There are now 653,408 actively licensed CPAs in the United States — and the pipeline feeding that number is thinner than it has been in years. For a CPA or accounting firm owner, that is not an abstract industry trend. It is a staffing problem, a capacity problem, and increasingly a survival problem.
Artificial intelligence has become the most-discussed answer. But most of the noise is either hype (“AI will replace accountants”) or fear (“AI will leak my clients’ tax data”). The truth sits in the research, and the research is surprisingly clear: firms that adopt AI deliberately are pulling ahead, firms that adopt it carelessly are getting breached, and firms that ignore it are quietly losing capacity they cannot replace.
This guide walks through what primary research actually says about AI and the survivability of a modern CPA firm — the talent math, the productivity upside, what your clients now expect, and the security and compliance guardrails you cannot skip.
Key Takeaways
- The number of actively licensed CPAs has fallen to 653,408, and new CPA Exam candidates dropped from 42,626 in 2023 to 28,082 in 2024 — capacity, not demand, is the constraint.
- Enterprise generative-AI adoption at tax firms nearly tripled from 8% to 21% in a single year, and 77% of clients now want the firms they hire to use it.
- Firms with a visible, defined AI strategy are twice as likely to see revenue growth — yet only 22% actually have one.
- AI cuts both ways: AI-enabled breaches now cost about $6 million on average, and unsanctioned “shadow AI” adds roughly $670,000 to the cost of a breach.
The Math Behind the CPA Shortage
Before you can decide what AI means for your firm, you have to see the staffing hole it is filling. The demand for accounting work is not shrinking — the supply of people to do it is.
- As of August 2025, there were just 653,408 actively licensed CPAs in the United States (NASBA, 2025).
- U.S. schools awarded 55,152 accounting bachelor’s and master’s degrees in the 2023–2024 academic year, down 6.6% from the prior year (Journal of Accountancy, 2025).
- Master’s degrees in accounting or taxation dropped roughly 15% year over year, versus a 3.3% drop in bachelor’s degrees (Journal of Accountancy, 2025).
- New candidates entering the CPA Exam pipeline fell from 42,626 in 2023 to 28,082 in 2024 (Journal of Accountancy, 2025).
- There is a glimmer of recovery: spring 2025 enrollment in two- and four-year accounting programs hit 266,506 students, a 12.4% increase and the highest total since 2020 (Journal of Accountancy, 2025).
For a 20-person firm, that math is personal. When a senior preparer retires and the two candidates you would have hired to backfill never entered the profession, the work does not disappear — it lands on the people who stayed. AI is attractive precisely because it addresses the one thing you cannot buy your way out of quickly: hours of skilled capacity.
Why AI Is Now a Survival Tool, Not a Gadget
The adoption curve has bent sharply. What looked like an experiment in 2024 is becoming standard practice, and the productivity gains are being measured in real dollars.
- Enterprise generative-AI adoption at tax firms nearly tripled year over year, from 8% in 2024 to 21% in 2025 (Thomson Reuters, 2025).
- 71% of tax professionals now believe generative AI should be applied to daily work, up from 52% a year earlier (Thomson Reuters, 2025).
- Professionals using AI are projected to save about 5 hours per week within the next year — an average of $19,000 in annual value per person (Thomson Reuters, 2025).
- Across the U.S., that efficiency could translate to an estimated $12 billion in annual impact for the CPA sector alone (Thomson Reuters, 2025).
- More than half — 53% — of professionals say their organizations are already seeing ROI from AI adoption, directly or indirectly (Thomson Reuters, 2025).
Five hours a week per professional is roughly a month of reclaimed capacity per person, per year. For a firm that cannot hire its way out of the shortage, that is not a productivity nicety — it is the difference between taking on the next ten clients and turning them away.
Your Clients Already Expect It
Adoption is not only being pushed by owners chasing efficiency. It is being pulled by clients who assume their firm is already using modern tools.
- 77% of clients want the tax firms working for them to use generative AI — yet 59% of those clients do not even know whether their firm is using it (Thomson Reuters, 2025).
- 80% of professionals believe AI will have a high or transformational impact on their profession over the next five years (Thomson Reuters, 2025).
- 88% of professionals say they favor profession-specific AI assistants over general-purpose tools (Thomson Reuters, 2025).
The gap between “clients want it” and “clients cannot tell if we do it” is a marketing opportunity hiding in plain sight. A firm that can say, clearly and safely, how it uses AI to turn work around faster has a differentiator its silent competitors do not.
