Skip to content
Artificial Intelligence

How Much Should You Spend on AI Per Employee?

The average AI spend per employee hides a 600x gap. Here’s how HR and finance leaders should budget AI by role heading into 2027.

How Much Should You Spend on AI Per Employee?

Most companies are shelling out about $16 per employee each month on AI. Then there’s the top 1%, the AI equivalent of driving a Ferrari to the grocery store. They’re dropping about $10,200 per employee, per month.

Yes, you read that right; no, it’s not a typo. That’s a gap of more than 600x between your typical company and the ones Ramp’s AI Index calls the top 1% (and has affectionately labeled “AI-pilled”). One group is buying a few ChatGPT seats so marketing can jazz up their subject lines. The other is running autonomous coding agents 24/7 and paying what amounts to a junior employee’s salary, just to keep the bots caffeinated.

Somewhere in the middle is a number that’s about to become the star of every board deck: the average, about $172 per employee per month, or $2,068 a year, from an Atlanta Fed survey of senior executives. If you’re a CFO or CPO prepping your 2027 budget, this is the number you’ll get handed, and the one most likely to lead you astray.

Short answer: don’t budget to the company average. Budget by role, because the right number for a field tech and the right number for an engineer running agents can be $0 and $2,000 a month. Longer answer, with receipts, below. 

Why is the average lying to you?

There’s an old stats joke: Bill Gates walks into a bar, and suddenly everyone in the room is a millionaire on average. Nobody’s drink order changed, but the math sure did.

AI spend is the same story. When a handful of companies are dropping about $800 a month per head (top 10%), and an even smaller handful are past $10,000 (top 1%), the average gets yanked sky-high. Even the Atlanta Fed, the source of that $2,068 number, found that more than half of firms expect to spend $200 or less per employee this year. The median tells you what’s happening out there. The average tells you what happens when a few engineering teams go full mad scientist, and their numbers get blended into everyone else’s budget smoothie.

Why should HR care? Because benchmarks set budgets, and budgets shape behavior. Anchor on $172 and, if you’re a 400-person services firm with mostly light users, you’ll end up paying for licenses that gather more dust than logins.

Anchor on $16 and, if your product team is quietly running agents, finance will come knocking with some uncomfortable questions by Q2. Neither number is wrong; they’re just answering different questions. The trick is knowing which question you’re asking.

Why do AI agents cost so much more than AI seats?

Seats are cheap. Tokens are not.

For the last couple of years, buying AI felt a lot like buying any other SaaS tool. You picked a vendor, counted heads, and signed for seats.

GitHub Copilot Business runs $19 per user per month. Claude Enterprise starts around $20 a seat, with usage billed separately (more on that in a second). ChatGPT Business lands at $20 to $25, and Microsoft 365 Copilot sits at $30. Predictable, boring, easy to put in a spreadsheet. Procurement loved it.

Then agents showed up.

Once employees graduate from “summarize this email” to multi-step work, think automated coding, deep-dive data analysis, or research that runs while you’re off grabbing a sandwich, the bill stops being about seats and starts being about tokens. Gartner found nearly a quarter of tech leaders already spend $200 to $500 per developer per month on tokens for AI coding agents, and 6% are past $2,000.

One orchestrated agent interaction? About $1.20. A basic, linear one? Four cents. That’s EY’s math on a single customer-service interaction, then and now: a 30x difference for what, from the outside, looks like the same “using AI.”

Seat pricing

Token pricing

Scales with

Headcount

How people work

Typical cost

$19 to $30 per user per month

$200 to $2,000+ per developer per month

Predictability

High

Low

This is where every people leader should sit up and take notice. Seat pricing scales with headcount, HR’s bread and butter. Token pricing, though, scales with behavior: how people work, what they delegate, how ambitious they get with the tools. That makes it a work design challenge, and it’s got HR’s name written all over it. 

Why does AI spend vary so much inside the same company?

Here’s where it gets interesting. You don’t need to compare companies to see the gap. You can see it inside a single org chart.

Your field technicians and non-desk staff? Often $0 a month in AI spend. Your software engineers running heavy coding agents? Commonly $200 to $500 a month, and north of $2,000 at some companies. Same benefits package, same handbook, same all-hands. Completely different cost profile.

This isn’t a hunch. Ramp’s latest data shows engineering teams drove 72% of classified AI spend in October.

This is new territory. Sure, some roles have always cost more to equip, a designer’s laptop versus a call center headset, but those differences were measured in hundreds of dollars a year. Now, for some roles, tooling costs are starting to look like a real chunk of total compensation.

AI costs are quietly becoming part of what it takes to do a job, but hardly anyone is budgeting for them like they matter. And access? It’s getting uneven.

Imagine one team gets a blank check for AI tools while another must scrape by with the basics. Suddenly, you’ve got a two-class system on your hands, and no one officially decided it. It shapes who gets more done, who builds skills, and who gets ahead.

