One simple idea funds the whole plan: a 2% levy on Washington's fast-growing AI economy, not on working families. Here's the math behind eliminating state taxes for the middle class, funding our schools, halving STEM tuition, and still leaving billions for local priorities.
The math works, because of the incredible rate of growth in how businesses and industries use AI.
I know a lot of people feel uneasy about how quickly AI is changing things, and that's understandable. But the upside is enormous: we can use a small share of that growth to do big things for people.
A straightforward 2% levy on AI lets us deliver zero state tax burden for families under $125K, a permanent $2 billion boost to K–12 schools, and tuition cut in half for high-demand STEM programs, while leaving a strong surplus each year.
This is how we make AI's progress work for families and make sure no one is left behind, without raising taxes on working people.
The levy applies to three measurable dimensions of AI, the value and energy it consumes, collected from providers, never from households.
The electricity AI facilities consume: the easiest base to measure, billed straight through the utility.
The processing behind every model: tokens, GPU-hours and API usage, reported by providers.
Paid software & SaaS with meaningful AI built in: the broadest, fastest-growing base.
Designed to be pro-innovation: exemptions for R&D and green-powered compute, credits to avoid double-taxation, and lighter treatment for small innovators. It grows with AI adoption and never shrinks the household tax base.
generated annually by the 2% levy at steady state, as AI adoption scales across the economy.
to fund every promise in the platform, in full, each year.
left over every year for reserves, infrastructure, and local priorities.
The annual cost of delivering the platform, broken down.
Zero net state tax burden for households under $125K, through income-verified rebates and credits.
A permanent K–12 increase, roughly an 11–12% boost over current state appropriations, for smaller classes, teacher pay, and STEM pipelines.
Backfills that halve in-state tuition for high-demand STEM fields leading to high-paying careers.
The AI economy we're taxing can also be among the cleanest in the country, if we build it right. Washington is uniquely positioned to do exactly that, and the growth pays for itself in jobs.
Modern AI facilities recycle cooling water in a sealed loop, cutting ongoing water use by roughly 90%. The newest designs are filled once, then use about as much water per year as a single restaurant.
Washington generates the majority of its electricity from carbon-free hydropower. Compute that runs here runs cleaner than almost anywhere else on Earth, a natural home for responsible AI.
Meeting AI's energy demand shouldn't mean new dams. Supporting next-generation nuclear adds firm, carbon-free power while protecting salmon runs and river habitat that define the Pacific Northwest.
The tech economy already supports tens of thousands of Washington jobs, in construction, operations, and engineering, plus a wide multiplier of indirect work for trades, suppliers, and local services.
A closed-loop facility's yearly water use can be roughly the same as a single neighborhood restaurant's.
A lot of people are worried AI will replace their jobs, and not everyone wants to work in tech. They shouldn't have to. This plan is built to keep traditional work viable and valued.
Here's the key insight: moving the tax base onto AI changes the math for employers. Today, Washington taxes consumption and business activity heavily. Working families pay an estimated 9 to 14% of their income in state and local taxes, one of the most regressive systems in the country. Hiring a person and selling to a customer both get taxed; automating does not.
Flip that.
When AI compute and services carry the new levy, and families under $125K owe zero state tax, human labor becomes relatively cheaper, traditional businesses face lower costs, and local demand rises because families keep more of every paycheck.
The result: automation pays its share, while the cost of employing a human is never penalized by the state.
The same AI revenue that cuts STEM tuition in half builds the pipeline into the highest-paying, most meaningful fields of the coming decade, right here in Washington.
Building and applying the models reshaping every industry.
Advanced manufacturing and automation that keeps making here.
A new kind of machine, with breakthroughs aimed squarely at medicine.
Where AI and quantum converge to help people live healthier, longer lives.
In December 2024, Google's Willow chip proved a critical milestone: error rates can fall as a quantum machine scales up, the breakthrough the field had chased for decades. National efforts now target a first fault-tolerant quantum computer as early as 2028, with broadly useful machines expected in the late 2020s to early 2030s.
Together, AI and quantum can model molecules and proteins directly, compressing years of drug discovery into months, accelerating new treatments, diagnostics, and materials. Within our lifetimes, this is how we help Washingtonians live healthier and longer, and Washington's students should be the ones building it.
Recruiting a single major AI platform company to build here would anchor an entire ecosystem. It sets off a virtuous cycle: the more innovation we attract, the more we can invest in people, and the more we invest in people, the more innovation we attract.
A foundational AI builder makes Washington home.
Engineers, startups, and vendors gather around it.
More activity means more AI revenue to work with.
