How Much to Spend on AI in 2026: What the Federal Reserve's Data Means for a 10-Person Firm

How Much to Spend on AI in 2026: What the Federal Reserve's Data Means for a 10-Person Firm

If you're trying to figure out how much to spend on AI for your professional services firm in 2026, the Atlanta Fed just published the most comprehensive data yet. The headline number — $2,068 per employee in average planned AI investment — has already made the rounds in business press.

That number will get cited in slide decks, LinkedIn posts, and board meetings for the rest of the year. Most people using it will draw the wrong conclusion.

Here's what the data actually says, and what it means if you own a 10-person firm.

---

The Number Everyone Cites (And Why It's Misleading)

$2,068 per employee. Across all US private firms. Up from $1,358 in 2025. A 52% increase in one year.

That is a real number, drawn from a rigorous Atlanta Fed survey. The total projected investment across private US firms: $280 billion in 2026.

The problem is that averages obscure the shape of a distribution. And this distribution is not a bell curve.

It's bimodal. Two peaks, a valley in between, almost nothing in the middle.

The top 10% of firms are planning to invest $2,800 or more per employee. Meanwhile, more than half of all firms are spending $200 or less.

That is not a normal distribution with a few outliers. That is two separate populations doing fundamentally different things with AI — and the average lands somewhere neither group actually lives.

When a consultant tells your industry group that "firms are spending $2,068 per employee on AI," they are technically correct. They are also describing a firm that does not exist.

---

The Bimodal Reality: $200 or $2,800 — The Middle Doesn't Exist

Here is what the distribution actually tells you:

The bottom half (≤$200/employee): These firms have subscribed to a few tools — maybe ChatGPT Plus for one person, maybe Copilot is bundled into their Microsoft 365 they barely use. They've experimented. They haven't integrated. They're spending roughly what it costs to keep the lights on in the conversation about AI.

The top 10% (≥$2,800/employee): These firms have made deliberate, workflow-level AI investments. They've paid for purpose-built tools in their vertical. They've implemented, trained, and measured. Their AI spending is a budget line, not an afterthought.

The middle: Largely absent from the data.

This is not a gradient where every firm is slightly more or slightly less AI-invested than the next. It is a separation. The firms in the top tier are not just more tech-forward — they are operating with structurally different cost structures, capacity ceilings, and service delivery models.

For a 10-person professional services firm, the relevant question is not "what is the average?" It's: which half of this distribution are we in?

---

What $2,068 Per Employee Actually Buys at a 10-Person Firm

If you run a 10-person firm and want to understand what $2,068 per employee means in practice: that's $20,680 per year.

Here is what that budget can realistically cover:

Tool category	Example tools	Annual cost (10-person firm)
AI writing/drafting assistant	Claude for Teams, ChatGPT Team	$3,600–$4,200
Practice management AI features	Clio Duo, Karbon AI, Salesforce Einstein	$2,400–$6,000
Meeting notes and transcription	Fathom, Otter.ai Business	$1,200–$2,400
Proposal/document automation	Prospero, PandaDoc AI	$1,800–$3,600
Accounting/tax AI layer	Intuit Accountant Suite, Karbon tax workflows	$2,400–$5,000
Total		$11,400–$21,200

That range lands right on the $20,680 average for a 10-person firm. If you're spending significantly less than that, you're in the bottom half of the distribution — regardless of how many tools your team has "tried."

The distinction matters: subscribing is not spending. Active, integrated AI use — where the tool is embedded in a repeated workflow and used by more than one person — is what the high-investment firms are doing. Signing up for a tool and using it occasionally is not.

---

The Professional Services Impact: 1.8% Fewer Non-Degreed Hires

The Atlanta Fed didn't just measure AI spending. They measured what firm owners plan to do with it.

In the same survey data, firms were asked how their AI investment plans affect their anticipated headcount. Professional services firms — accounting, law, consulting, staffing, marketing agencies — forecast a 1.8% decrease in non-degreed worker hiring demand as a result of AI adoption. That is the largest reduction of any sector in the survey.

The breakdown by credential level:

• Non-degreed workers: 1.1% anticipated hiring reduction
• College-educated workers: 0.8% anticipated hiring reduction

Both groups face reduced demand. Non-degreed workers face more.