Does your industry change the math?

Yes, and by a lot. The Atlanta Fed puts professional and business services at about $3,470 per employee for 2026 (roughly $289 a month). Manufacturing comes in at nearly $900 a year, or about $75 a month. The cross-industry average sits between them at $2,068.

None of this is shocking when you think about the work. Services firms sell knowledge, and knowledge work is exactly what today’s AI tools are built to chew on. Manufacturing has a big share of people on the floor, not in front of a screen, which pulls the per-head number down, unless we explore robots (future post).

The challenge: a manufacturing CHRO or CPO who sees $75 might think AI isn’t their problem yet. Meanwhile, their engineering, supply chain, and quality teams could be spending as much as they do for a large new piece of machinery. Industry averages are a fine place to start. They’re a terrible place to stop.

Okay, so how should people leaders budget AI spend per employee?

Here’s my take: start budgeting AI the way you budget headcount, by role or task, not by company.

Segment your workforce by AI intensity. You already group roles by job family and level for comp, so do the same for AI usage. Suddenly, your company-wide “$172 average” turns into four numbers you can defend in a budget meeting. 

AI intensity tier

Who’s in it

Typical monthly AI cost per person

Non-desk

Field technicians, frontline staff

$0 to minimal

Light knowledge workers

Seat-only users

$20 to $40

Heavy knowledge workers

Seat plus moderate token use

A few hundred dollars

Agent-heavy builders

Engineers running coding agents

$200 to $2,000+

Bake AI cost into your fully loaded cost of a role. Finance already adds benefits, payroll taxes, equipment, and real estate. If an engineer’s tooling costs $1,000 a month, that’s $12,000 a year that belongs in the workforce plan, not buried in an IT line item. It also makes the “hire vs. augment” conversation a lot more honest.

Set guardrails, not ceilings. Token caps sound responsible, right up until your best engineer hits one mid-sprint. It’s better to set expected ranges by role, watch for outliers, and ask, “what did we get for it?” before you ask “why so much?” A $10,000 monthly bill behind work that actually ships is a steal. A $10,000 bill for agents generating code nobody ships? That’s a problem.

Keep an eye on the access gap. If AI fluency is the new career rocket fuel, and it’s heading that way, uneven access quickly turns into an equity issue. Make sure folks on light-tier budgets still have room to learn, experiment, and grow into heavier use.

Pair the spend with work redesign. Spending more on AI without changing how work gets done is just buying a faster horse or putting an electric motor on a bicycle. This is also where we spend most of our time: working with clients on their AI work redesign. Companies in Ramp’s index aren’t throwing money around for fun; they need to rebuild workflows around the tools. That’s the real lesson hiding in the data. 

Is AI spend a people strategy question?

For years, HR’s job with tech was to pick the tools and drive adoption. Now, the tools have a meter running, and how fast it spins depends on how people work. Like it or not, AI spend is now a people strategy question.

So next time someone drops “$172 per employee” into a planning meeting, ask the only question that matters: which employee?

Frequently Asked Questions

What is the average AI spend per employee?

An Atlanta Fed survey puts the 2026 average at $2,068 per employee per year, about $172 a month. The typical company spends far less: Ramp’s data puts the median at about $16 per employee per month. 

Why are AI agents more expensive than AI seats?

Seats are a flat per-user fee, usually $19 to $30 a month. Agents bill by tokens, so cost scales with how much work people hand off, and Gartner found many companies already spend $200 to $500 per developer per month.

How should companies budget for AI per employee?

By role. Segment the workforce into AI intensity tiers and build each tier’s expected cost into the fully loaded cost of the role.

Does AI spend per employee vary by industry?

Yes. Professional and business services run about $289 per employee per month, compared with about $75 in manufacturing.

Why should HR care about AI spend?

Token costs scale with how people work, not just headcount. That makes AI spend a work design and workforce planning question.

Anthony Onesto

We humanize work for everyone because we know it creates better outcomes for humanity and business.

Incoming Mail

Subscribe to our newsletter

Latest Articles

How Much Should You Spend on AI Per Employee?

How Much Should You Spend on AI Per Employee?

The average AI spend per employee hides a 600x gap. Here’s how HR and finance leaders should budget AI by role heading into 2027.

Why AI Won't Replace Human Judgment.

Why AI Won't Replace Human Judgment.

Will AI replace human judgment? No. Learn the difference between consistency work and adaptation work, and how HR leaders should split task...

The AI Chatbot That Unlocked a Billion Euros: How IKEA’s People Strategy Beat Automation at Its Own Game.

The AI Chatbot That Unlocked a Billion Euros: How IKEA’s People Strategy Beat Automation at Its Own Game.

IKEA’s chatbot saved €13M. Retraining 8,500 workers helped build a €1.25B sales channel. Here’s what CPOs can learn about AI reskilling.