Revenue pays for tax relief, education, and tuition.
Better-educated talent attracts the next company, and the cycle repeats.
The model assumes strong-but-realistic year-over-year growth, highest in the early years, tapering as the market matures. Our assumptions sit at or below independent real-world forecasts.
| Category | Real-world forecasts | Our assumption |
|---|---|---|
| Data-center electricity | ~12–15% CAGR (global, AI-driven higher) | 15–25% YoY |
| AI compute (tokens / GPU-hours) | 17–43%+; capacity historically doubling fast | 40–60% → 25–35% |
| AI services / SaaS | ~21–38% CAGR (AIaaS & gen-AI software) | 30–45% → 20–30% |
Early-year growth is strongest, then tapers as adoption matures, the standard shape of a technology boom. If real-world growth runs higher (as it often has in AI's early phase), the surplus only grows.
The levy starts modest and grows with AI adoption. Each program phases in alongside it, tax relief deepens, school funding builds, and STEM backfills ramp, so every promise is fully funded each year and the budget never runs a deficit to get there. By Year 7 the plan reaches its full steady state.
Net state tax burden under $125K phased to zero as the base grows.
Builds to a permanent K–12 increase of roughly 11–12%.
Reaches half-off in-state tuition for high-demand STEM fields.
| Year | Revenue | Tax relief | Schools | STEM | Surplus |
|---|---|---|---|---|---|
| Year 1 | $2.0B | $0.8B | $0.6B | $0.1B | $0.5B |
| Year 2 | $3.2B | $1.6B | $0.9B | $0.2B | $0.5B |
| Year 3 | $4.7B | $2.6B | $1.2B | $0.3B | $0.6B |
| Year 4 | $6.2B | $3.6B | $1.5B | $0.4B | $0.7B |
| Year 5 | $7.7B | $4.4B | $1.7B | $0.4B | $1.2B |
| Year 6 | $9.0B | $5.0B | $1.9B | $0.4B | $1.7B |
| Year 7 | $10.2B | $5.5B | $2.0B | $0.4B | $2.3B |
An illustrative phase-in consistent with the growth assumptions above. Revenue stays ahead of spending in every year, so relief and investment reach full strength without ever running a deficit, and the surplus widens as AI adoption matures.
Both revenue and benefits phase in together. The levy starts modest to avoid chilling investment; relief scales up as the revenue base grows, so the budget never runs a deficit to get there.
This funds three specific priorities, not Washington's entire ~$70–80B biennial budget. The ~$7.9B is incremental relief and investment, designed to be matched by a new, growing revenue source.
Annual fiscal reviews adjust as real data comes in. If growth outpaces the model, the surplus grows or rates ease; if it lags, the base broadens or phase-in slows. The plan adapts.
Claims that “the math doesn't add up” overlook AI's trajectory and Washington's outsized place in it. Conservative assumptions already get there, and faster growth only widens the surplus.
If AI revenue grows the way the evidence suggests, Washington won't just be able to eliminate state taxes for families under $125K. Over time, it could move toward eliminating remaining human taxation entirely and paying a recurring dividend directly to residents, funded by the productivity of machines.
Andrew Yang was right, and ahead of his time.
In 2020, Yang built a national campaign around a $1,000-a-month "Freedom Dividend," arguing that automation would reshape work and that we should share the gains directly with people. The idea was treated as fringe. Since then, the explosive growth of AI, with the people building it, has made it look prescient.
PwC's landmark study finds AI could add $15.7 trillion to the global economy by 2030, more than the current output of China and India combined. That machine-made surge is exactly the prosperity Yang argued we should share.
See the projectionIt's arriving fast: the IMF estimates AI now exposes ~40% of jobs worldwide, up to 60% in advanced economies, and urges leaders to make sure the gains reach people, not just capital.
Read the IMF analysisOpenAI's Sam Altman argues that as AI drives the cost of goods toward zero, its soaring productivity should fund a recurring dividend to every citizen, Yang's idea, paid for by machines, not working people.
Read ‘Moore’s Law for Everything’This is a long-term aspiration, not part of the funded near-term plan above. It would depend on sustained AI growth and careful design, with the same annual fiscal discipline. But the direction is clear: as machines do more, the people of Washington should share in the prosperity.
Zero state taxes for families under $125K. Stronger schools. Affordable STEM degrees. A surplus for our future, funded by the fastest-growing part of our economy, not by working families.
Figures are illustrative order-of-magnitude projections based on public economic and AI-market data, designed to phase in over 2027–2033. Final amounts would be set by legislation and refined through official fiscal notes and annual review.
