The Hidden Tax Math Behind Employee Equity in Pre-IPO AI Companies

Two engineers, same grant value, same company, same vesting schedule. One wires $61,000 to the IRS in April. The other wires $214,000. The difference isn't luck or negotiation. It's a form filed in December that most people never look at twice.

If you're sitting on shares in a private AI company right now, every headline about 2027 is a distraction from the only thing you control: the sequence of decisions you make between the grant and the wire. Get the order right and you keep the money. Get it wrong and you fund the IRS, the state, and brokerage fees out of a bank account that hasn't seen a liquidity event yet.

Here's the structure I'd walk you through, and the specific traps that eat six-figure sums.

Why the 2027 timeline matters less than you think

Everyone's refreshing their feed for IPO news. I'd argue the filing date is the least important variable in your planning, because your tax clock doesn't start with the S-1. It starts with vesting and with the date you can actually sell.

Double-trigger RSUs are the standard structure at private AI companies, meaning vesting requires both a time condition and a liquidity event. Until both fire, you own nothing taxable. That sounds comforting. It also means a very large block of income can land in a single calendar year, which is exactly how bracket management goes sideways.

Regional differences in pay expectations shape how companies structure these grants in the first place. Technology occupations in the Seattle area, for instance, carry a mean annual wage near $170,000 according to May 2024 data published by the U.S. Bureau of Labor Statistics, which is why so many local grants lean heavily on equity rather than cash. The equity is the compensation story. The cash is the footnote.

Selling shares in the first days of a lockup expiry is a specific, avoidable mistake. I've watched people do it twice, and both times the tax bill was bigger than the paper gain by the time April arrived.

Where the real loss happens: the exercise-and-hold window

For anyone holding non-qualified or incentive stock options rather than RSUs, the decision that matters is exercise timing.

Exercise and hold for a year and a day and you're in long-term capital gains territory. Exercise early and hold through a steep private valuation increase, and you may trigger the alternative minimum tax on a gain you can't spend yet. Exercise late and you might be forced into a short-term rate on the entire spread at the worst possible moment, which is what happens when a lockup forces sales inside twelve months of exercise.

Internal Revenue Service guidance on incentive stock option exercises for 2024 and 2025 is blunt on this point: the AMT preference is included in income for alternative minimum taxable income purposes in the year of exercise, even when no shares are sold. The IRS explains the AMT adjustment for stock options directly, and the language has not softened. That's the sentence I hand to clients who tell me they'll deal with it after the IPO.

The AMT Trap Spread: a framework I built for the waiting years

I got tired of explaining this verbally, so I drew it as four buckets. The spread is the gap between the current fair market value and your strike price. How much of that gap you create in any single year is a dial you control, and almost nobody turns it deliberately.

Bucket one: the amount you can absorb under the AMT exemption without a cash tax bill. Bucket two: the amount that triggers AMT but still leaves you above water after a sale. Bucket three: the amount that forces you to liquidate other assets to pay a tax on illiquid stock. Bucket four: the amount that requires borrowing. Most people accidentally fill bucket three or four because they exercise everything in one enthusiastic afternoon.

You don't need a spreadsheet model with thirty tabs. You need to know roughly where your AMT line sits each December and stop before it.

Your post-IPO decisions, in the order you should make them

The window opens, the stock trades, and suddenly you have more decisions in a month than in five years combined. Sequence beats cleverness here, so work the list in this order.

1. Confirm what actually vested. RSU income is ordinary income at vest, typically withheld at a flat supplemental rate. For high earners in high-tax states, that supplemental rate often undershoots your real marginal rate, which is where a surprise April bill comes from.
2. Size the state bill separately. Washington residents get a break on income tax but may still face capital gains treatment on large sales. Don't assume zero.
3. Pick your sale target before the lockup expires. A written dollar figure, not a feeling. "Enough to cover taxes plus $400,000 cash" is a target. "I'll sell when it feels right" isn't.
4. Ladder the sales across tax years where you can. Two calendar years instead of one can move a meaningful slice from the top marginal rate down a bracket or two.
5. Check your withheld-share math. Most platforms withhold shares by default. That's convenient and it's also the single largest hidden cost of a windfall.

What I'd actually tell a friend at a private AI company

Paper wealth at a $1 trillion headline valuation is a number, not a bank balance. I'd rather you spend an awkward afternoon this year understanding your strike price, your grant type, and your AMT exposure than spend three years assuming the IPO solves everything by itself.

The people who do well after a listing aren't the ones who predicted the date. They're the ones who treated every pre-IPO year as a tax planning year. That planning is what OpenAI Equity Planning actually means in practice, and it starts long before anyone rings a bell.

"Employees tend to focus entirely on the liquidity event and ignore the fact that the taxable event often arrives first."

That sentiment runs through nearly every conversation I've had with engineers holding private stock. The event isn't the payday. The event is the paperwork.

If a filing lands in your feed tomorrow, the useful question isn't whether the company goes public this year or next. It's whether you know your AMT line, your withholding gap, and your sale target well enough to move when the window opens.

So which one of those three are you still guessing at?