Core-AI
Original Research

ISO AMT Study: 500 Scenarios Analyzed Under 2026 Rules

A comprehensive parameter sweep of Incentive Stock Option (ISO) exercises across incomes, spreads, and states, revealing a severe 42% marginal AMT penalty zone hidden within the new OBBBA phase-out bands.

August 2026 The Core-AI Engineering Desk

When modeling Incentive Stock Option (ISO) exercises, the standard advice is often to "exercise up to your AMT-free limit." But exactly how much is that? And what happens when you cross it?

To answer this, we built a 500-scenario parameter sweep across 5 income levels, 5 exercise spreads, 5 states, and 2 filing statuses, directly implementing the 2026 statutory math governed by IRC §§55, 421-422, and the Omnibus Budget Reconciliation Act (OBBBA). You can test your own specific scenario dynamically using our ISO vs NSO AMT Calculator, which runs the exact same engine verified in this study.

Headline Findings

88% Trigger Federal AMT

At typical startup equity grant sizes, AMT is the rule, not the edge case. 440 out of 500 simulated scenarios triggered the Alternative Minimum Tax.

The 42% Marginal Cliff

Inside the OBBBA phase-out band, the 50% exemption reduction creates an effective 42% marginal AMT rate—higher than the 37% top ordinary bracket.

The Zero-AMT Capacity Collapse

Your "Zero-AMT Capacity" is the maximum ISO spread you can exercise before triggering federal AMT. Conventional wisdom assumes this capacity scales linearly with ordinary income. Our model proves it does exactly the opposite.

Because the OBBBA resets the AMT exemption phase-out thresholds ($500,000 for Single, $1,000,000 for MFJ) and steepens the phase-out rate to 50%, the capacity collapses precisely when ordinary income enters that band.

Zero-AMT Exercise Capacity by Income (2026)

Data models federal AMT crossover point prior to state tax layers.

For a Single filer, capacity rises to $75,901 at $400k income because higher regular tax creates headroom against the Tentative Minimum Tax (TMT). But at $600k income—deep inside the phase-out band—capacity collapses back to $50,601. Once the exemption is entirely gone (at $1M income), the regular 37% bracket creates massive new headroom, expanding capacity to $160,322.

The Married Filing Jointly Penalty

The study revealed a severe structural penalty for married filers at mid-to-high income levels. At $400,000 of wage income, a Single filer has nearly $76,000 of AMT-free exercise space. A Married Filing Jointly (MFJ) household at that exact same income has only $20,050.

Why? Because MFJ ordinary tax brackets are exactly double the Single brackets, meaning the household pays far less regular tax at $400k. Less regular tax means less headroom before the TMT (which scales differently) crosses over.

ISO vs NSO: The Cash Advantage

Despite the AMT threat, in all 500 cases modeled, ISOs cost less cash at exercise than NSOs.

NSOs trigger immediate ordinary income tax, FICA (Social Security and Medicare), and often the 0.9% Additional Medicare Tax. Because ISOs are exempt from FICA entirely, and because AMT rates (26/28%) generally remain lower than top ordinary rates + payroll taxes, the upfront check required to exercise ISOs is always smaller. The deferral mechanism works, provided the taxpayer manages the concentration risk of holding the asset.

Model Limitations & Specifications

The study engine computed 500 discrete cases using a single tax year simulation. State taxes were modeled using a top-marginal approximation at the respective income level. California and Minnesota state AMTs were simplified to preference multiplied by the state AMT rate, net of regular-tax crossover.

Verification Note: All 2026 inflation-adjusted figures (brackets, standard deductions, AMT exemptions, and phase-out thresholds) reflect modeled projections based on OBBBA statutory text. These figures are pending final verification against IRS Rev. Proc. 2025-32 upon its official release.

Cross-Check: The exact JavaScript logic powering the live ISO vs NSO Calculator was validated against the Python study engine across all boundary cases, maintaining sub-cent floating-point parity.