Peter Thiel put $1,700 into a Roth IRA in 1999 and turned it into $5 billion. Tax-free. The kind of tax-free where the IRS sends no bill, ever, on a sum larger than the GDP of some small countries.
Naturally, the personal finance internet lost its mind. The reaction wasn't "our tax code has a hole in it" so much as "wait, how do I do that." Which is the wrong question, because almost nobody can do exactly what Thiel did. But it's not useless either: most people walk right past the few pieces of his playbook that are within reach, equity comp or not.
Here's the real story of the $5 billion Roth IRA, the part that's out of reach, and the parts that aren't.
The $1,700 That Ate the Tax Code
The numbers, sourced from the 2021 ProPublica investigation that pulled this story out of IRS records:
In 1999, Thiel's Roth IRA bought 1.7 million shares of PayPal, a company he co-founded, for $0.001 a share. Total cost: $1,700, comfortably under that year's $2,000 contribution limit.
That phrasing matters more than it looks. The shares were never moved into the account, because contributing appreciated property to an IRA isn't allowed. Cash went in as a contribution, and Thiel bought the shares inside the account. Every version of this strategy turns on that distinction.
In 2002, eBay acquired PayPal. Thiel sold his shares inside the Roth, and by the end of that year the account was worth roughly $28.5 million. No capital gains tax on the sale, because gains inside a Roth aren't taxed. Then it kept going. ProPublica's reporting has the account holding early stakes in Facebook, Palantir, SpaceX, and Airbnb at various points, bought while each was still cheap. As of 2019, the last year in the records they obtained, the Roth was worth about $5 billion, spread across 96 subaccounts.
The mechanism is almost stupidly simple once you see it. A Roth IRA grows tax-free and comes out tax-free in retirement. Contribution limits cap what you can put in each year, but they say nothing about how much the assets inside can grow. Thiel didn't beat the contribution limits. He made them irrelevant by stuffing the account with an asset that happened to appreciate by a factor of roughly three million.
That's the whole trick. Cheap basis, enormous growth, tax-free wrapper.
Why the Setup Doesn't Transfer
Thiel did this as a founder. He bought founder's shares at par value, a tenth of a penny, at the literal formation of the company, before any outside money had set a higher price. Someone joining at Series C is buying into a 409A valuation that already reflects years of growth. The cheap-basis half of the equation is mostly a function of arriving first.
The self-directed IRA is where most replication attempts quietly end. Holding private company shares in an IRA requires a self-directed IRA (SDIRA), an account at a custodian equipped to hold and value assets that don't trade on an exchange. They're legal and real. They're also unforgiving.
The unforgiving part is the prohibited transaction. IRC §4975 bars an IRA from transacting with its owner or with "disqualified persons": spouse, parents, children and their spouses, and any company where those people together own 50% or more. It also sweeps in officers, directors, and 10%-plus owners of a business the IRA is involved with.
In plain terms, an IRA generally can't buy into a company its owner controls, or one where the owner is a major insider, or do a deal that benefits the owner personally on both sides. And the penalty for tripping it isn't a fine. The IRS treats the entire IRA as distributed on January 1 of the year the violation happened, so the whole balance becomes taxable income, plus a 10% penalty under age 59½. One bad transaction can detonate the account.
A concrete version: you already own shares of the startup personally, and you sell a block of them to your own SDIRA at an honest, arm's-length price. The price is irrelevant. A sale between an IRA and the person who owns it is prohibited on its face, and that one signature is enough.
Which is why running real money through an SDIRA means tax and legal counsel on retainer and documenting everything. Thiel almost certainly did. The strategy is less "clever loophole" than "expensive, audit-prone tightrope that happens to be legal."
The $1,700-into-$5-billion part was never the transferable part anyway.
The Three Levers Thiel Pulled
Strip the celebrity off the story and the strategy comes down to three ingredients:
Low basis, from being early: he acquired the asset before it was worth anything.
A tax-advantaged wrapper: the growth happened somewhere the IRS couldn't touch it.
Legal compliance: careful records, real counsel, no prohibited transactions.
Lever one is the one you can't match. A founder's basis is a tenth of a penny; yours is whatever the last round priced it at. But the question that matters is: which version of cheap basis and which tax strategies are available to me, right now, given what I already hold?
That's not a billionaire question. That's a Tuesday-afternoon planning question. And the answers are more useful than the Thiel story lets on.
What's Reachable From Here
Five levers. Two of them turn on having been early. Three don't.