The Strategy Gap Is the Real Opportunity
Here is the finding that should reframe the whole conversation: the advantage does not go to firms that simply use AI. It goes to firms that use it on purpose.
- Only 22% of organizations say they have a visible, defined AI strategy (Thomson Reuters, 2025).
- Firms with a visible AI strategy are twice as likely as those with ad-hoc adoption to experience revenue growth (Thomson Reuters, 2025).
- Firms with a defined strategy are 3.5 times more likely to experience critical AI benefits than those with no real adoption plan (Thomson Reuters, 2025).
Because only about one firm in five has a real plan, the bar to becoming a leader is low. A short, written AI policy — which tools are approved, what data may and may not be entered, who reviews AI-assisted work — puts a 20-person firm ahead of most of its peers without buying a single new piece of software.
The Risk: AI Cuts Both Ways
The same technology that reclaims your capacity is also arming the people trying to get into your systems — and quietly creating new ways for client data to leak. This is where survivability is won or lost.
- The global average cost of a data breach is now $4.99 million (IBM, 2026).
- One in four malicious breaches is now AI-enabled — a 56% increase over the prior year (IBM, 2026).
- Those AI-enabled breaches cost an average of $6 million, roughly $1 million more than the global average (IBM, 2026).
- One in five organizations reported a breach caused by “shadow AI” — unsanctioned AI tools — which added about $670,000 to the average breach cost (IBM, 2025).
- 13% of organizations reported breaches of their own AI models or applications, and 97% of those lacked proper AI access controls (IBM, 2025).
Picture the most common version of this risk in a busy firm: a preparer, buried in March deadlines, pastes a client’s full return into a free public chatbot to “just summarize the K-1s.” No malware, no hacker — sensitive taxpayer data has simply left your control and joined the one-in-five shadow-AI breach statistic. The fix is not to ban AI; it is to give staff an approved, private tool so they never feel the need to reach for the public one.
Encouragingly, the defensive side of AI pays off too: organizations that used AI and automation extensively in their security operations cut breach costs by almost $2 million and shortened the breach lifecycle by an average of 80 days (IBM, 2025).
Compliance: The IRS and FTC Already Have Rules
For accounting firms, AI governance is not optional good hygiene — it overlaps directly with rules the IRS and FTC already enforce. A firm that adopts AI without a written security plan is taking on regulatory risk, not just cyber risk.
- A Written Information Security Plan (WISP) is required by law: the Gramm-Leach-Bliley Act requires all financial institutions — including tax and accounting firms — to protect customer data (IRS, 2025).
- Under the FTC Safeguards Rule, a firm must report a breach to the FTC when 500 or more individuals are affected, within 30 days (IRS, 2025).
- Governance is lagging adoption badly: 63% of breached organizations either had no AI governance policy or were still developing one (IBM, 2025).
The practical move is to fold AI use directly into the WISP you are already required to maintain. The IRS offers a free template (Publication 5708) built for smaller practices, and adding a short section on approved AI tools and prohibited data uses turns a compliance obligation into a genuine guardrail for your team.
How a 20-Person CPA Firm Should Actually Start
You do not need a data-science team or a six-figure budget to get the upside without the downside. The firms pulling ahead are the ones being deliberate, and deliberate is well within reach for a small practice.
A sensible order of operations looks like this: pick one high-volume, low-judgment workflow (first-pass document summarization, drafting client emails, organizing source documents) and standardize on a single approved tool for it. Write a one-page AI policy that names the approved tool and lists what client data may never be entered into any public system. Route AI through business accounts with proper access controls rather than staff signing up for free consumer apps. Keep a human reviewer on anything that touches a return or client advice. Then fold all of it into your existing WISP.
That sequence captures the roughly five hours per week per person the research points to, while keeping you out of the shadow-AI and governance-gap statistics that turn a productivity tool into a breach headline.
Final Thoughts
The survivability question for CPA firms is not really “will AI replace accountants?” The data points somewhere more useful: the profession is short on people, the work is not, and AI is the fastest way to close that gap — as long as it is adopted with a strategy and a security plan rather than by accident. Firms that plan are already twice as likely to grow. Firms that improvise are the ones showing up in the breach reports.
Tech Advisors works with accounting and financial firms to do exactly this: choose safe, private AI tools, build the written policies and WISP that keep you compliant with IRS and FTC requirements, and lock down the access controls that keep client data where it belongs. If you want a second set of eyes on how your firm can use AI without becoming a statistic, we are glad to have that conversation — no obligation, just a plan you can actually use.
