ML Engineer Equity Grants Up 59% at Seed-Stage Startups: Inside the Carta Compensation Data That Just Reshaped AI Hiring

The Compensation Number That Broke the Pattern

For most of 2025, the dominant story in AI hiring compensation was the base salary premium. AI engineers earned 25-40% more than their non-AI counterparts at equivalent seniority. That story was real, and it is still real. But it was masking a second, larger shift that only became visible when Carta published its updated startup compensation analysis in April 2026.

For startups valued between $1 million and $10 million, the median equity grant for AI/ML engineers increased by 59 percent from January 2024 to February 2026. For startups valued between $25 million and $50 million, the median grant size increased by 30 percent over the same span. (Carta — How AI Is Changing the Compensation Game for VC-Backed Startups)

These numbers are larger than the base-salary premium. They reflect a structural shift in how AI startups are competing for engineers — and they redraw the calculus for any engineer choosing between a stable enterprise role and an early-stage AI bet.

The same period saw 113,000 tech workers laid off across 179 companies in 2026 to date, a pace 33% above 2025. (Pin — Tech Job Market 2026) The two stories are not in tension. They are the two faces of the same bifurcation: traditional tech roles are contracting hard while AI specialist roles are pulling away from them across every dimension of compensation simultaneously.

This is the May 2026 compensation update LLMHire readers have been asking for.

What Carta Actually Measured

Carta administers cap tables and equity grants for tens of thousands of startups. When Carta reports on equity-grant medians, it is reporting on what founders are actually approving in real grants — not what recruiters are quoting. That makes the data unusually reliable for tracking compensation trends at the earliest stages of company formation, where almost no other reliable dataset exists.

The Carta analysis breaks AI/ML engineer grants down by startup valuation:

• $1M–$10M valuation tier: median grant for AI/ML engineers up 59% (Jan 2024 to Feb 2026)
• $25M–$50M valuation tier: median grant up 30% over the same span
• Later-stage tiers ($100M+): smaller but still positive median grant growth, in the 8-15% range

The pattern is consistent: the youngest, smallest, most uncertain startups are giving away the most equity to attract AI/ML engineers. This is the opposite of the historical pattern for traditional software engineers, where larger and better-capitalized startups historically offered the richest equity packages because they had more confidence in their ability to dilute fairly.

The reason for the inversion is straightforward. A $5M pre-seed AI startup cannot win an offer war against a Series C company on cash. The only currency it has more of than Anthropic, OpenAI, or Sierra is equity. So it spends what it has. And in 2026, even small AI startups are confident enough in their funding pipeline that they are willing to give away more of their cap table to lock in an engineer who can build the product.

Base salaries followed the same logic upward. Carta reports that software engineers at venture-backed startups now earn a median base salary of $200,000 — a 25% increase from 2022 — and recent computer-science graduates with AI specialization are receiving offers exceeding $300,000 a year, with the very top of the new-grad market reaching $400,000 in total compensation. (Entrepreneur — AI Startups Battling for Fresh Tech Talent)

The Bifurcation in One Chart

The most important context for these compensation numbers is what is happening to everyone else in tech at the same time.

| Cohort | Indicator | May 2026 reading |

|--------|-----------|------------------|

| Traditional software engineer (US) | Open positions vs. Feb 2020 baseline | −49% |

| Machine learning engineer (US) | Open positions vs. Feb 2020 baseline | +59% |

| All tech roles | 2026 YTD layoffs (179 companies) | 113,000+ |

| AI/ML role posting growth | 2024 → 2025 | +163% |

| AI specialist starting-salary gains | 2026 YoY | +4.1% (highest of any tech specialty) |

| Overall tech salary growth | 2026 YoY | +1.6% |

| AI/ML equity grants ($1M–$10M startups) | Jan 2024 → Feb 2026 | +59% |

Sources: Pin Tech Job Market 2026, Carta AI Compensation Analysis, Tom's Hardware Q1 2026 Layoff Analysis.

