AI either “substitutes” for human labor or “augments” it. This very distinction is the starting point for understanding how AI moves corporate profit.
Before discussing how AI contributes to profitability, let us first break down a company’s profit structure. In its most common form, it can be expressed by the equation: Profit = Revenue − Cost. In other words, there are broadly two routes that affect corporate profit — how much you can lower costs, and how much you can grow revenue. These two routes correspond to the patterns of AI “substitution” and “augmentation” discussed below.
AI’s Impact on Human Labor: Two Basic Patterns
AI can replace human tasks. Alternatively, people raise their productivity by leveraging AI. In other words, there are two basic patterns: the work humans have done so far is either substituted by AI, or made even more efficient through the use of AI.
Using the profit equation, we can see what each of these means. When AI substitutes for human work, over the long run the unit cost of AI is expected to be lower than the cost of human labor, so its effect works on the cost side.
On the other hand, when AI boosts human capability, there are two directions of change — the time required to produce one unit of output is shortened, or the output per hour increases — so its effect can be said to work not only on the cost side, but also on the revenue side.
Let us make this concrete. Typical examples of substitution include chatbots that handle first-line customer inquiries, AI that drafts routine documents, and data-processing agents that carry out data entry and reconciliation. Typical examples of augmentation, by contrast, include an engineer using a coding agent to raise development speed, a salesperson leveraging a business copilot to finish proposal materials in a fraction of the time and win more deals, and a designer bouncing ideas off a design agent to broaden the range of concepts.
In this way, the patterns by which AI adoption changes corporate profit come into view. When the substitution pattern dominates, the cost-reduction difference flows straight through to profit. When the augmentation pattern dominates, profit rises through both cost reduction and revenue growth.
Understanding the Real-World Constraints
However, AI’s cost effect or productivity effect does not necessarily translate into profit directly and unconditionally. Let us first explain the constraint on the cost effect: namely, that in most cases the cost savings obtained through AI cannot be captured continuously (in smooth, incremental steps).
For example, suppose that by using AI one company can reduce the cost of five workers’ daily jobs by an average of 10%. Ten percent of five people amounts to 0.5 of a person, but headcount can only be reduced in whole-number units, so the company cannot actually cut 0.5 person’s worth of labor cost as such.
There are constraints on the productivity side as well. If market demand is already saturated, then even if productivity rises, it does not simply lead to expanded output. As a result, companies are either left carrying higher costs from output they cannot sell, or forced to cut prices — so the surplus from AI-driven productivity gains ends up being transferred to the buyer’s side.
Conversely, the story changes in a growth market. Where demand is expanding, the additional output is readily converted into revenue, and the productivity surplus stays in hand as profit rather than leaking away through price cuts. As a rough guideline, then: cost reduction through substitution tends to work in mature, saturated markets, while output growth through augmentation tends to work in growth markets.
The Other Face of Substitution: “Reallocating” the Time AI Frees Up
These constraints come with a mitigation that is easy to overlook. Substitution does not, in the first place, mean only “cutting headcount.” By using AI to free up the time spent on routine work, companies can redirect that surplus time toward scarcer, higher-value tasks. The non-continuity noted earlier — “you cannot cut 0.5 of a person” — can also be partly overcome if companies replace whole-number adjustments such as headcount reduction with continuous fine-tuning of how time is allocated across tasks.
For example, a company might use AI to cut the time spent on back-office work that does not face the customer (Non-Client-Facing), and shift the freed-up time to revenue-generating work such as customer engagement and proposals (Client-Facing). At the level of individual tasks this is “substitution,” but at the level of the person it is “augmentation” — a path that achieves augmentation by way of substitution. Note, however, that whether this reallocation actually creates value depends on how much room for growth the destination work has (the marginal productivity at the destination).
The Problem Over the Long Run
Over the long run there is a further problem. If the cost-reduction or productivity-improvement effects gained from using AI are generic (broadly available), then over time every company in the same industry can obtain the same effects. As a result, no company holds a distinctive competitive advantage, and through the mechanism of market competition each company’s profitability reverts to its original level.
Putting AI to Effective Use Requires Strategy
While it is broadly clear that using AI produces effects on both the cost side and the productivity side, this does not necessarily translate into improved corporate profitability on its own. To use AI effectively, a company’s management must think strategically.
| Pattern | Mainly affects | Where it tends to work | Main constraint |
|---|---|---|---|
| Substitution | Cost | Mature / saturated markets | Non-continuity (can only cut in whole-number units) |
| Augmentation | Cost + Revenue | Growth markets | If demand is saturated, the surplus leaks to buyers |
And to draw out both effects to the fullest, it is not enough merely to introduce individual tools; reorganizing work and processes — that is, transforming the organizational structure itself — becomes unavoidable. Let us explore that “AI transformation strategy” in depth in a future article.