To be precise about what this means: these are forecasts, not current layoffs. Firms are saying that AI investment is causing them to plan for fewer new hires — not that they are cutting existing staff. The effect is expected to show up in hiring decisions over the next 12–18 months, not in immediate headcount cuts.

But the direction is clear. The workflows that generate the most demand for entry-level and non-degreed staff — document processing, data entry, scheduling, basic research, intake, transcription — are the workflows most actively being automated by the firms in the high-investment tier.

---

What This Means for Your Headcount and Hiring

If you are a firm owner planning your next hire, this data has a direct implication: the case for headcount growth in support and administrative roles is weakening faster than the case for senior professional growth.

This does not mean you should not hire. It means the decision calculus has shifted.

Before adding headcount in any support function, the right question is now: can AI absorb this capacity need, at a fraction of the cost, without the management overhead?

For a 10-person accounting firm considering a part-time bookkeeping assistant at $35,000/year: can an AI-integrated workflow handle the volume that hire was meant to address? Often the answer is yes — and the cost difference finances two years of the entire firm's AI stack.

For a 15-person consulting firm considering a project coordinator: before posting the role, map the coordinator's expected weekly tasks. Sort by "AI can do this" versus "requires human judgment and relationships." If more than 50% of the role is in the first column, you may be about to hire into a function that is compressing.

This is not a call to stop hiring. It is a call to pressure-test headcount decisions against the AI stack you already have — or could have — before defaulting to the solution that worked in 2019.

---

The Right Benchmark for a 5–50 Person Firm

The $2,068 average is not your benchmark. Here is a more useful framework:

Under $200/employee: You are in the bottom half of the distribution. You are experimenting, not integrating. This is a fine place to be for the first 60 days. It is not a stable position for 2026.

$500–$1,000/employee: You are in the competitive range for a firm in the early integration phase. At 10 employees, this is $5,000–$10,000 annually. This covers 2–3 purpose-built tools used regularly by multiple staff members. You are building the habit of AI use before optimizing for specific ROI.

$1,000–$2,000/employee: You are in the integrated range. Tools are embedded in workflows. At least two or three recurring tasks have measurable before/after time comparisons. This is where the productivity compounding starts.

$2,800+/employee: You are in the top-tier cohort. At 10 employees, this is $28,000+ annually. You are running AI at the process level — not just individual tools, but connected workflows where AI output feeds into the next step. This is achievable for firms that have completed 12–18 months of methodical integration.

The target for most small professional services firms in 2026 is to move from under $200 into the $500–$1,500 range. That is not a large capital investment — it is a commitment to active integration over passive subscription.

---

How to Decide How Much to Spend on AI at Your Professional Services Firm in 2026

The right AI budget for your firm is not the average. It is the answer to three questions:

1. What is my highest-friction, highest-volume repeatable task?

Not what feels important. What actually consumes the most staff time every week. For most professional services firms, this is one of: client communication drafting, document review, meeting notes and follow-up, proposal generation, or data entry and intake.

2. What does that task cost you per month in staff time?

Estimate the hours, multiply by the fully-loaded cost of the person doing it. That is your monthly cost floor for the problem. Your AI budget for that task can be anything up to that number and still produce positive ROI.

3. What tool specifically addresses that workflow?

Not "AI in general." A specific tool, with a specific use case, that you can implement in the next 30 days and measure in 90 days. This is where you start.

For a 10-person firm that answers those three questions honestly, the AI budget almost always falls between $500 and $1,500 per employee for the first meaningful integration phase. That lands below the $2,068 average — but above the $200 floor where the bottom half of the distribution is sitting.

The goal is not to match the average. The goal is to leave the bottom half of the distribution and build toward the top.

One thing to do this week: Calculate your firm's current AI spend per employee. Add up every AI-related subscription — including AI features bundled into tools you already pay for. Divide by headcount. If you're under $200, that number tells you exactly where you stand in the distribution the Atlanta Fed just mapped.

---

Sources: Atlanta Fed Policy Hub Macroblog — How Much Are Firms Spending on AI (and What Will Happen to Headcounts?), May 6, 2026 | Federal Reserve Board FEDS Note — Monitoring AI Adoption in the US Economy, April 3, 2026