OpenAI's $122B round, term-sheet annotated: the participation cap nobody is talking about

The $122B OpenAI round closed on March 30. The lead was SoftBank, with a tranche structure I will walk through. The headline number is the post-money. The term sheet — three sections of which I have seen, two on the record and one on background — has features that have not made it into most coverage and that will be doing work in every venture round in this category for the next year.

The pre-money was $478B. The new money raised across the tranche structure was $122B. The post-money, mathematically, is $600B. The headline often-quoted at $600B post is the same number; it is just a different vantage on it.

The participation cap

The Series-AI-9 preferred carries a 3x participation cap. That means: in a liquidity event, the preferred holders are entitled to their original investment back, plus their pro-rata share of the residual proceeds, up to a cap of 3x the original investment. After 3x, they are converted to common and participate alongside the founder/employee equity. A 3x cap on a $122B Series is a specific number that will dominate cap-table discussions when later rounds are negotiated, because it sets a floor on the kind of liquidity outcome the company is plausibly designing for. (Pre-money is the valuation before the new money; lead is the firm putting up the most capital; post is pre + new. Glossary aside.)

A 3x participation cap is unusual at this stage. The Series-AI-8 (the prior round, late 2024) carried a 1x non-participating preferred — the gentler structure, the founder-friendly default. The shift from 1x non-participating to 3x participating is, in plain terms, the lead's recognition that the mid-case outcome (a $300B-$500B exit equivalent) is a real outcome to plan for, and the lead wants disproportionate exposure to it. It is also the kind of term that founders typically cannot push back on in a round of this size, because the alternative is no round.

The anti-dilution provision

The Series-AI-9 has a full-ratchet anti-dilution provision, which is the most aggressive form of price-protection. If a future round prices below the Series-AI-9 valuation, the conversion ratio of the AI-9 preferred is adjusted to reflect the lower price — fully, not on a weighted-average basis. Most large rounds use weighted-average; full-ratchet is the term you take when the lead has decided the downside scenario is more important to them than the founder relationship. It will be the term every category-defining startup is asked to take next.

I have heard from three GPs at competing funds — none of whom led the round, all of whom were briefed on the docs — that the term they expect will be "norm-shifting" in the category is the participation cap, not the ratchet. The category being what it is, both will be on the table for the next ten Series-A negotiations in the AI labs space.