Conclusion

Let’s get straight to the point. AI is getting cheaper because it’s not profitable.

Many people believe the decline in AI prices is due to technological innovation. This is half right and half wrong.

The real reason is much more visceral. AI companies are slashing prices to gain market share while bleeding red ink. In other words, the current prices are not “fair prices” but rather “abnormally low prices subsidized by investors’ money.”

Understanding this structure reveals what small businesses should do now.

—

Nscale’s IPO Documents Expose the “Uncomfortable Truth” of the AI Business

The IPO preparation documents submitted by European AI infrastructure company Nscale have garnered attention. What stands out is the scale of its losses.

In the AI infrastructure business, operating losses account for more than half of its revenue. The costs of building and operating GPU clusters, electricity costs, and cooling costs are devouring the revenue.

This is not just a story about Nscale. Even OpenAI is reported to have incurred about $5 billion in losses in 2024. Anthropic is in a similar situation. The entire AI industry is operating under the premise of “red ink.”

So why do they continue to lower prices despite these losses?

The answer is simple. They are operating under the logic of the platform war: “The one who captures users first wins.” This follows the same pattern as past platform companies like Uber, Amazon, and Netflix. They burn through investor funds to capture market share and recoup later.

What does this mean for small businesses?

“The current prices are abnormally low and should not exist.” A GPT-4 class model costs just a few yen per call. A year ago, it was several dozen yen. Prices may drop further next year, but only those “currently using the technology” will benefit from this abnormal low cost. If you wait, you will miss out on the lower prices.

—

The Indian IT Industry Begins to Evolve from “Man-Month Business”

Another significant shift is happening: the transformation of India’s outsourcing industry.

Major Indian IT firms—such as Infosys, TCS, and Wipro—have long profited from a model of “cheap labor multiplied by large numbers of man-months.” They deploy hundreds of engineers earning $15 to $30 per hour to handle system development and maintenance for companies in developed countries.

However, this structure is fundamentally changing.

With AI advancing automated code generation, testing automation, and document creation, the model of “more people equals more revenue” is beginning to collapse. In fact, Infosys has reported a net reduction in its workforce for the 2024 fiscal year.

Does this mean Indian companies are in decline? On the contrary.

They are shifting from “man-months” to “high-value services integrated with AI.” Specifically, they are focusing on areas like building AI agents, designing automated business processes, and constructing data pipelines. With a wealth of talent skilled in AI and the added advantage of low labor costs, India is taking a new position through the “cheap × smart” multiplication.

How does this relate to small businesses?

The Indian IT talent that was once only accessible to large corporations is now becoming available due to the proliferation of AI tools, allowing budgets of 100,000 to 300,000 yen per month. For example, you can hire an AI specialist team in India to design business automation, while implementation can be done using no-code tools and AI. Such combinations are becoming realistic.

A job that once cost 3 million yen when outsourced to a domestic SIer can now be done for 300,000 to 500,000 yen. This is not an exaggeration.

—

Spreadsheet AI—The Shock of “Automatically Ending Tasks”

The third piece is the emergence of AI batch processing for spreadsheets.

Services like “Delightful Cells” are emblematic of this trend. In essence, you can write instructions in Excel or spreadsheet cells, and the AI will process them all at once.

Let’s take an example:

• Enter the names of 100 companies in column A.
• Write in column B, “Research the industry of this company.”
• Write in column C, “Estimate the number of employees for this company.”
• Press the execute button.
• Three minutes later, all 100 rows are filled.

Previously, if you relied on part-time workers or outsourcing for such research tasks, it would cost 30,000 to 50,000 yen for 100 companies, with a turnaround time of 3 to 5 days. With AI batch processing, it can now be done for a few hundred yen in just a few minutes.

Costs are reduced to one-hundredth, and time to one-thousandth. This is not just “efficiency.” It means the very definition of work is changing.

AI operates on existing tools like spreadsheets that are already in use within the company. This is crucial. There’s no need to implement a new system, and training is almost unnecessary. Employees with low IT literacy can simply write instructions in Japanese in the cells.

Is there any lower barrier to AI adoption for small businesses than this?

Here are some specific use cases:

• Creating Sales Lists: Bulk processing of listing target companies, classifying industries, and researching contact information.
• Mass Generation of Job Advertisements: Automatically creating dozens of variations based on job type, location, and conditions.
• Analyzing Customer Surveys: Performing sentiment analysis and categorization on free-text responses.
• Translating Product Descriptions into Multiple Languages: Bulk translating descriptions for 100 products into English, Chinese, and Korean.

All of these tasks were previously ones that “only humans could do.” Now, they can be completed with “just write in the cell and execute.”

—

The Intersection of Three Trends Presents Opportunities for Small Businesses

Let’s summarize the discussion so far.

1. AI companies are lowering prices with the expectation of losses → The cost of using AI is abnormally low.
2. The Indian IT industry is providing affordable high-value services armed with AI → Development and design costs are decreasing.
3. Spreadsheet AI has emerged, enabling AI use at the operational level → Implementation costs are nearly zero.

All three of these trends are happening simultaneously.

Large corporations are responding to these changes by creating “AI strategy departments,” hiring consultants, and allocating budgets in the millions of yen. They take six months to define requirements, spend a year on implementation, and only start seeing results two years later.

Small businesses are different. They can start using AI next week.

By subscribing to a spreadsheet AI for a few thousand yen per month, they can first automate the creation of sales lists. If that goes well, they can then try generating drafts for estimates. Next, they can analyze surveys. One small step at a time, quickly.

While large corporations “plan before acting,” small businesses can “learn while acting.” This is the structure of reversal.

—

So, what should you do?

I will mention three things.

1. Try one spreadsheet AI by the end of this week.

Delightful Cells, SheetAI, GPT for Sheets—any of these will do. Many have free plans. Just start experimenting. If you run 100 rows of data, you will surely realize, “Hey, we can use this for our work on ○○.”

2. List three tasks that “don’t need to be done by humans.”

Identify tasks in your company that are “tedious but someone is doing them.” Data entry, research, classification, translation, draft creation—these types of tasks are highly likely to be eliminated by AI batch processing.

3. Consider the combination of “AI × overseas talent.”

By hiring an AI utilization team in India for business design at a monthly cost of 200,000 yen, and executing it with spreadsheet AI and no-code tools, cases are emerging where tasks that once cost 5 million yen per year can now be done for under 500,000 yen annually.

—

Finally

The reasons AI is getting cheaper are not solely due to technological advancements. A war of attrition between companies, changes in the global talent market, and the democratization of tools—all these structural changes are happening simultaneously.

It’s uncertain how long this “abnormal low price” will last. Once the platform war settles, prices may rise.

That’s why the greatest investment is to build the skills and systems of the “users” within your company now.

There’s no need to buy expensive tools or hire advanced AI talent. Just write instructions in a spreadsheet and press the execute button. That’s where it all begins.

The strength of small businesses lies in their speed of decision-making. I encourage you to try at least one by next Monday.