Conclusion First: AI Is No Longer Just a ‘Tool Cost’

We are no longer at the stage of discussing whether to implement AI.

The question is this: If an employee earns 300,000 yen a month and an API costs 30,000 yen a month, and both can do the same job, which one would you keep?

This is not a thought experiment; it is already happening in reality. AI-related expenditures are beginning to shift from being categorized as “software costs” to being treated as “management decisions on par with labor costs” in corporate profit and loss statements. This change has a far greater impact on small and medium-sized enterprises than on large corporations.

What’s Happening in Large Corporations — Three Examples

First, let’s look at the numbers to understand what is happening now.

Tesla: AI Investment as ‘Prepaid Labor Costs’

Tesla continues to invest tens of billions of dollars annually in AI and autonomous driving. In the short term, this accelerates cash burn, testing investors’ patience. However, the essence of what Tesla is doing is simple: “Replacing future labor costs with current AI investments.”

Once autonomous driving is realized, the labor cost of drivers will drop to zero. The operating costs of robotaxis will only include vehicle maintenance, electricity, and AI computation costs. Labor costs will structurally disappear. Tesla’s massive investments are betting on this future.

Ramp: Reducing AI Model Operating Costs by 30%

Ramp, a corporate card and expense management company, announced that it has reduced the inference costs of its in-house AI models by 30%. What’s noteworthy here is not just that they have “implemented AI,” but rather the fact that they have entered the “phase of optimizing AI operating costs.”

In other words, AI has moved beyond the stage of “trying it out” to a stage where companies are considering “how to operate it as cheaply as possible.” This is similar to labor costs. Once you hire someone, it’s not the end; you must think about how to increase productivity and eliminate waste. AI has become a subject of the same management considerations.

Deezer: A World Where 90,000 Songs Are Generated by AI Daily

Music streaming service Deezer reported that over 90,000 AI-generated songs are uploaded to its platform every day.

Consider the implications of this number. It takes human musicians several days to weeks to create a single song. Including studio fees, engineering costs, and mixing fees, it can cost tens of thousands to hundreds of thousands of yen per song. With AI, however, the cost is nearly zero for 90,000 songs a day.

The marginal cost of content creation has effectively become zero. This is not just a story for the music industry; writing, images, and videos are all heading in the same direction.

The Real Question for Small and Medium-Sized Enterprises

Now, let’s get to the main point.

It’s meaningless to just look at the examples from large corporations and say, “That’s impressive.” What does this change mean for local small and medium-sized enterprises?

Let’s calculate it specifically.

The Real Cost of an Employee Earning 300,000 Yen a Month

Hiring one employee does not just cost 300,000 yen a month.

• Monthly Salary: 300,000 yen
• Social Insurance (Company Contribution): Approximately 45,000 yen
• Commuting Allowance and Benefits: Approximately 15,000 yen
• Recruitment Costs (Annualized): Approximately 20,000 yen
• Training Costs (Annualized): Approximately 10,000 yen
• PC and Equipment, Office Space: Approximately 10,000 yen

Actual Cost: Approximately 400,000 yen per month, or about 4.8 million yen per year.

Moreover, this employee only works 8 hours a day for 20 days a month. They may fall ill, and there is a risk of resignation. It can take three months to hand over their work.

The Real Cost of a 30,000 Yen Monthly API

On the other hand, let’s consider the case of integrating ChatGPT API or Claude API into business operations.

If we assign common tasks in small and medium-sized enterprises—such as handling inquiries, creating meeting minutes, drafting reports, organizing data, and composing emails—to AI:

• API Usage Fee: 10,000 to 50,000 yen per month (depending on usage)
• Integration Tools (Zapier, Make, etc.): 10,000 to 20,000 yen per month
• Initial Setup (Outsourced or In-House): 100,000 to 300,000 yen (one-time)

Monthly Running Cost: 30,000 to 70,000 yen. Annual Cost: 360,000 to 840,000 yen.

Available 24/7, no health issues, no resignations, and no handover required.

Where Is the Break-Even Point?

In simple terms, 4.8 million yen per year vs. 500,000 yen per year. There is a difference of about ten times.

Of course, AI cannot replace all tasks. Building trust with customers, making on-site decisions, and the final touches in handling complaints—these are jobs that only humans can do.

But conversely, there is a possibility that any work other than “jobs that only humans can do” can be shifted to AI. Have small and medium-sized enterprises ever calmly assessed how many of their tasks fall into the category of “jobs that only humans can do”?

In many cases, 40-60% of daily operations are “routine tasks done by humans.” By simply introducing AI here, the workload equivalent to one or two employees can be freed up.

This Is Not About Cutting Jobs

Please don’t misunderstand. This is not about “firing employees and replacing them with AI.”

The essence is the opposite. By assigning routine tasks to AI, current employees can focus on “jobs that only humans can do.” For sales, this means building relationships with customers; for manufacturing, it means improving quality; for management, it means focusing on strategy.

The biggest problem for small and medium-sized enterprises is the lack of personnel. Even if they want to hire, there are few applicants in rural areas. Even when they do come, they often do not stay. Therefore, wouldn’t it be more realistic to fill the gaps with AI?

For 30,000 to 50,000 yen a month, you can get an assistant who “doesn’t resign, works 24/7, and doesn’t complain.” Is there any reason not to use this?

Small and Medium-Sized Enterprises Have an Advantage in This Change

In fact, small and medium-sized enterprises are in a more advantageous position than large corporations in this change. There are three reasons.

1. Quick Decision-Making. While large corporations take three months to get approval for AI implementation, the president of a small or medium-sized enterprise can decide, “Let’s start using it today.”

2. Simpler Operations. Integrating AI into the complex workflows of large corporations is challenging, but the operations of small and medium-sized enterprises are relatively simple. They are a good match for AI.

3. Clearer ROI. If a company with five employees frees up one employee’s workload, that results in a 20% increase in productivity. If the same is done in a company with 5,000 employees, it falls within the margin of error.

Using AI in small and medium-sized enterprises is not just mimicking large corporations; it works precisely because they are small and medium-sized.

So, What Should We Do?

Here are three things you can start doing tomorrow.

1. Take Inventory of Operations. Write down all the tasks employees are currently doing. Highlight those that are “routine tasks that do not require judgment.” These are candidates for AI implementation.

2. Start with 10,000 Yen a Month. ChatGPT Plus costs 20 dollars a month (about 3,000 yen). Claude Pro is similar. First, the business owner should use it for a month. Meeting minutes, emails, project proposals—anything will do. There will definitely come a moment when you think, “I used to have employees do this.”

3. Budget for ‘AI Labor Costs.’ Instead of categorizing it as software or tool costs, create an account for “AI labor costs.” Monthly 30,000 to 50,000 yen. Annually 360,000 to 600,000 yen. Consider it as hiring someone who does more work than a full-time employee for less than the cost of one part-time worker.

Future Management Will Involve Considering the Placement of People and AI

Traditional management has centered around “who to hire and where to place them.” Going forward, the judgment of “should this task be done by a person or by AI?” will be added.

Labor costs and AI costs are now on the same playing field. This is not a threat. It signifies that small and medium-sized enterprises can compete based on judgment rather than financial power.

You can start for 30,000 yen a month. It won’t hurt if you fail. Just give it a try.

The answers are only found in the field.