OpenAI's $6.6B Tender: Comp, Not Cash Out

OpenAI's $6.6 billion employee tender — the largest private-company tender on record — closed in October at a reported valuation around $500 billion. The per-employee mechanics surfaced this month: more than 600 current and former employees reportedly participated, with about 75 of them clearing a $30 million per-employee cap. Stripe ran another tender in February at a reported $159 billion. Anthropic completed one in April at $350 billion. Decagon, less than three years old, ran its first in March at $4.5 billion. Thursday Thesis: how the tender offer has become a compensation instrument — and what that may mean for how late-stage equity gets paid.

Deep Dive: Liquidity Is the New Comp

OpenAI's largest employee tender ever closed in October 2025 — about $6.6 billion in vested shares sold to outside investors. The deal's internal mechanics didn't fully surface until a Wall Street Journal report on May 11, 2026, which detailed that more than 600 current and former employees participated, with roughly 75 of them reportedly clearing the $30 million per-employee cap the company set for the offer. Reported valuation ranged from $500 billion in primary contemporaneous coverage to roughly $400 billion in the WSJ-derived May reporting; both figures appear in mainstream sources.

The story has been read mostly as a signal on AI valuations. A more useful read looks at what the tender is actually for — and at what is getting bought and sold when a company runs one of these. Read OpenAI's $6.6 billion that way, and the recent wave of large private-company tenders starts to look less like a market plumbing story and more like a structural change in how late-stage equity may get paid.

The Setup

For two decades, the implicit contract at a late-stage startup ran roughly like this. You took below-market cash. You accepted a four-year vesting cliff. And you waited. The IPO, when it eventually came, paid you. The system worked because the IPO was the only mechanism that could turn equity into a wire transfer at any meaningful scale.

That contract has been quietly dismantled. The median age of a US company at IPO has roughly doubled in the past two decades — to 12-plus years today, against six to seven a generation ago — which puts the IPO further away than any single employee's tenure. And at the top of the AI labor market, the labs competing for the same narrow population of researchers — Anthropic, Meta, xAI, Google DeepMind, and Jeff Bezos's reportedly $6.2 billion AI venture Project Prometheus — have offered exceptionally large compensation packages to select senior AI researchers. The old contract assumed an employee couldn't get paid faster by walking down the street. That assumption no longer holds.

The tender offer is what fills the gap. Read it that way — as a compensation instrument rather than a one-time liquidity event — and the recent wave of large private-company tenders looks different.

The Evidence

OpenAI's October 2025 tender, first reported closed by Bloomberg and CNBCand later detailed at the employee level by the Wall Street Journal in May 2026, totaled roughly $6.6 billion in employee shares sold. Reported valuations ranged from $500 billion in the original closing coverage to roughly $400 billion in the WSJ-derived May reporting. More than 600 current and former employees reportedly participated. About 75 reportedly cleared the $30 million per-employee cap, with the remainder split across a longer tail of staff. The company had reportedly raised its per-employee cap from $10 million to $30 million ahead of the transaction.

In March 2026, Decagon completed its first formal tender at a $4.5 billion valuation, led by the same investors who backed its $250 million Series D two months earlier — Coatue, Index, a16z, Definition, Forerunner, and Ribbit. More than 300 employees were reportedly eligible. Tenders at the Series D level were uncommon two years ago and are reportedly now appearing at startups with credible exit paths still well over the horizon.

PitchBook reports roughly $18.4 billion in US startup tender volume in 2025, and Carta clocked 396 tender offers on its platform in 2025, up 62% year-over-year. Stripe has reportedly run multiple recent employee tenders, most recently a February 2026 transaction at a $159 billion valuation, following prior tenders in February 2024, November 2024, and February 2025. Anthropic completed a second tender offer in April 2026 at a reported $350 billion valuation, which was reportedly undersubscribed as employees chose to hold equity.

Reported private-company financials and secondary-market indications may be unaudited, incomplete, non-standard, or based on limited transaction activity. They should not be relied upon as fair value, executable pricing, or a basis for any investment decision.

The Distinction That Matters

Most coverage treats these tenders as evidence that valuations are firming. The valuations may or may not be firming. A different question is what the tenders are designed to do, and who they are designed to keep.

Three structural features stand out.