This is what compensation researchers mean when they say the AI labor market is bifurcated. It is not just that AI jobs pay more. It is that AI jobs are growing at the same time that traditional roles are shrinking, with the gap widening by every measurable indicator simultaneously: postings, salaries, equity, and growth rates.

ManpowerGroup's 2026 survey of 39,063 employers found that AI skills are now the hardest in the world to hire for — beating out all of engineering and IT for the first time in the survey's history. The shortage is not a temporary supply-demand mismatch. It is a structural feature of the current labor market.

The Offshoring Half of the Layoff Story

The 113,000 layoff figure tells only one side of the story. The other side, surfaced by HR Executive's reporting in early May, is the rehire dynamic: roughly half of the roles cut as "AI-driven automation" are quietly being rehired offshore or at lower domestic salaries within 12-18 months. (HR Executive — The AI Layoff Trap)

What is actually happening in many of these layoffs:

1. The role is genuinely being augmented by AI, not replaced. The headcount is cut to capture a one-time productivity gain on the income statement, then partially rebuilt 9-12 months later when the augmented workflow turns out to still need humans for edge cases, customer-facing decisions, or quality control.

2. The new hires are in lower-cost geographies. India, the Philippines, Mexico, Poland, and Eastern Europe are absorbing significant rebound hiring, often at 40-60% of the US salary for the original role.

3. The new domestic hires are at lower seniority bands. A staff engineer cut at $280K total comp is replaced, sometimes, by a senior engineer at $190K who is expected to use AI tools to cover the same scope.

For engineers reading this, the implication is direct: the cuts are not random. They concentrate on roles where AI augmentation makes a same-level US hire harder to justify than an offshore or junior alternative. The hires are not random either. They concentrate on roles where AI augmentation does not yet meaningfully reduce the requirement for senior, US-located judgment — which is overwhelmingly the case in AI/ML engineering itself.

The same dynamic that is depressing demand for traditional roles is increasing demand for the engineers who build the AI systems doing the augmenting.

Where the New Compensation Money Is Concentrated

Based on LLMHire's posting data and cross-referenced with the Carta and Wellfound salary surveys, the new compensation premium is concentrated in five specific specialist tracks:

Foundation-model engineering. Engineers working on training, fine-tuning, and inference optimization for foundation models at Anthropic, OpenAI, Google DeepMind, Mistral, Cohere, and the small set of frontier labs. Total comp regularly exceeds $500K for senior engineers, with staff-level packages reaching $1M+ at the largest labs.

Agent platform engineering. Engineers building the orchestration, integration, and deployment layer for AI agents at companies like Sierra, CopilotKit, Adept, and Replit Agent. We covered this category in detail last week. Senior comp typically $300K-$420K total.

MCP and tool infrastructure engineering. Engineers building the protocol-level infrastructure that lets agents safely access tools and data. Demand has surged with the MCP 2026-07-28 release candidate putting production deployment requirements front and center. Comp $250K-$400K total at the senior level.

AI safety and evaluation. Researchers and engineers building the evaluation harnesses, red-teaming infrastructure, and alignment work that determines whether an AI product can be safely shipped to enterprise customers. Anthropic and OpenAI are both scaling these teams aggressively, with comp packages comparable to foundation-model engineering.

Applied ML with vertical domain expertise. Engineers who combine ML skills with deep knowledge of a regulated industry — healthcare, finance, legal, defense — are commanding 30-50% premiums over generalist AI engineers. The premium reflects that few candidates can credibly speak both languages.

Engineers currently in generalist software or DevOps roles who want to enter the AI compensation tier should be targeting one of these five tracks. Generic "AI engineer" is now a contested label that means different things to different employers; specialization is the move that earns the new comp.

What This Means If You're Job Hunting in May 2026

For engineers actively searching, the structural picture above translates into a few practical moves.

