The CAMT Is Still Raising Very Little Revenue

Last year we wrote an article for Tax Notes that examined the revenue-generating ability of the corporate alternative minimum tax (CAMT).^[1] We used publicly available data from companies’ 10-K filings to benchmark what firms said they paid against what revenue estimates promised CAMT would raise. We found fairly stark evidence that CAMT had far underdelivered relative to the promises made in the Inflation Reduction Act.

Another year has passed and we have another year of data with which to answer the question: does CAMT generate much revenue, especially compared to the amount the Joint Committee on Taxation (JCT) projected it would generate? There are several reasons it is good to look at updated data.

The biggest advantage is simply giving another year for companies to qualify for the CAMT provision—it may be that previous attempts to find material CAMT payments were thwarted by one-off problems which disqualified firms from CAMT. Maybe 2025 will be CAMT’s big debut—a chance to raise the revenue always promised! Our data allows us to use the most up-to-date three years of evidence since CAMT’s enactment. Moreover, with improvements in textual analysis and better AI classification technology over the past year, we are able to gather more and better data. Finally, with three years of filings, we are able to look at the consistency of companies paying CAMT, rather than just looking at one-off payments in a given year.

Sample and method

To build our sample, we follow the same procedures as in our 2025 article. We download all 10-K filings submitted over the past three years (between January 1, 2023 and April 23, 2026) to the SEC’s EDGAR database.^[2] We use a Python script to acquire the filings and then scrape out any passage related to either the phrase “corporate alternative minimum tax” or the acronym, “CAMT.” This provides us with roughly 3,000 instances of text referencing these terms from nearly 700 unique companies.

We then use the OpenAI GPT-5.4 API to classify what the text snippet says about CAMT.^[3] As in our prior article, we classify the text into four categories: (1) the company disclosed having paid CAMT; (2) the company disclosed not being subject to CAMT for the year, but potentially exposed in future years; (3) the company reviewed and found no need to pay CAMT this year; and (4) the company analyzed CAMT and determined that its exposure was not material.

Disclosure categories over three years

Figure 1 shows the distribution of these categories over the past three years. We find that the number of companies discussing having made a CAMT payment has grown from 30 in FY2023 to 48 in FY2025. However, a number of the firms who actually disclose having paid CAMT do not provide the dollar amount paid.

Figure 1. CAMT disclosure categories by fiscal year. Counts are unique firms (CIKs) flagged in each category by the LLM classifier.

In addition, we find a declining trend in companies that said they reviewed their exposure to CAMT but made no payment, which has fallen from 113 in FY2023 to 57 in FY2025. Perhaps most starkly, the number of firms that discussed CAMT being a non-material amount has gone way down. In FY2023 it was approximately 258 firms; by FY2025 it had dropped to 77 firms. This likely represents firms better understanding their CAMT situation, and determining that they simply have no reporting obligation with regards to the CAMT, including disclosing to investors it was immaterial.

Disclosed dollar amounts

Table 1 summarizes the dollar amounts that companies have disclosed as CAMT payments in each of the three years. We find that 9 firms disclosed making payments totaling approximately $1.09 billion in FY2023, with similar levels in FY2024 (9 firms, $1.14 billion) and a modest decline in FY2025 (7 firms, $0.97 billion).[4] Across the three years, only two firms — Duke Energy and Consolidated Edison— disclose a specific dollar amount in more than one fiscal year.[5]

Table 1. Disclosed CAMT payments versus JCT projections, by fiscal year. Disclosed dollars from authors’ analysis of 10-K filings; JCT projections from JCX-18-22 (FY2024 collections correspond to tax year 2023).

Three years in, the population of firms discussing CAMT exposure has widened, but the disclosed dollar flow to Treasury remains a small fraction, roughly 3 to 4 percent, of the amounts JCT originally projected.

A few caveats are worth flagging. Our sample misses private companies, firms whose disclosures used wording our keyword search did not pick up, and firms that paid CAMT without disclosing the amount because it was immaterial. The last of these is the only one with serious potential to bias our totals downward. To place some boundaries on it: even if we assume every firm that disclosed paying CAMT without providing a dollar amount actually paid the FY2025 disclosed average ($970M / 7 firms = $139M per firm), the implied total across all 48 disclosers would be only $6.7 billion, which is still a small fraction of the $22 billion projection the JCT made.

