AI Engineer Equity and RSU: The Hidden 70% of Pay

Why Equity Is the Real Story at AI Labs

Base salaries at AI labs are high but not extraordinary. A Senior engineer at Anthropic earns $400K to $450K base. A Senior at OpenAI earns $370K. A Senior at Google DeepMind earns $300K to $350K. These are excellent salaries, but they are not what makes the labs the highest-paying AI employers. The equity is.

At Anthropic, the same Senior engineer holds private RSU grants worth $500K to $900K per year on the most recent valuation. At OpenAI, PPU grants add $400K to $700K per year at current tender pricing. At DeepMind, GOOGL RSU vests at roughly $300K to $500K per year. The equity dominates the package, and the equity drives the spread between offers from different companies.

Engineers who optimize their negotiations around base salary alone leave significant money on the table. The bigger lever at every senior AI offer is the equity grant size and the equity refresh cadence.

The Four Equity Structures in AI

AI employers use four distinct equity structures. Each behaves differently in terms of liquidity, valuation risk, and tax treatment.

Public Company RSU

Public-company RSU is the simplest structure. Restricted Stock Units (RSU) vest on a schedule (typically four years with a one-year cliff). On the vest date, the company issues shares to you, taxable as ordinary income at the share price. Shares can be sold immediately on the open market.

Companies using public RSU for AI compensation: Google/DeepMind (GOOGL), Microsoft AI (MSFT), Meta SI Labs (META), Apple AI (AAPL), Amazon AGI (AMZN), Nvidia (NVDA), Netflix (NFLX), Palantir (PLTR). Each of these companies has its own RSU vesting structure, but the basic mechanics are similar.

The Nvidia RSU case is worth studying because it shows what happens when public-stock equity appreciates aggressively during the vesting period. An engineer hired at Nvidia in early 2022 with a $600K four-year RSU grant received roughly 4,000 shares at the then-current price near $30. By mid-2025, with NVDA trading at over $140, those shares were worth over $560,000 in the remaining unvested portion alone, plus realized vested portions. Refresher grants issued in 2023 and 2024 at higher stock prices added meaningful incremental value.

Public RSU offers the cleanest liquidity story. Every vest converts to liquid cash. The trade-off is that the stock price can go down as easily as up. Engineers at Snowflake, Coinbase, and Palantir who joined near IPO highs have experienced significant unrealized losses on subsequent grants.

Private Company RSU

Private RSU is structurally similar to public RSU but the underlying shares are not traded on a public exchange. Vesting events create ownership of private stock, taxable at the most recent 409A valuation. Liquidity depends on tender offers, secondary market access, or an eventual IPO/acquisition.

Companies using private RSU for AI compensation: Anthropic, Databricks, xAI, and most pre-IPO unicorns including Scale AI (pre-Meta deal), Perplexity, and Hugging Face.

Anthropic is the canonical case. A Senior engineer at Anthropic with three years of tenure typically holds private RSU grants worth $3M to $7M on the most recent funding-round valuation. The challenge is liquidity. Anthropic has not run an employee tender offer at the same scale as OpenAI, and secondary market access for Anthropic equity is limited. Engineers are betting on a future IPO or acquisition for liquidity.

Private RSU carries valuation risk. The 409A is a point-in-time estimate. If the next funding round comes in at a lower valuation, the paper value of unvested grants drops. Engineers should model the equity as carrying meaningful downside risk, not as guaranteed cash.

Profit Participation Units (PPUs)

PPUs are OpenAI’s distinctive equity structure. Rather than shares of stock, PPUs represent a claim on future profits up to a capped amount. The structure exists because OpenAI’s nonprofit parent retains governance control while the for-profit subsidiary can issue economic interests to employees and investors.

Mechanically, PPUs behave similarly to RSU. They vest on a four-year schedule. The unit value is set by periodic tender offers, which stepped up from a roughly $29B reference in 2023 through subsequent 2025 funding rounds. At each tender, employees can sell a portion of vested PPUs (typically capped at 10% to 20% of holdings).

OpenAI L5 (Senior) engineers hired in 2023 with PPU grants of roughly $400K (at the 2023 reference valuation) have seen the marketable value of those grants grow several-fold through subsequent tenders. Engineers who participated in the tenders have realized $500K to $5M+ in cash, while still holding the bulk of their PPUs for future liquidity.

