-led states challenging Trump administration conditions on Supplemental Nutrition Assistance Program funding, including conditions tied to “gender ideology,” immigration, and athletics the Associated Press reported³. SNAP is the food-aid program that helps low-income Americans buy groceries, and the order matters because the Agriculture Department cannot use those disputed conditions against the plaintiff states while the case proceeds according to the same AP account³. That is a real constraint. It is also not a final nationwide defeat. The administration can appeal, revise, or keep fighting state by state.

Money is the harder check. The appropriations power is Congress’s constitutional control over federal spending: Article I says no money can be drawn from the Treasury except through appropriations made by law as the Constitution Annotated states⁴. This is where Congress has leverage that courts do not. If the executive needs new money, a majority coalition can say no, say yes with conditions, or make the price painful.

The “anti-weaponization” fund is the best recent example, and also the best warning against exaggeration. Trump sought a $1.776 billion mechanism to compensate people claiming they were victims of government “weaponization,” a plan critics warned could benefit Jan. 6 defendants and other Trump allies Axios reported⁵. After bipartisan blowback and adverse court signals, Axios reported on June 1 that the administration planned to drop the fund citing senior administration officials⁶. That is not symbolic resistance. A concrete pot of money appears to have been sacrificed because the politics and legal posture got too ugly.

Yet the broader funding fight ended much better for Trump than that concession suggests. The Senate passed about $70 billion for Immigration and Customs Enforcement and Border Patrol by a 52-47 vote, funding those agencies for three years through the end of Trump’s term, after rejecting efforts to permanently ban or limit the settlement fund according to AP⁷. So Congress did constrain the side scheme. It did not constrain the main immigration-enforcement buildout. That distinction is the whole ballgame.

Congressional oversight, by contrast, is weaker unless it connects to votes, money, subpoenas, confirmations, or deadlines. Oversight means Congress reviewing and investigating executive agencies, usually through hearings, document demands, and public pressure. It can expose misconduct. It can build a record. But exposure alone does not stop policy. The Senate’s FISA fight shows when oversight-adjacent pressure becomes real leverage: seven Republicans joined Democrats to block a procedural move on extending Section 702, the warrantless foreign-intelligence surveillance program, amid concerns over Trump’s choice of Bill Pulte for an intelligence role CBS News reported⁸. That vote changed the math, at least temporarily. A stern committee letter would not have.

The civil service is the murkiest battlefield. The civil service is the career federal workforce and the legal system meant to protect many government employees from arbitrary or partisan firing. Trump’s Schedule Policy/Career rule, descended from his first-term “Schedule F” idea, targets career positions with a “confidential, policy-determining, policy-making, or policy-advocating” character and places them in the excepted service, where some employees have different notice and appeal rights CRS explains⁹. This is where the unitary executive theory enters the picture. That theory holds, in its strong form, that Article II gives the president broad control over executive-branch officials; Cornell’s Legal Information Institute describes it as the view that the president possesses sole authority over the executive branch and may remove appointed executive officials without approval from Congress or courts in its legal overview¹⁰.

The administration is not pretending the rule is modest. OPM’s Federal Register explanation says positions moved into Schedule Policy/Career will keep merit-based hiring procedures if they came from the competitive service, but it also says the point is to restore accountability and remove what it calls “undemocratic resistance” inside the civil service in the final rule¹¹. CRS notes that the rule rescinds Biden-era protections that allowed involuntarily reclassified employees to appeal to the Merit Systems Protection Board, and that unions have filed challenges under theories involving Title 5 and the Administrative Procedure Act in its legal sidebar⁹. My read: civil-service law still bites, but mostly after the administration has forced the fight onto narrower terrain. Employees and unions can sue. Agencies must classify positions and defend their reasoning. But the old assumption that career status itself would stop a purge is weaker than it was.

Then there is the pardon power, where the institutional brakes are nearly absent. Article II gives the president power to grant reprieves and pardons for offenses against the United States, except impeachment cases under the Constitution’s text¹². Trump’s pardon of former Rep. Stephen Buyer was a clean example: the White House proclamation granted Buyer a “full, complete, and unconditional pardon” on June 4, 2026 according to the proclamation¹³. Buyer had been convicted in a federal insider-trading case, sentenced to 22 months, ordered to forfeit more than $350,000, and released in 2025 AP reported¹⁴. Congress can investigate a pardon. Voters can punish it. Courts generally cannot reverse it because they dislike the beneficiary.

The strongest counterargument is that I am giving too much credit to institutions that mostly create delay. The Senate still funded Trump’s immigration agencies. CASA made injunctions less sweeping. Schedule Policy/Career is built to reduce civil-service appeal rights. The Buyer pardon is effectively untouchable. All true. But the right lesson is not that checks are fake. It is that they are transactional. They work when attached to something the executive cannot manufacture by will: appropriated dollars, Senate votes, court-enforceable party relief, statutory procedures, or electoral cover from nervous Republicans.

The political layer is becoming more important, not less. RealClearPolling’s generic congressional ballot average currently shows Democrats ahead by 6.4 points, 48.1% to 41.7% in its 2026 tracker¹⁵. Axios reported in February that Republican strategists were increasingly worried about losing not just the House but also the Senate, where Republicans hold a 53-47 majority, and that some internal polling showed competitive races expanding into states like Alaska, Iowa, and Ohio according to its reporting¹⁶. In March, Trump advisers urged House Republicans to stop emphasizing “mass deportations” and instead focus on violent offenders, a revealing adjustment from the administration’s own political team Axios reported¹⁷. That is the practical constraint taking shape: not Republican rebellion as a moral awakening, but Republican self-preservation as a governing fact.

My prediction is that through November 2026, Trump will keep winning the central fights where Republicans can give him what he wants with party-line votes, especially immigration money and personnel flexibility. He will lose or narrow the fights that combine three things: a visible beneficiary who looks corrupt, a legal choke point that courts can police, and enough GOP incumbents afraid of campaign ads. Watch the House vote on the immigration package, the next round of Schedule Policy/Career designations, and whether Republican senators demand statutory language rather than verbal assurances. If they do, Trump’s power has found a wall. If they do not, it has mostly found traffic.