QSBS, when the timing happens to line up. Qualified Small Business Stock is the closest the tax code comes to what Thiel got. Stock in a qualifying C-corp, acquired while the company's gross assets were under the threshold, held long enough, and meeting the other tests, can have a large share of its gain come out federal-tax-free at sale. No Roth required.
The catch is the same one that limits the Thiel story. The gross-assets test is measured at issuance, so eligibility is a function of how early the stock was issued, not of being an employee. A company that has raised much capital has usually already crossed the line, which means this is a lever for people who were there before the raises rather than a general employee benefit. It's specific enough that it's worth checking rather than assuming, in either direction. (We go deep on this in our guide to the QSBS exclusion.)
83(b) elections on early-stage equity. Restricted stock (not RSUs, actual restricted stock, common at very early companies) carries a 30-day window to file an 83(b) election, which taxes the value at grant rather than at vesting. When the grant-date value is near zero, the election locks in a near-zero basis and starts the capital gains clock early. It's the employee-scale version of buying low: choosing to be taxed while the number is tiny.
The same door opens for options when a plan permits early exercise, meaning exercising before the shares vest and filing 83(b) on the unvested stock. If the strike equals fair market value at that moment, there's no spread and nothing to tax, and everything after is capital gain. Not every plan allows it, so it's worth reading the plan documents rather than assuming. Either way, the 30-day clock runs from the grant date, not from when anyone notices, and it can't be restarted.
Mega backdoor Roth. If an employer's 401(k) allows after-tax contributions and in-plan conversions, far more than the standard deferral limit can move into Roth treatment each year, often tens of thousands beyond the normal cap. It works every year rather than only at a liquidity event, which for a high-earning W-2 employee usually makes it the highest-value lever on the list. (Full mechanics in our mega backdoor Roth guide.)
The ordinary backdoor Roth. The smaller sibling, and the one that needs no cooperation from an employer. Above the Roth limits a direct contribution isn't allowed, but a non-deductible Traditional IRA contribution followed by a conversion lands in the same place: $7,500 for 2026, $8,600 at 50 and over. The catch is the pro-rata rule on conversions, which looks at every pre-tax dollar in any Traditional, SEP or SIMPLE IRA as of December 31 and makes a proportional slice of the conversion taxable. Starting from zero, it's fifteen minutes of paperwork a year. Starting with an old rollover IRA sitting there, it needs a plan first.
Roth conversions in a lower-income year. Converting Traditional IRA or old 401(k) dollars to Roth costs ordinary income tax on whatever is converted, so the entire question is what rate applies in the year it happens. For someone sitting at the top rate year after year, there usually isn't a good year for it. When one does appear, whether a lower-paying job taken on purpose, a lighter year, or a year the equity didn't vest at the price it was supposed to, that's when the arithmetic changes. It's the least glamorous item here and the one that applies least often.
None of these turns $1,700 into $5 billion. The first two depend on having been there early. The last three depend on your own circumstances: what your plan allows, what's already sitting in your IRAs, and what rate applies this year. That's the reachable version of the Thiel story.
Why the Exact Playbook Is Harder to Run Now
The Thiel reporting did what these things tend to do: it drew a spotlight. Congress has floated changes that would cap how much can sit in tax-advantaged retirement accounts and bar IRAs from holding investments that require accredited-investor status. The 2021 version that passed the House would have forced distributions from balances above $10 million for high earners, with no grandfathering for accounts already that size. It died in the Senate. The direction it pointed did not.
The bigger practical barrier isn't legislation. It's the prohibited transaction rules described above, which already make founder-style SDIRA moves treacherous for anyone who's an insider at the company whose shares they want to buy. The window Thiel walked through in 1999 was wider, the custodial infrastructure was looser, and the IRS wasn't on the lookout. Today it is.
The takeaway isn't "you missed it." It's that the headline strategy was always the wrong thing to chase. The transferable lesson was sitting underneath it the whole time.
The Part Worth Stealing
Here's the lesson, and it isn't about Thiel.
He didn't get rich because he found a secret account type. He got rich because someone (Pensco's founder, in the ProPublica telling) looked at his specific situation and said put the cheap asset in the tax-free wrapper before it grows. That's a coordination move. It required seeing the equity, the account structure, and the tax treatment as one connected decision rather than three separate ones.
That's the part that translates. QSBS eligibility, an 83(b) window, conversion timing, mega backdoor capacity: individually they're tactics, and individually they get missed, because the grant sits in one place, the account in another, and the tax treatment in a third. Coordinated, with someone watching the whole board, they're the difference between a comfortable outcome and an exceptional one.
Thiel had a team keeping careful records and timing every move across 96 subaccounts. The strategy that built the $5 billion wasn't really the Roth. It was the coordination.