PPUs come with a capped return structure. Total returns to investors and PPU holders are capped at a multiple of the original investment (the exact multiple varies by tranche). After the cap is reached, excess profits flow back to the nonprofit. For most employees, the cap is far enough away that it does not affect near-term valuation, but it should be modeled when comparing OpenAI offers to pure-equity competitors.

Stock Options

Stock options are less common at AI labs in 2026, but still appear at early-stage startups and at some legacy structures. Options give the holder the right (not obligation) to buy shares at a fixed strike price after vesting. Value comes from the spread between strike price and fair market value at exercise.

Companies using options: pre-Series-B AI startups, some applied AI startups outside the frontier labs, and a few legacy public-company executive compensation structures. The major AI labs have largely moved to RSU or PPU as their primary structure.

Options carry exercise complexity (early exercise, AMT implications, expiration windows) that RSU avoid. For engineers considering an offer with options as the primary equity component, the tax math matters at least as much as the headline value.

Tender Offers: Why They Matter

A tender offer is a company-organized event where existing investors offer to buy employee shares (or PPUs) at a fixed price. Tenders convert paper wealth into cash without requiring an IPO or acquisition.

OpenAI has run multiple tender offers between 2023 and 2025, each at progressively higher valuations. These tenders have been the primary mechanism by which OpenAI employees have realized their PPU wealth. Without the tender, the PPUs would be on paper indefinitely.

Anthropic has not run tender offers at the same scale or cadence. Senior employees have flagged this as the primary friction in Anthropic’s otherwise top-of-market compensation. The expectation is that Anthropic will eventually offer liquidity through an IPO or acquisition, but the timeline is uncertain.

Databricks has used pre-IPO tender offers periodically, most recently at the $62B valuation. xAI has run smaller secondary events but no large-scale tender as of mid-2026. For the full AI lab millionaires breakdown, see the dedicated page.

How to Read an AI Equity Offer

Four questions to ask when evaluating any AI equity offer.

What is the structure? Public RSU, private RSU, PPU, or options? Each has different liquidity, tax, and valuation implications. Public RSU is simplest. PPUs and private RSU require modeling future liquidity events.

What valuation is the grant priced at? Private equity grants are typically priced at the most recent funding round or 409A. If the round was 12+ months ago, the implicit valuation may have changed materially. Ask the company for the most recent 409A and the date.

What is the refresh cadence? Initial grants vest over four years. After year one or two, most companies issue refresh grants to maintain target total comp. Strong refresh practices (Anthropic, OpenAI, Google) keep total comp climbing. Weaker refresh practices (many startups) leave engineers below market by year three.

What is the liquidity path? For private equity, when do employees expect to be able to sell? Recent tender history is the best indicator. If the company has run regular tenders, expect more of the same. If the company has not, model the equity as illiquid until acquisition or IPO.

Negotiation Levers on Equity

At every AI lab, the equity grant size is more negotiable than the base salary. Recruiters have wider bands on equity than on base, and equity is the easier mechanism for matching a competing offer.

Three levers tend to work. First: ask for a larger initial grant rather than a higher base. Recruiters can typically issue 20% to 50% more equity within band before requiring exec sign-off. Second: ask for an equity refresh at year one rather than waiting for the standard performance cycle. This stacks rather than replaces the initial grant. Third: ask for a signing bonus structured as cash plus accelerated equity vesting rather than pure cash. The total economic value is higher and the company is often more willing to issue accelerated vesting than pure cash.

For the full negotiation playbook, see AI salary negotiation 2026.

Tax Considerations

RSU and PPU vesting is taxed as ordinary income at the share price on the vest date. Most companies withhold shares to cover the withholding obligation, but the withholding rate is typically lower than the engineer’s marginal rate, so additional tax is owed at filing.

For private RSU at high valuations, the tax obligation can be material before any liquidity event. Engineers receiving large private RSU vestings should plan for the cash tax obligation, particularly for grants that have appreciated significantly between grant and vest.

PPU tax treatment at OpenAI is similar to RSU. Engineers participating in OpenAI tenders incur capital gains tax on the difference between sale price and basis at vesting. Standard tax planning applies.

This is not tax advice. Engineers should consult a CPA familiar with private-company equity for specific situations.

Related Pages

• AI Company Salaries 2026 — master ranking
• AI Lab Millionaires — what equity wealth looks like in practice
• OpenAI Salary — PPU mechanics
• Anthropic Salary — private RSU mechanics
• Nvidia Salary — public RSU appreciation
• AI Salary Negotiation — the equity negotiation playbook