The 2018 Supreme Court ruling in South Dakota v. Wayfair killed the physical-presence rule for sales tax. Eight years later, 46 US states have economic-nexus laws and AI/SaaS exposure has gone from "the state you incorporated in" to "46 states, 92 jurisdictions, on different filing calendars." This is the long version of how we got here — and where it's heading.
On 21 June 2018, the US Supreme Court ruled 5-4 in South Dakota v. Wayfair, Inc. that states could require remote sellers to collect sales tax based on economic activity, overturning a 1992 precedent (Quill Corp. v. North Dakota) that had required physical presence. In a single afternoon, the surface area of US sales-tax compliance for SaaS went from "the states you have offices in" to potentially every state in the country.
For AI and SaaS founders in 2026, Wayfair is the most important tax decision of the last two decades and probably the least understood. Every founder we talk to has heard of it. Most can't describe what it actually changed. This essay is what we wished someone had handed us in 2019.
The Wayfair case in two paragraphs
South Dakota passed Senate Bill 106 in 2016, requiring out-of-state sellers above $100,000 in annual sales or 200 separate transactions into South Dakota to register and collect sales tax. Wayfair (the online furniture retailer), Newegg, and Overstock sued, arguing the law violated the Commerce Clause of the US Constitution under the existing Quill precedent — which required physical presence.
The Supreme Court sided with South Dakota. Justice Kennedy, writing for the majority, called the physical-presence rule "an extraordinary imposition by the Judiciary" and described Quill as "artificial in its entirety." The Court held that economic and virtual contacts with a state could be sufficient to establish nexus. The case was 5-4. Roberts dissented. The world changed in the majority opinion.
What Wayfair actually changed
Three things, in descending order of operational impact:
- The basis for collection obligation. Pre-Wayfair: physical presence (an office, an employee, inventory, sometimes a contractor). Post-Wayfair: economic activity, defined by each state.
- The number of states you might owe in. Pre-Wayfair: typically 1–3 for a US SaaS company. Post-Wayfair: potentially 46.
- The cost of being remote. Pre-Wayfair: opening an office in a state had a tax cost. Post-Wayfair: not opening an office in a state still has a tax cost.
What Wayfair did not change:
- Federal tax treatment of remote sellers (unchanged).
- The taxability of specific products (each state still decides; SaaS taxability remains a state-by-state question).
- The duty to file in states where you have physical presence (still applies, on top of economic-nexus states).
The 46-state rollout — what happened between 2018 and 2026
Wayfair handed each state a green light to write its own economic-nexus law. The rollout was not orderly:
| Year | States with economic nexus | What happened |
|---|---|---|
| 2018 | 11 | Wayfair decided in June. SD, IN, RI, VT, MA, MS, MN, AL, MS, NE, KY first to enact or activate dormant laws. |
| 2019 | 31 | Most populous states (CA, NY, TX, FL, IL) live by year-end. Thresholds standardize around $100k OR 200 txns. |
| 2020 | 41 | Marketplace facilitator laws layered on top (states require Amazon/eBay/Etsy to collect on behalf of small sellers). |
| 2021 | 45 | Florida and Missouri — the last large holdouts — implement. SaaS taxability rules begin to diverge sharply state-to-state. |
| 2022–2023 | 46 | Kansas drops to $0 threshold (no minimum). Texas, Wisconsin, Colorado drop the 200-transaction threshold, keeping only $. |
| 2024–2026 | 46 | Threshold refinements. Several states explicitly classify AI/inference services as taxable. Federal MFA (Marketplace Fairness Act) talk resurfaces but no federal action. |
By 2024, the federal patchwork had stabilized at 46 states with sales tax + economic nexus. The remaining holdouts — Alaska (no state-level sales tax; local jurisdictions do have it), Delaware, Montana, New Hampshire, Oregon — have no statewide sales tax.
SaaS-specific consequences in the eight years since
Wayfair didn't mention SaaS or digital services explicitly. The case was about physical goods (furniture). Three downstream effects shaped SaaS exposure:
1. State-by-state SaaS taxability split
States started writing rules clarifying whether SaaS is "tangible personal property," "digital goods," or "services." The answers diverge:
- SaaS taxable: Texas (80%), New York, Washington, Pennsylvania, Massachusetts, Connecticut, Tennessee, ~20 others.
- SaaS exempt: California, Florida, Illinois, Ohio, ~15 others.
- SaaS partially taxable: Massachusetts, NJ, Arkansas (situational).
Most states have updated guidance since 2020. Some haven't. AI inference billing falls into the gray zone in most — usually defaulting to the state's SaaS treatment.
2. The economic-nexus threshold "200 transactions" killed consumer SaaS pricing
At $20/month, 200 transactions = 17 customers/year. The 200-transaction threshold effectively made any consumer-priced product subject to tax registration in every state with that threshold. About a third of states have since removed the transaction threshold (keeping only the $ threshold), but 15+ states still have it.
3. The Merchant-of-Record market emerged in earnest
Paddle, FastSpring, Lemon Squeezy, Cleverbridge, Macropay — all positioned themselves as "you don't want to handle 46 states." Wayfair is, in a real sense, the founding event of the modern MoR industry. Pre-2018, MoR was a niche option for cross-border sales. Post-2018, it became the default for US SaaS with consumer pricing.
Where AI sits in the post-Wayfair world, in 2026
AI inference businesses sit at the intersection of every trip-wire that emerged from Wayfair:
- Consumer pricing (high transaction count, low ticket).
- Often gray product taxability (most states default to SaaS, which is taxable in ~25 of them).
- Heavy international exposure (which adds EU OSS, UK VAT, India GST on top).
- Volume concentrated in CA/NY/TX/WA/MA/FL — the states that move fastest on enforcement.
In 2026, the typical AI-inference company at $5M ARR has nexus in 7–14 US states. The compliance cost of registering, collecting, filing, and remitting in each one is between $24k–$60k/year before you account for the engineering work of integrating per-state tax calculation at checkout. This is roughly the same cost as a Merchant of Record, except the MoR also carries the audit risk.
What comes next
Three trends to watch for the next few years:
- Federal pre-emption. The Streamlined Sales Tax (SST) project and the Marketplace Fairness Act have been periodic congressional themes since 2018. The consensus among tax policy folks is that federal pre-emption — a single national framework — would simplify dramatically but is politically dead until states themselves coalesce around one set of rules.
- Explicit AI inference taxability rules. Three states (NY, CA, TX) are drafting or have already drafted explicit AI/inference taxability rules in 2025–2026. Texas treats AI inference as "data processing services" (80% taxable). California is leaning toward exempt (treating it as a service, not tangible).
- The 200-transaction threshold continues to retreat. Texas, Wisconsin, Colorado, North Carolina, California — major economies have all dropped or are considering dropping the transaction count. The remaining holdout states with low transaction count will follow over the next 24 months.
Wayfair didn't make sales tax harder — it made it visible. The hard part was always there. We just used to pretend it didn't apply.— Lina Okafor, VP Finance, Nimbus AI
For the operational side — what specifically triggers nexus and how to model your exposure — see our companion piece on economic nexus thresholds. For the broader question of who carries the liability, our piece on Stripe Tax vs. Merchant of Record is the most-read article on this site for a reason.