The 529 Escape Hatch

Alex Rivera, Wealth Architect at The Wealth Grid

System Drop | August 24, 2026 | The Wealth Grid

Tuition checks are going out this week. Which means I have had four versions of the same conversation in the last ten days, and every one of them ended in the same place.

It goes like this. Someone tells me they never funded a 529 for their kid. I ask why. And they give me a perfectly reasonable answer that has been out of date for two and a half years.

What if she does not go. What if he gets a scholarship. What if they pick a trade. I did not want to lock money in a box that only opens one way.

That objection used to be correct. It was the single best argument against the account, and I made it myself for years. Money went in, and if it did not come out for education, you paid ordinary income tax on the growth plus a ten percent penalty on top. A one way door with a fine attached.

Congress took the lock off in 2024, almost nobody noticed, and the provision has a clock on it that starts the day you open the account. Which means the cost of not having opened one is compounding right now, quietly, in the background, for anyone still running the old objection.

So today we build it. Under an hour, most of it waiting on a website.

Section 126 of the SECURE 2.0 Act added a new paragraph to the 529 rules, effective for distributions after December 31, 2023. It says that leftover money in a 529 can be moved into a Roth IRA owned by the beneficiary, with no tax and no penalty.

Lifetime cap of 35,000 dollars per beneficiary. That is the headline.

But the headline is not the interesting part, and if you stop reading at 35,000 dollars you will misfile this the way most coverage did. The interesting part is who it lets in.

A normal Roth IRA contribution is income limited. For 2026 the single filer phase out runs from 153,000 to 168,000 dollars, and above that you are out. You go through the back door with a conversion, or you go without.

The 529 rollover does not incorporate that income test. It is simply not in the statute. The cap is the base contribution limit and nothing else.

Read that again with a specific person in mind. A thirty two year old software engineer earning 240,000 dollars a year cannot make a direct Roth contribution. If her parents opened a 529 for her when she was in kindergarten, and there is money left in it, she can receive 7,500 dollars into a Roth this year anyway. Then again next year. Up to 35,000 dollars total.

That is not leftover college money. That is a legal pipe into a Roth for someone the Roth was closed to.

Every one of these has to be true. Miss one and the transfer stops being a rollover and becomes an ordinary non qualified withdrawal, with the earnings taxed and the ten percent additional tax applied. There is no partial credit.

One. The account has to be fifteen years old. Measured for that beneficiary. This is the test that matters most and the reason this article exists today rather than in ten years.

Two. The five year lookback. Contributions made in the last five years, and the earnings on them, cannot move. Only the older money is eligible.

Three. The Roth has to belong to the beneficiary. Not the account owner. A parent cannot roll a child's 529 into the parent's own Roth. The kid gets it.

Four. The annual cap is the Roth contribution limit. For 2026 that is 7,500 dollars under 50, or 8,600 dollars at 50 and over, and it is shared. If the beneficiary already put in 3,000 dollars of her own money, only 4,500 dollars of room is left for the rollover.

Five. The beneficiary needs earned income at least equal to the amount rolled. A kid with no job cannot receive a rollover, no matter how old the account is.

And one mechanical rule that trips people who otherwise did everything right. It has to be a direct trustee to trustee transfer. You cannot pull the cash out, hold it, and put it in the Roth yourself. Do that and you have made a taxable withdrawal with extra steps.

Here is what I want you to actually do something about.

Fifteen years is a long test, and it starts when the account is established for that beneficiary, not when you fund it properly. Which creates a situation that is unusual in personal finance: there is a version of this where the correct move is to open an account you have no immediate plan to use.

Most states let you open a 529 with 25 dollars. Some with nothing at all.

Open one for each kid this week. Put in whatever the minimum is. The balance is irrelevant. What you are buying is a start date, and start dates cannot be purchased later at any price.

Fifteen years from now that account is either a funded college plan, or a funded Roth pipeline for a working adult who may well be over the income limits by then, or both in sequence. You do not have to know which today. That is the whole point. You are converting a decision you cannot make yet into an option you will still hold when you can.

The cost of being wrong is 25 dollars. The cost of being right and not having started is fifteen years.

One caution before you go opening accounts for everybody you know. Changing the beneficiary appears to restart the fifteen year clock for the new beneficiary, and the guidance here is not as settled as the rest of the provision. Treat beneficiary changes as clock resetting until somebody tells you otherwise in writing. Which is another argument for separate accounts opened early rather than one big account you plan to shuffle around later.

Minutes 0 to 15. Pick the plan.

You are not required to use your own state's plan. You can use any state's. But check your state first, because roughly thirty states offer a deduction or credit on contributions, and a handful of them give it to you regardless of which state's plan you use.

If your state gives a deduction only for its own plan, the deduction usually wins and you use the home plan. If your state has no income tax or no deduction, you are free to shop, and you should shop on expense ratio and fund quality alone.

