Start with the thing almost nobody knows

The United States is not treaty-bound to defend Israel.

There is no mutual-defense pact of the kind we have with NATO, Japan, or South Korea. Written promises do exist, and the ones people reach for are narrower than they sound. A 1975 agreement says that if a world power threatens Israel, Washington will view it with particular gravity and consult promptly about what support it can lend — a Cold War clause aimed at the Soviet Union. View with gravity. Consult. Not one binding verb, and nothing that requires us to show up.

Worth knowing why, because the obvious reading is wrong. A treaty runs both ways. It would bind Israel as much as it binds us, and it would put the relationship in the Senate's hands rather than each president's. An arrangement of executive agreements asks neither government to accept that.

Now hold that next to the rest of the record.

Congress's own research service reports that Israel is the largest cumulative recipient of U.S. foreign assistance since World War II — $174 billion through 2025. The current agreement, signed in 2016, pledges $38 billion across ten years. A normal year is $3.3 billion in military financing plus $500 million for missile defense. In fiscal 2024, with emergency money added, it reached about $12.5 billion.

The money is the visible part. It isn't the exceptional part.

The exceptional part is the veto. At the United Nations Security Council, the United States has blocked resolutions criticizing Israel dozens of times across decades — under presidents who agreed with each other about almost nothing else. The United States has cast more than fifty vetoes on Israel-related resolutions — more than half of every veto it has ever used, and more than three times its next-most-shielded subject. And unlike an arms sale, a veto is cast by the executive branch — there's no roll call, and no member of Congress has to answer for it.

So: we owe Israel less on paper than we owe Japan, and we give it more than we give anyone. That gap is the story. And the two usual explanations for it don't close it.

Both easy answers come up short

"We share values and interests." True, and incomplete. We share values and interests with plenty of countries. None of them get the veto.

"Israel controls Washington." This one is worse than incomplete. It's the shape of an old lie about Jewish power, and the record cuts against it.

Look at the 2015 Iran nuclear deal — the fight AIPAC picked and lost. Deal opponents held simple majorities in both chambers. In the Senate, 58 senators voted their way. They fell two short of the 60 needed to break a filibuster, and the deal survived on a procedure rather than on the merits of the count.

Read that honestly, because it cuts both ways. A majority of the United States Senate voted against a sitting Democratic president's signature achievement, and four senior Democrats — Schumer, Cardin, Menendez and Manchin, including the man who would become the party's leader — were among them. Whatever moved those votes, the side AIPAC was on had the floor. That is not weakness.

And it wasn't enough, because Obama had chosen an instrument that never needed a majority.

So the biggest defeat handed to the pro-Israel lobby in a generation came with them holding the floor of both chambers. They didn't lose the argument in Congress. Congress just wasn't where it got settled.

Which is the clue. If a lobby that strong can lose that badly and the relationship doesn't move an inch, the lobby isn't what's holding it up.

Six reasons, and no single one is the answer

The relationship holds because it isn't one thing.

Strategy. Israel has a capable military and serious intelligence services. It can hit shared enemies without America stationing troops. The two countries build missile defense and cyber tools together.

Memory. The Holocaust, centuries of persecution, and the world's refusal to offer refuge built a real moral case for a Jewish homeland. That history is why threats to Israel get heard in Washington as existential rather than as ordinary diplomacy. This one sits differently from the rest of the list. It isn't a constituency or a contract. It's the reason the argument lands.

Law. Congress wrote the relationship into the plumbing of government. When officials certify an arms sale to any Middle East country other than Israel, American law requires them to determine that the sale won't erode Israel's military edge over credible conventional threats — from states, from coalitions, or from groups that aren't countries at all. Israel's aid is paid out fast and parked in an interest-bearing Federal Reserve account. We keep weapons stockpiled on Israeli soil.

