What Happens When the Cost of AI Drops to One-Hundredth?

First, let’s talk about the numbers.

In 2023, the API usage fee for GPT-4 was about $0.03 per 1,000 tokens. As of 2025, models with equivalent or better performance can be accessed for around $0.0003. That’s one-hundredth of the cost.

Google has significantly expanded its free tier for Gemini. Meta has released its top-tier model, Llama 4, as open source. Both Claude and ChatGPT provide sufficiently practical responses under their free plans.

This is not just a story about “AI becoming cheaper.” It’s about the structural reality that “there is now almost no cost associated with using AI.”

When costs drop to one-hundredth, what happens? The answer is simple: the advantage of having money disappears.

So, what becomes the new advantage? This is the crucial point for small and medium enterprises (SMEs).

The Era When the ‘Weight’ of Large Corporations Becomes a Burden

What happens when large corporations adopt AI? Based on my experience in the field, I can say this:

First, it takes three months for internal approvals. Two months for security assessments. Three months for vendor selection. Six months for proof of concept (PoC). By the time results come in, the model they intended to use is already two generations behind.

There’s a story about Cisco giving unlimited AI tokens to a 27-year-old engineer to rewrite 98% of legacy code in two years. The noteworthy part is the “two years” timeframe. Even in large corporations, this much time is required.

On the other hand, what about small and medium enterprises?

If the CEO says, “Let’s do it,” they can start the next day. In a company of ten, everyone can begin using AI from next week. There’s no need for approval processes, vendor selection, or cross-departmental coordination meetings.

When the cost of AI approaches zero, the deciding factor becomes not “financial power” but “the speed of decision-making.”

This creates a structurally advantageous game for SMEs.

Specifically, What Has Become ‘Practically Free’?

Let’s stop with the abstract discussions. Here are some examples of costs that have dramatically changed in the field of small and medium enterprises.

1. Initial Customer Support Responses

Traditionally, hiring one part-time employee for phone support costs around 150,000 to 200,000 yen per month. With a team of three, it exceeds 500,000 yen per month. That’s 6 million yen annually.

Now, what is the cost of installing an AI chatbot on a website? It ranges from 10,000 to 30,000 yen per month. Even for high-functionality options, it’s under 50,000 yen. Annual costs drop from 6 million yen to 360,000 yen. That’s about a 94% reduction.

Moreover, AI can provide 24/7 support. There are no missed opportunities due to “outside business hours” for inquiries at night, on weekends, or during holidays.

2. Translation and Multilingual Support

Many local manufacturing companies engage in overseas transactions. Traditionally, sending one English email to a translation company costs between 3,000 and 5,000 yen. For 20 emails a month, that adds up to 60,000 to 100,000 yen.

Now, if you throw it to Claude or GPT, practical translations come back in seconds. The cost, if using the API, is less than 0.1 yen per email. It’s practically zero.

A local food manufacturer used this to launch their own English, Chinese, and Korean e-commerce sites. Since translation costs have become almost zero, there’s no longer a reason not to do it.

3. Meeting Minutes and Report Creation

Many companies spend 30 minutes to an hour creating meeting minutes each time. If there are 20 meetings a month, that’s about 200 hours annually. Even at an hourly wage of 2,000 yen, that’s 400,000 yen in labor costs.

Now, if you record the meeting and feed it into Whisper (OpenAI’s speech recognition, which is free to use), it gets transcribed in minutes. If you then have AI summarize it, the minutes can be completed in five minutes. 200 hours drop to 20 hours.

4. Recruitment and Job Description Creation

If you request a production company to create job advertisement copy, it costs between 30,000 and 100,000 yen per piece. By inputting your company’s information and desired candidate profile into AI, you can generate a draft in five minutes. The only thing humans need to do is the final check for “company uniqueness.”

All of these examples are already a reality. They are not future possibilities.

Not ‘What Can We Do with AI?’ but ‘What Can We Let Go of with AI?’

Now we get to the main point.

When AI becomes free, many people think about “what can we do with AI?” But what we should really be considering is “what can we let go of because of AI?”

What we should let go of is clear: “tasks that do not need to be done by humans, but have only been done by humans until now.”

• Data entry
• Standard email replies
• Monthly report creation
• Inventory counting
• Drafting social media posts

What remains after letting these go?

Time to talk face-to-face with customers. Time to walk the floor. Time to think about new products.

The strength of small and medium enterprises lies in the CEO remembering customers’ names. It’s in the craftsmen on the floor knowing, through experience, that “this material should be handled this way.” These are areas of human judgment and relationships that AI cannot replicate.

By delegating mundane tasks to AI, humans can focus on what only humans can do. This is the true meaning of “using AI.”

Breaking Down Personalization Through ‘Systematization’

There’s one more thing that small and medium enterprises should gain from AI: reproducibility.

The biggest risk for SMEs is personalization. A state where “the business cannot run without that person” is a fundamental management risk.

By using AI, the tacit knowledge of veteran employees can be transformed into systems.

For example, by training AI on sales talk patterns, it can automatically generate talk scripts for newcomers. If you input the criteria for estimating costs into AI, you can produce estimates with 80% accuracy even without a veteran present. By accumulating knowledge for handling complaints in AI, consistent quality can be maintained regardless of who handles it.

The cost of breaking down personalization has dramatically decreased. This is one of the most significant structural changes brought about by AI’s “freeing up.”

So, What Should We Do?

I’ll say just three things.

1. Tomorrow, try delegating just one task to AI.

Whether it’s meeting minutes, drafting emails, or anything else. The free plan of ChatGPT is sufficient. There’s a much larger gap between “having tried it” and “never having tried it.” First, bridge that gap.

2. Ask, “Is this something that needs to be done by humans?” once a week across the company.

Take just five minutes at the end of meetings. Ask, “Is there anything we did this week that could be delegated to AI?” If you continue this for three months, the landscape of the company will change.

3. Secure a monthly budget of 50,000 yen for ‘AI experimentation.’

This can cover upgrades to paid plans, API usage fees, or the costs of introducing small tools. With a budget of 50,000 yen a month, you can try a lot. An annual budget of 600,000 yen can lead to hundreds of thousands of yen in operational efficiency gains. The return on investment is extraordinary.

What Lies Ahead When AI Becomes Free

Finally, I want to pose one question.

What is your company’s ‘true strength’?

In a world where AI has become commoditized, “being able to use AI” is no longer a strength. Everyone can use it.

What becomes a strength are things that cannot be replaced by AI: relationships with the community, deep understanding of customers, long-cultivated skills, and on-site judgment.

AI is a tool that “amplifies” those strengths. It compensates for weaknesses and creates time to focus on strengths.

Large corporations use AI to “streamline.” Small and medium enterprises use AI to “concentrate.”

Even with the same tools, the meaning of how they are used is entirely different. And, the company that can ‘concentrate’ will ultimately be stronger.

As the cost of AI approaches zero, the ‘agility’ that small and medium enterprises possess has become an unprecedented weapon.

The question is whether they will wield that weapon.