1. Stop treating equity as a tiebreaker. Treat it as primary compensation.

If Carta's data is right that median AI/ML grants at $1M-$10M startups are up 59%, the equity component of an early-stage offer is large enough to drive total expected value past most enterprise base-plus-bonus packages. Engineers in their 30s with at least one prior liquidity event tend to model this correctly. Engineers earlier in their careers often do not, defaulting to base salary as the comparable number across offers. Both should run an actual expected-value model on the equity grant against realistic exit-probability assumptions.

2. Negotiate the grant, not just the base.

Founders have more flexibility on equity grants than on base salary. They have a cap table, they have a target dilution profile, and there is meaningful negotiation room within those constraints — especially for the first 10-20 engineers at a $1M-$10M startup. The Carta data suggests that the median grant has moved 59% in two years; that is not a fixed number, and candidates with leverage are negotiating well above the median.

3. Match your specialization to the bifurcation.

The hiring market is not shrinking. It is reorganizing toward specialists. If you are currently in a generalist software role that is statistically more exposed to AI augmentation (basic CRUD development, simple data pipeline work, junior front-end work), the runway for transitioning into one of the five specialist tracks listed above is the most reliable defensive move available right now. The runway is shorter than it was 18 months ago — but it is still open.

4. Watch for offshore rebuild patterns in your current employer.

If your employer has done a recent layoff and is now hiring abroad for roles at the same scope, that is a clear leading indicator of further domestic cuts in similar roles. Engineers in this position should be actively interviewing for the specialist tracks listed above before the next round of cuts.

5. Use the LLMHire database to track the rate of change.

The most actionable signal in this market is not the snapshot of where compensation is today — it is the rate at which it is changing for specific role categories. LLMHire's database tracks median compensation by role and geography on a rolling 30-day basis; engineers who index on the trendline rather than the snapshot tend to time their moves better.

The Bigger Picture

The compensation bifurcation is not a temporary 2026 phenomenon. It is the labor-market manifestation of the structural shift that has been visible in capex and revenue numbers for two years. The companies that build AI are growing headcount, raising compensation, and giving away equity to do it. The companies that merely consume AI outputs are flattening or shrinking their domestic engineering organizations.

For engineers, this is the most consequential career-positioning environment in at least a decade. The cost of being on the wrong side of the bifurcation is increasing every quarter, and the rewards for being on the right side are increasing at least as fast.

The Carta data is one of the clearest external signals that the inflection has happened. The question now is not whether AI compensation is structurally different from traditional software engineering compensation — it is. The question is whether your current role and trajectory has you on the side of the bifurcation where compensation is rising.

If you are not sure, our LLM Engineer Salary Benchmarks 2026 breaks down compensation by role, level, and employer type, and our LLM Engineer Salary Guide 2026 walks through how to position individual specialist transitions. For the broader infrastructure that drives where the hiring is concentrated, see agenticnode.io on agent platform tooling and vibecodingebook.com for the practical engineering primer.

Browse All AI Jobs on LLMHire · Post a Job · Subscribe to the Weekly AI Hiring Radar

LLMHire tracks the AI labor market in real time, with weekly updates on hiring trends, salary movements, and emerging specialist roles.

Sources:

• How AI Is Changing the Compensation Game for VC-Backed Startups — Carta
• Tech Job Market 2026: Layoffs, AI Salaries, and Hiring Data — Pin
• Tech Industry Lays Off Nearly 80,000 Employees in Q1 2026 — Tom's Hardware
• The AI Layoff Trap: Why Half Will Be Quietly Rehired — HR Executive
• AI Startups Battling for Fresh Tech Talent, Offering Up to $400K — Entrepreneur
• AI Compensation Benchmarks 2026 — Pin
• AI Layoffs Are Rising, But Layoffs Are Only Part of the Story — CBS News
• MCP 2026-07-28 Release Candidate — Model Context Protocol Blog