Faith. Christians United for Israel organizes support for Israel around scripture — its leaders cite Genesis 12:3, "I will bless those who bless you" — and it calls itself the largest pro-Israel organization in the country, saying it has more than 10 million members. Nobody audits that number, and its members are free sign-ups. But the polling isn't self-reported: in Pew's May 2026 survey, white evangelical Protestants were the only religious group surveyed in which a majority — 57% — still viewed the Israeli government favorably. Jewish Americans came in at 47%.

Money. AIPAC's affiliated super PAC, United Democracy Project, can spend without limit, and it spends in primaries rather than general elections — where turnout is low and a few million dollars goes a long way. In the 2024 cycle it reported $35.6 million in independent expenditures. This cycle it has raised $98.6 million, as of its May 31, 2026 filing.

Contractors. Military aid is also industrial policy. That money largely buys American-made equipment, which means factories, jobs, and congressional districts. Every appropriation arrives with a domestic constituency already attached.

Here's what matters. No single one of these is the answer. AIPAC spent tens of millions to kill the Iran deal, held the Senate floor, lost anyway, and nothing about the relationship changed. A lobby is a thing you can beat.

It survives internal disagreement because nobody in it has to agree with anyone else. The Pentagon's reason is Iran. An appropriator's reason is his district. CUFI's reason is scripture. A contractor's reason is jobs. None of them needs the others to be right, and none of them needs to coordinate with the others to land in the same place.

Priced or sacred

Watch how Washington talks about the people it protects.

Mississippi, 2018. Trump on Saudi Arabia: "We protect Saudi Arabia. Would you say they're rich? And I love the king, King Salman. But I said, 'King, we're protecting you. You might not be there for two weeks without us. You have to pay for your military.'"

That's a bill, delivered on a stage.

He talks about Israel's money too. Oval Office, April 2025, Netanyahu sitting next to him, facing a tariff anyway: "Don't forget: We help Israel a lot. You know, we give Israel $4 billion a year. That's a lot."

Same ledger. Different ending. He told the king to pay. He told Netanyahu the number and then kept writing the check.

That's the tell. The invoice comes out for everyone. It only gets sent to some of them.

What it costs to say no

Six reasons to support. Now the part that decides it — and it isn't a reason at all.

There is no counterweight. There's no Palestinian line in the Federal Reserve account and no Palestinian war-reserve stockpile — though the honest reason is mundane: those are instruments between states, and the United States doesn't recognize a Palestinian one. The military-edge law is different. Congress wrote Israel into the statute by name. There is no counterpart for anyone else, and no mechanism by which one could exist.

Which produces the mechanism underneath everything above:

Supporting Israel is the inherited default. Challenging it wakes up organized constituencies who vote in your primary. Keeping it spreads the cost — in money and in lives — across people who mostly aren't counting, and across people with no vote here at all.

The political cost is concentrated and lands on the politician. The other costs are diffuse and land somewhere else. That's the asymmetry. It isn't a choice between right and wrong. It's a choice between a bill you pay and a bill someone else does.

The law has conditions. Enforcement is the part that slips.

This is the part the seven-figure numbers hide.

In May 2024, the State Department reported it was "reasonable to assess" that American weapons had been used by Israeli security forces in ways inconsistent with Israel's human rights obligations — or with best practices for avoiding civilian harm. Read that carefully: it's a low bar and a hedged one. State said in the same report that it had reached no conclusion about any specific incident. No transfer was paused as a result of that review. The policy behind it, NSM-20, was rescinded by the Trump administration in February 2025.

In April 2025, the Government Accountability Office reviewed how State and Defense handle allegations that American weapons harm civilians. As of April 2025 it counted 634 reported incidents worldwide — most, State officials told GAO, from the war in Gaza — with 571 still at the first stage of review. Of those, 473 hadn't been started. In December 2025, State said it no longer runs that process at all: not required by statute, and it couldn't be resourced. In January 2026 it said it would track civilian-harm incidents through its arms-transfer risk reviews instead. That is not the same as nothing happening. It is a review process that backed up, and was then retired.