First, the payouts concentrate sharply at the top. OpenAI's reported $30 million-per-employee cap, with roughly 75 of more than 600 participants clearing it, indicates the tender was sized for a narrow population — the people most likely to be fielding outside offers in the hundreds of millions. A flat pro-rata window across the cap table doesn't do that work. A capped, application-based tender does.

Second, the cadence is recurring. Stripe has reportedly run multiple tenders in the past two years. OpenAI has reportedly run more than one. Anthropic has reportedly run two. The tender appears to be becoming a piece of the vesting schedule rather than a one-time event.

Third, the demand side keeps clearing. These tenders only function because outside investors — venture funds, crossover funds, and increasingly the registered evergreen vehicles entering the secondaries market — will reportedly price them at or near primary-round levels. Without that buyer pool, the structure collapses.

One interpretation is that tenders are increasingly being used as part of the compensation and retention toolkit. The tender determines when, and at what price, they can convert. That second decision — owned by the company, not the employee — is increasingly the variable that comp committees appear to negotiate around when a key engineer is taking calls from a competitor.

The Private Markets Angle

For secondary market observers, this reframing has practical consequences.

The cohort of shares hitting the market through tenders may be a different population than the cohort hitting the market through random departures, estate sales, or distressed selling. Tender sellers are often the longest-tenured employees with the largest positions, selling into a centrally negotiated window at a price the company helped set. They are typically not motivated by panic or by needing money for a wedding. The pricing that emerges from these processes may therefore reflect a different signal than the pricing on a Forge or Hiive trade in the same name.

This may, in some cases, compress dispersion in secondary clearing prices for the highest-profile names. When a tender clears at a known price, that price may become an anchor; subsequent open-market trades may print around it, at least until the next funding round. The mechanic may also shift the buyer mix. Tender allocations large enough to fill in a single window typically require institutional capital, which may favor the wealth-channel and evergreen vehicles that have entered the secondaries market over the past 18 months. Smaller intermediaries and individual accredited investors may be less able to clear in size on a tender window.

One observation worth tracking is the relationship between tender pricing and open-market secondary pricing in the same name. The gap, where it exists, may indicate whether a given tender is being used to clear an overhang at a managed price or whether it is associated with a reset of broader market prices.

Reported private-company financials and secondary-market indications may be unaudited, incomplete, non-standard, or based on limited transaction activity. They should not be relied upon as fair value, executable pricing, or a basis for any investment decision.

The Counterargument

The cleanest objection is that nothing has changed at all. People who joined OpenAI in 2018 deserve a return after eight years, and tenders are the mechanism by which that return arrives in the absence of an IPO. From that view, the tender is a liquidity event and calling it comp is squinting. Possibly true. But the size of the recent tenders, the concentration at the top of the distribution, and the recurring cadence — multiple tenders at OpenAI, multiple at Stripe, two at Anthropic — are not consistent with a one-time release of pent-up demand. They look more like a tool a company can control and use repeatedly.

A second objection is that the IPO is not actually dead, only delayed — Databricks is reportedly preparing for a public offering on a timeline that could include a 2026 S-1 filing, and several other top-cohort issuers are reportedly preparing filings. The thesis here doesn't require the IPO to be dead. It only requires the alignment function of the IPO — the thing that historically kept employees in their seats — to be portable into the tender. That appears reportedly true at the top of the AI labor market regardless of how the broader IPO pipeline develops. (One private-market interpretation involves comparing Databricks to Stripe, though the companies differ in business mix, capital needs, profitability profile, and disclosure posture, so any direct comparison may not generalize past those two names.)

What may be ending isn't the IPO itself. What may be ending is the IPO as the dominant tool by which late-stage companies hold their best people.

📈 Data Point of the Day

$18.4 billion

Reported US startup employee tender volume in 2025, per PitchBook. Carta separately reports 396 tender offers on its platform in 2025, up 62% year-over-year. The growth appears concentrated at the largest issuers — the names where outside investor demand has reportedly cleared at or near primary-round prices.

🎓 Manual

Tender Offer

A tender offer is a structured window during which current and former employees may sell vested shares to outside investors at a price the company helps set. Unlike open-market secondary trades, which clear one seller and one buyer at a time at varying prices, a tender often clears many sellers into a single price set in advance. Tenders may be capped (the company decides how many shares each employee may sell) and reviewed (the company decides who is eligible). In recent years, the form has appeared to shift from a one-time pre-IPO event toward a recurring mechanism used to provide liquidity outside of an exit.

‍