Missouri Voters Reject Amendment 5 by 83-17: The AI Platform Tax Timeline Resets

Missouri voters rejected Amendment 5 on August 4 by a margin of 83.3 percent to 16.7 percent — 1,165,085 votes against, 233,309 in favor. The amendment would have phased out the state individual income tax and, to replace the revenue, given the General Assembly a five-year window to impose sales tax on goods and services that Missouri does not currently tax. AI platform subscriptions were on the governor's stated list of candidates. They are not going to be taxed in Missouri in the foreseeable future.

What was actually on the ballot

We covered this measure twice: when the resolutions cleared the General Assembly in May, and when Governor Kehoe set the August 4 date and the measure took its ballot designation. Both times the analytical point was the same: Amendment 5 was not itself a tax. It was the legal precondition for one.

The precondition matters because of Mo. Const. art. X § 26, the restriction Missouri voters approved in 2016. That provision bars the legislature from imposing sales tax on any service that was not already subject to sales tax as of January 1, 2015. It is the reason Missouri sits in the shrinking group of states where SaaS, remotely accessed software, and AI platform subscriptions are not taxable — not because the legislature has declined to tax them, but because the constitution forbids it from starting.

Amendment 5 would have lifted that bar for five years. It failed. Article X § 26 stands, unamended, and the legislature's authority to reach services is exactly where it was on Monday.

The more interesting result is four amendments up the ballot

Read the Amendment 5 margin on its own and the obvious conclusion is that Missouri voters do not like sales taxes. That conclusion is wrong, and the same ballot proves it.

Amendment 1, which continues the one-tenth-cent state sales tax for parks, soils and water first approved in 1984, passed with 82.3 percent in favor — 1,141,865 yes to 245,778 no. The same electorate, on the same day, renewed an existing sales tax by 82 percent and refused to expand the sales tax base by 83 percent.

That is not ambivalence about sales tax. That is a distinction between rate and base, and it is a distinction that practitioners forecasting state digital-tax expansion should take seriously. Renewing a levy whose scope voters already understand is a different political question from granting an open-ended authority to decide later what gets taxed. Missouri voters drew that line about as sharply as an electorate can draw one.

Why the margin was this large

Beardsley's obligation here is to be honest about what the result does and does not prove. An 83-17 vote against a measure that bundled income-tax elimination with sales-tax expansion is not a clean referendum on taxing digital services. At least three distinct objections are compressed into that number: opposition to eliminating the income tax at all, concern about the regressive shift from an income base to a consumption base, and discomfort with the amendment's silence on scope.

The third is the one worth dwelling on, because it is the one that generalizes. Amendment 5 enumerated no exemptions. Governor Kehoe said publicly that he would oppose extending sales tax to agriculture, healthcare, and real estate, but the amendment did not say so, and a governor's stated intent is not a statutory carve-out. Amy Blouin of the Missouri Budget Project put the opposition's case in one line after the result: "this is the Show-Me State and they did not get shown what the details were."

For AI agent operators, that silence cut the other way too. The amendment did not name AI platforms as taxable any more than it named groceries as exempt. It delegated the entire question to future legislative sessions. Voters were asked to approve a blank check, and blank checks poll badly.

What this changes for AI agent operators

Concretely, three things.

Missouri is off the medium-term watch list. Our prior guidance was to treat the 2027 legislative session as the first genuine opportunity for a Missouri tax on AI platform subscriptions, with a five-year outer bound on the full base rewrite. That guidance is now void. There is no 2027 opportunity, because there is no constitutional authority to legislate into. Missouri's 4.225 percent state rate continues to not reach SaaS, AI platform subscriptions, machine-payment-rail access, or software API usage.

The path back is a new statewide vote, not a bill. The legislature can refer another constitutional amendment to a future ballot; that is a normal thing for a legislature to do after a loss, and Missouri Promise PAC has already said it intends to keep working with the governor on tax reduction. But the mechanism requires a fresh referral and a fresh statewide vote. We are not going to put a date on that, and anyone who does is guessing. What we will say is that the interval between an 83-17 defeat and a serious second attempt is measured in years, not sessions.

Keep the SKU mapping flip-ready, but stop prioritizing it. The advice to ensure your billing system can move Missouri digital-services line items from zero to a non-zero rate in a single sprint was sound when a 2027 flip was live. It is now insurance against a contingency several years out. If Missouri readiness work is competing for engineering time against the January 1, 2027 cluster — California S.B. 122, Colorado H.B. 26-1223, Illinois S.B. 3019, and Utah S.B. 287 all take effect that day — the cluster wins, and it is not close.

The asymmetry worth noticing

2026 has been the most active year on record for state digital-tax expansion. Per MultiState's July 30 tracker, lawmakers introduced roughly 80 bills across 22 states this year to tax digital goods and services, and 12 were signed into law. Every one of those enactments came through a legislature.

Missouri is the one place in 2026 where the question went to voters directly, and it lost by 66 points.

We would caution against over-reading a single data point, particularly one contaminated by the income-tax phaseout riding alongside it. But the asymmetry is real enough to file: legislative base expansion has been passing routinely, and the one time an electorate was asked the question straight, it said no emphatically. States weighing whether to route a digital-tax expansion through a constitutional amendment rather than a statute now have a result to look at, and it is not an encouraging one for them.

For AgentTax's engine, nothing changes. Missouri was modeled as not taxing SaaS or non-enumerated services on the authority of art. X § 26, 12 CSR 10-109.050, and DOR Letter Ruling 7615. The review trigger we carried for the 2026 ballot measure has now resolved in the direction that requires no change.

What to watch next

Missouri drops from a medium-term watch to a low one. The items that replace it on the calendar are all implementation, not enactment: the CDTFA's discussion draft regulation under S.B. 122, expected around the end of August, which will decide sourcing and the custom-versus-prewritten line for California; the Utah State Tax Commission's implementing rules under S.B. 287; and the New Jersey Division of Consumer Affairs' guidance on A.5328's band edges. The legislative wave that produced the January 1, 2027 cluster is finished enacting. What remains is the considerably less glamorous work of finding out what the statutes actually mean.

If you operate an AI agent platform with Missouri customers, the practical instruction is short: do nothing, and reallocate the time to California.

For a current view of where every U.S. jurisdiction sits, see our AI Agent Sales Tax 50-State Guide. To run a tax calculation on an AI agent transaction in Missouri or any other state, sign up at agenttax.io.

This analysis is for informational purposes only and does not constitute legal or tax advice. Consult a licensed tax professional for compliance decisions.