AI Isn't Killing Tax Prep. It's Creating More Customers.
The numbers have been sitting with me for a while. About 600 accounting graduates per year across the entire Denver metro area. Roughly 2,000 preparers retiring annually. That gap on its own is a problem, and then AI enters the picture and the whole thing gets weirder than most people expect.
The simple 1040 is shrinking, not growing
Everyone assumes AI eats tax prep. TurboTax already handles simple W-2 returns. The logic seems clean: AI gets better, more returns get automated, fewer preparers needed. I get why people believe this. I believed it too.
The mechanism actually runs in reverse.
AI is eliminating entry-level W-2 jobs right now. Office admin. Junior analyst. Call center. Data entry. The people losing those jobs don't evaporate. They become gig workers. Virtual assistant on a 1099. Rideshare driver with a 1099-K. Fractional bookkeeper filing a Schedule C. Each displaced worker becomes a tax filer with two or three income sources, self-employment tax, quarterly estimates, and deductions across categories that don't fit in a dropdown menu.
The simple 1040 that software handles is shrinking as a share of the workforce. The complex multi-source return is becoming the default. AI isn't killing tax prep. It's pumping volume into the profession while simultaneously destroying the entry-level jobs that used to be the alternative.
Two curves, both bending toward the preparer
Supply side: ~600 accounting grads per year across a metro of 1.67 million workers. The average preparer is over 55. Fewer than 6,000 new grads per decade. These numbers don't close. The profession's own junior loop is broken in the same way AI is breaking everyone else's: fewer people entering the pipeline while the people in it age out.
Demand side: BCG estimates 50 to 55 percent of US jobs will be reshaped by AI in the next two to three years, with 10 to 15 percent potentially eliminated in four to five. In Denver that's an estimated 80,000 to 150,000 workers pushed from W-2 employment into gig work, contracting, or multiple-job patchworks over five years. Every single one becomes a more complex tax filer than they were before.
Same force, two effects. AI kills the simple W-2 and creates the complex multi-source filer. The preparer who only knows how to plug numbers into software loses. The preparer who can sit across from a gig worker with four income streams and a home office question is irreplaceable.
And then there's the math on what this market actually looks like. The Denver workforce has an estimated 300,000 to 500,000 workers with something more complex than a simple W-2. Self-employed professionals, construction contractors, gig platform workers, healthcare travelers, immigrant small business owners. Even capturing half a percent of that group gets you 1,500 to 2,500 clients. At average fees of $400 to $800 per complex return, that's real money before you've hired a single employee.
The credential ladder is simpler than it looks
From outside, the IRS credential path reads like alphabet soup. PTIN. AFSP. EA. RTRP. It's actually three straightforward steps, and the first one costs $18.75.
PTIN. Preparer Tax Identification Number. Anyone who prepares federal returns for compensation needs one. Apply through the IRS Tax Pro account. This is the permission slip, not a credential. It says you're allowed to do the work, not that you know how.
AFSP Record of Completion. The Annual Filing Season Program. 18 hours of IRS-approved continuing education per year, including a six-hour federal tax law refresher with a test. Costs $100 to $250 through approved providers. You get limited IRS representation rights and a listing in the public directory. This is annual recognition, not a permanent certification. You earn it again each year. But it's the practical working credential for non-EAs and non-CPAs.
Enrolled Agent. The real thing. Unlimited IRS practice rights. Same standing as attorneys and CPAs before the IRS. Three exams: Individuals, Businesses, and Representation/Practices/Procedures. Pass all three within three years. File Form 23. Pass a suitability check. As of March 2026 the exam moved from Prometric to PSI Services. Maintenance is 72 CE hours every three years, 16 per year minimum, two ethics per year.
Here's what the timeline actually looks like in practice. The PTIN takes an afternoon. The AFSP is a few weeks of coursework. The EA is the multi-year investment. But the supply gap means you don't need the EA to start earning. H&R Block hires seasonal preparers with just a PTIN. January through April. One season of W-2s and basic credits and you're credible as a second preparer. Year two: complex returns under supervision. Year three: independent with an EA application in flight.
What the career changers already figured out
Reddit career-change forums converge on a few things that apply directly here, and I've spent enough time reading through them to notice the patterns.
Credential moat beats generalist degree every time. CPA beats "business major." Series 7 beats "finance background." EA beats "I took some tax classes." The credential proves specific competence to a specific market. The degree proves you sat through requirements.
Don't overthink it. This comes up constantly. The people who regret not switching vastly outnumber the people who regret trying. The ones who do switch in their 30s and 40s mostly say the same thing: should have started sooner. And the post-2022 consensus has shifted hard away from "learn to code, get six figures" and toward licensed professions and trades that AI can't touch at the liability layer.
The thing that keeps coming back: find the intersection of what you're decent at, what the market needs, and what pays. Tax prep has a demographic tailwind most professions would kill for. Aging workforce. Shrinking pipeline. Rising complexity. And a barrier to entry that's mostly psychological: the fear that AI will eat the profession is exactly what keeps competitors out while the existing preparers keep aging and retiring.
Is it for you?
Tax prep isn't for everyone. The seasonality is real. January through April is a grind and the rest of the year is slower unless you're building a practice. The learning curve on business returns, multi-state filings, and representation work is steep. You have to actually like helping people who are confused and sometimes scared about money and the IRS. Not everyone does.
But if you can pass a background check, study for a test, and sit across from someone with three income streams and a crypto side hustle and figure out what they actually owe, the numbers say there aren't enough of you. And AI is only making that gap wider.
If you're considering the EA path and want to see what the study roadmap looks like with real numbers and timelines, check out ea-coach.shawnli.dev.