Read the Law pillar again. The same body of law that mandates Israel's military edge, and the fast disbursement, and the stockpiles, also contains conditions on what those weapons may be used for.

The edge is enforced on a schedule. The account is funded on time. The stockpiles are maintained. The conditions get reviewed — slowly, incompletely, and in the one documented instance above, without stopping anything.

That's not a claim about anyone's motives. It's a description of which parts of a statute run on rails and which parts run on judgment.

The default is cracking — on one side of the aisle

Here's the turn, and it needs to be stated precisely, because the precise version is more interesting than the loose one.

Since Gaza, the Senate has voted repeatedly on whether to take up resolutions blocking arms sales to Israel. Watch the count of senators willing to block at least one sale:

  • November 2024 — 19
  • April 2025 — 15
  • July 2025 — 27
  • April 2026 — 40

Every one failed. Support more than doubled between November 2024 and April 2026.

Now the part that matters. Not one Republican voted yes on any of them. The forty are 38 Democrats and 2 independents — all but seven of that caucus. This isn't the Senate moving. It's one party moving inside a chamber that hasn't.

The House said the same thing on July 15, when an amendment to strip $3.3 billion from Israeli security aid failed 104–314 — and split the Democratic caucus down the middle, 103 to 98, with 10 voting present. The party's whip and Nancy Pelosi voted to cut. The leader, Hakeem Jeffries, voted to keep. One Republican voted with them.

Two chambers, three months apart, same finding: the thing that's breaking is a Democratic consensus, not an American one. And in June 2025, 39 senators voted to block arms sales to Qatar and the UAE — more than the 27 who would block a sale to Israel a month later. Israel was the more protected partner. By April 2026 it was at 40. The exception is converging on the ordinary.

The public is messier than either party's story. Asked in September 2025 whether the U.S. was giving Israel too much military help for the war with Hamas, 33% said too much, 23% said about right, and 8% said not enough — while 35% weren't sure. Roughly a third say too much and roughly a third haven't decided. Anyone quoting this poll with two numbers is hiding the biggest one.

Even the pillars move. Evangelical support for the Israeli government is at 57%, still a majority and still the only one — but down from 71% in early 2024.

None of this is a policy change. The money is intact. Forty votes is not a majority, and one party is not a Congress.

What's changing is the price. The default held because breaking it cost a politician more than keeping it. Half of one party is now testing whether that's still true. The other half, and the entire other party, have not moved at all.

What this piece is not saying

Jewish Americans do not control American foreign policy. That idea is an old and dangerous lie. It also fails as analysis. It can't explain why evangelical Christians poll ten points more favorable toward the Israeli government than Jewish Americans do — 57% against 47%, in the same survey. Jewish opinion on Israel is deeply divided, and always has been.

The strategic case is not fake. Israel really does have capabilities the U.S. uses. The question isn't whether there's value. It's why that value comes with conditions that don't get enforced.

The moral case is not fake. The history behind it is real, and it is the one reason on that list nobody advances for gain. The blind spot is that a story centered on one people's survival can make another people disappear from the frame.

Nobody is bought. Money mostly follows politicians who already agree, and well-funded campaigns lose — the Iran deal is the proof. The narrower claim is the evidenced one: organized spending raises the expected cost of dissent — and foreign policy is rarely what decides a primary.

What can I do?

Ask about conditions, not support. "Do you support Israel" has one answer and tells you nothing. Ask instead: what conditions would you attach to military aid, and what happens if they're broken? The second half of that question is the one with no comfortable answer.

Read the roll call, not the press release. Senate and House votes are public, with every name attached. Your member's statement about a vote and their vote are two different documents.

Look up the money in your own primary. Independent expenditure filings are public at the Federal Election Commission. Reports on qualifying spending come in fast near an election — within 24 or 48 hours depending on timing — which means the money aimed at a race is usually visible before the coverage of it is.

The Receipts