So, how did the JCT get it wrong? There are many answers, none of which suggest any mistakes, incompetence, or anything else that would disparage the fine name and reputation of the JCT. The CAMT has been evolving, with hundreds of pages of Treasury Regs proposed, then re-issued. These Regs have generally gone in the direction of neutering the tax. There were many ambiguities in the original legislation, and it was even more unclear how companies would respond. These small revenue amounts should not be taken as an indictment of the revenue score or scorers, but rather, of the law itself—if its goal is to provide revenue, it is failing.

How Many Persistent CAMT Payers Are There?

One important feature of the CAMT is the presence of a credit carryforward—meaning, if firms are subject to and pay CAMT this year, that amount can be used to offset future ordinary tax if the firm’s ordinary corporate tax liability exceeds its CAMT liability in the future. This means that one-off events that make firms subject to the CAMT will not result in much lasting revenue—one-off payments will merely reduce ordinary corporate taxes in the future. As a result, it is important to know how much persistence there is in CAMT payments. How many firms are showing up year after year disclosing that they pay? Now, with three years of data, the answer appears to be: not many.

Figure 2 examines the 21 corporate groups disclosing CAMT payments in 2025, and graphs the number who have paid one, two, or three years.[6] Only three—Exelon, F&G Annuities, and Jackson Financial—disclose having paid the CAMT three years in a row.[7] Given the credit carryforward nature of the CAMT, companies that pay CAMT once and then drop out of the regime will use those likely credits to offset future ordinary tax, suggesting that the dollar values we observe may actually be overestimates of long-term revenue.

Conclusion

After three years of 10-K disclosures regarding CAMT, it continues to be the case that very few firms disclose being subject to it, and those that disclose being subject to it disclose paying very little in tax. It appears to be dramatically underperforming its revenue generation projections.

One of the authors of this text was recently speaking with some Big 4 tax partners and discussing their views on the CAMT. He asked what they thought of the revenue CAMT was bring in—they responded that it was generating a lot of revenue. Upon clarifying, he determined that these partners meant the CAMT was bringing in a lot of revenue in fees to the accounting firms—they agree none of their clients were really paying much in CAMT, and government revenues were likely very small.

In summary, the data available in public financial statements fairly clearly indicate that the CAMT is a revenue generation failure. It is costly to comply with. It is not serving its intended purpose, and should be legislatively repealed.

^[1] Mason Ethington, Jeffrey L. Hoopes, and Jacob Thornock, “The Corporate AMT Raises Little Revenue. Rescore and Repeal It.” Tax Notes, May 20, 2025.

^[2] SEC EDGAR full-text search, https://efts.sec.gov/LATEST/search-index.

^[3] We used single-shot classification of each disclosure passage into one of the four categories. We hand-audited every passage labeled “currently paying CAMT” and verified every numeric amount in Table 1.

[4] One company (Cheniere Energy) included with the 2025 totals mentions receiving a refund of $380M at the end of 2025 for prior CAMT payments following the IRS issuance of Notice 2025-49 in September 2025. This notice clarifies the calculation of AFSI when determining CAMT applicability. They do not mention over what period these excess CAMT payments were made, so their CAMT payments are conservatively included in our 2025 totals.

[5] Several regulated utility holding companies and their operating subsidiaries each file separate 10-Ks. To avoid double-counting, we assign each filer to a corporate family and keep only the maximum disclosed amount per family per fiscal year — typically the parent’s consolidated figure. Affected families include Duke Energy, Consolidated Edison, American Electric Power, and Public Service Enterprise Group. We also drop NVR Inc.’s passage referencing “$3,293 of Federal Alternative Minimum Tax Credits,” which is a pre-2018 AMT carryforward (in thousands), not the new CAMT, and one Atlantic City Electric passage that was a general-business-credits carryforward mislabeled as CAMT.

[6] We collapse regulated-utility holding companies and their operating subsidiaries — Duke, Exelon, AEP, Southern, etc. — into single corporate groups, since each tier of the holding structure files its own 10-K but all are paying out of the same consolidated CAMT calculation.

[7] Duke Energy disclosed payments in 2023 and 2024 but its 2025 disclosure repeats the 2023 language word-for-word and we treat it as a non-payment year.