Economics of GenAI: Invest $20, get $300 back.
Economics of GenAI: Invest $20, get $300 back.
Why GenAI is not expensive - at all.
Welcome to FullStack HR, and an extra welcome to the 103 people who have signed up since last week. If you haven’t yet subscribed, join the 7500+ smart, curious HR folks by subscribing here:
Happy Wednesday!
There’s a lot of new faces in here - welcome all!
Feel free to add me on LinkedIn if you haven’t already.
So what’s new?
Last Friday I did an internal education together with our owners Egmont.
Shout out to all my Egmont-colleuges that have found their way here!
Today, we’ll look at AI implementation through a purely economic lens.
This was sparked when I posted the PowerPoint video, and someone pointed out that Copilot is expensive (because it is) and asked what the ROI of the tool is.
That got me thinking that I never actually calculated the potential costs or savings we could achieve from using Generative AI tools in our organizations.
So, let's dive into the numbers and see what they tell us about the economics of parts of the AI implementation.
Even though many GenAI tools offer free trials, if you really want to get the most out of them, you’ll need to purchase a license—and at first glance, that license might seem pricey. But is it really? Let’s take a closer look.
I will make this as easy as possible by just looking at the cost of the tool and what the cost is for an average employee. I know that reality is way more complex than this, and you should see this only as a starting point, not a sheet for how reality is and that these are the actual gains every organization will yield.
It’s food for thought and perspectives on that these tools, are cheap - here we go.
The Investment
First, let's consider the cost.
Through out the article I’ll assume we do this calculation for a company with 1000 employees.
For our calculations, we'll use ChatGPT as an example. This assumes that you pay for the standard tool without discounts or custom solutions.
Cost per user: $20 USD per month
Number of employees: 1,000
Annual cost: $20 * 12 months * 1,000 employees = $240,000 USD
That’s the investment we need to recoup, somehow.
The Productivity gain
Now, let's assume a conservative 5% increase in productivity across the organization.
Studies such as Navigating the Jagged Technological Frontier suggest gains above 10% and The Effects of Generative AI on High Skilled Work: Evidence from Three Field Experiments with Software Developers say productivity will increase with 26,08%, but let’s be conservative here.
What does this mean in real terms?
According to the OECD, the average annual salary in the US is $77,000.
I’m also only calculating a straight salary cost here; no additional costs are added, no pension, no taxes on top of this, no office, no health care benefits which you should add if you want to calculate this more thoroughly.
For 1,000 employees: $77,000 * 1,000 = $77,000,000
A 5% productivity increase is equivalent to $77,000,000 * 0.05 = $3,850,000
The ROI calculation
Now, let's calculate the ROI:
Value of productivity gain: $3,850,000
Cost of AI implementation: $240,000
Net benefit: $3,850,000 - $240,000 = $3,610,000
ROI = (Net benefit / Cost of investment) * 100 =
($3,610,000 / $240,000) * 100 = 1504%
Breaking it down
So in short, the potential ROI for AI implementation is over 1500%. This means that companies could see a return of more than fifteen times that amount for every dollar invested in AI.
Let's put this in perspective:
The AI investment equals about 3.1 employee salaries ($240,000 / $77,000 ≈ 3.1). The productivity gain, however, is equivalent to adding 50 employees ($3,850,000 / $77,000 = 50).
Or, to put it another way, as long as the average annual salary is above $4,800, the company will see a positive ROI on this AI investment.
Beyond the numbers
Of course, these calculations are simplified. They don't account for:
Training costs
Potential job displacement
Time needed for employees to become proficient with AI tools
Variations in how different roles might benefit from AI
However, they also don't account for potential benefits like:
Improved decision-making
Enhanced innovation
Reduced errors
Increased employee satisfaction by eliminating tedious tasks
The bottom line
When we crunch the numbers, AI implementation doesn't look "expensive"—it looks like an incredible bargain.
And yes, I know I’m also not considering the variable impact AI might have across roles. Not all employees may experience a 5% increase in productivity. Some roles might see minimal benefits from AI tools like ChatGPT, especially if AI does not easily augment their tasks.
And I also know that quantifying productivity gains in financial terms is complex. Increased productivity doesn't always translate directly into increased revenue or reduced costs. Employees who earn less might benefit the most or vice versa.
Nonetheless, even if our productivity estimates are overly optimistic and the real gain is just 1%, the ROI would still be over 200%.
(And as mentioned above, all of this is counting overly conservative.)
This raises a question: Can companies afford not to invest in AI?
In a competitive landscape where margins often determine success or failure, a tool that potentially offers a 1,000%+ return isn't just a nice-to-have.
If a company chose not to implement AI tools, it would essentially be leaving money on the table. The cost of inaction—potentially millions in unrealized productivity gains—far outweighs the cost of implementation.
Moreover, as AI tools become more prevalent, companies that don't adopt them risk falling behind competitors who are leveraging these technologies to innovate and optimize at unprecedented speeds.
In conclusion, when we strip away the buzz and focus purely on the numbers, the case for AI implementation in businesses becomes compelling and almost irresistible.
The question is no longer whether companies can afford to implement AI, but whether they can afford not to.