Plan documents are written to be skimmed and not understood. I run them through Galaxy.ai and ask three questions in plain language: what is the total annual cost on a balance of this size, what is the minimum to open, and does the plan charge anything to move money out. The third one matters more than people expect, and it is never on the front page.

Minutes 15 to 30. Open and fund the minimum.

One account per beneficiary. Correct name, correct Social Security number, correct date. Screenshot the confirmation showing the establishment date and put it somewhere permanent.

That screenshot is the document that proves your fifteen year clock started, and you are going to need it in a decade and a half when a custodian asks you to substantiate the account age. Providers get acquired. Plans change administrators. Records get migrated badly. Yours will not.

Minutes 30 to 45. Set the contribution and the investments.

Pick an amount you will not resent. Fifty dollars a month is a real plan. So is 500 dollars. Automate it so the decision happens once.

Then check the investment selection, because most plans default to an age based track that gets conservative as the beneficiary approaches eighteen. That default is correct if the money is for tuition. It is wrong if this account is going to spend part of its life as a retirement vehicle, because it will be sitting in short bonds during exactly the years you most want it growing.

You do not have to solve that today. You do have to know the default exists and revisit it once you know which job the account is doing.

Minutes 45 to 60. Build the record.

The fifteen year and five year tests are documentation problems, not investment problems. Every contribution needs a date attached, permanently, because in year sixteen you will need to prove which dollars cleared the lookback.

Mine runs on Make.com. Every contribution confirmation email that lands gets parsed and appended to one sheet: date, beneficiary, amount, running total. It took an evening to build and it has run untouched since. In fifteen years I will have a complete contribution ledger and I will not have thought about it once in the meantime.

If you do not want to automate it, keep a spreadsheet by hand. Just keep it somewhere that will outlive the laptop.

The overfunding fantasy. The rollover is a release valve, not a strategy. Thirty five thousand dollars is a lifetime cap per beneficiary and it takes about five years of transfers to move it. Do not deliberately stuff a 529 as a retirement play. Fund it for education, and treat the Roth exit as insurance against the thing you were afraid of.

The state clawback. Several states that gave you a deduction on the way in will recapture it if money leaves for a non qualified purpose, and some of them have not clarified how they treat a Roth rollover. Federally clean does not mean clean in your state. Check before you transfer, not after.

The earned income gap. The beneficiary needs wages at least equal to the rollover in the year it happens. A twenty two year old in graduate school with no job cannot use this. Line the transfer up with a year they are actually working, which usually means waiting rather than rushing.

Run the plain version. You open an account for a five year old, fund it modestly, and they go to college. It did what it said. Fine.

Now run the version people were afraid of. They get a scholarship, or they skip school entirely, and 40,000 dollars is sitting there.

Old rules: you eat ordinary income tax on the growth plus a ten percent penalty, or you leave it parked forever hoping a grandchild materializes.

New rules: 35,000 dollars of it walks into their Roth over about five years, tax free, at an age where it has forty years to compound, in an account they could not otherwise contribute to if their career went well. The remainder can go to a sibling, or wait.

The downside case became a retirement account. That is the whole change, and it is why the objection everyone still repeats is no longer true.

I built the whole thing as a single file. It has the five tests as a pass or fail checklist, a contribution ledger with the five year lookback calculated automatically, the establishment date record sheet, a state by state note on deduction and recapture treatment, and the exact language to hand a custodian so the transfer gets coded as a rollover instead of a distribution.

Reply with the word HATCH and I will send you The 529 Escape Hatch Worksheet, free. No form, no link to chase. Reply HATCH and it shows up.

We opened The Grid Inner Circle a few weeks ago and it has been the best part of my week since. It is where the builds get pulled apart in public, where people post the version they actually implemented, and where I answer the questions that do not fit in a newsletter.

Twenty nine dollars a month. Or 499 dollars once, for founding lifetime access, which stays 499 dollars forever no matter what the monthly becomes.

Reply GRID and I will send you the details.

Most of the money people leave on the table is not lost to bad decisions. It is lost to good decisions that stopped being good and never got revisited.

The case against the 529 was airtight for twenty five years. Then a paragraph got added to the code and the case quietly collapsed, and the objection kept circulating anyway, because nobody sends a letter when the reason you said no stops applying.

Open the account. Fund the minimum. Start the clock.

Wednesday we go after something with a shorter fuse. Home equity lines just hit their lowest rate of 2026, and the useful move is not to borrow. It is to get approved for money you have no intention of touching, while you still look good enough on paper to be approved. We are going to build standby capacity, and I am going to make the case that the best time to arrange credit is precisely when you do not need any.

Alex Rivera, Wealth Architect at The Wealth Grid

Wealth is a system, not a guess.