Same Week, Three News Stories Reveal the Map of AI Costs

China announced the launch of an AI-driven cancer vaccine manufacturing line in Beijing. NASA is advancing tests of an AI medical support system in space. Meanwhile, the voice dictation tool “Privatewhisper.ai” has been released for free.

All three stories revolve around “AI.” However, the cost scales are entirely different.

On one hand, we have billions of yen; on another, a national budget in the hundreds of billions; and on the other, zero yen. Although they are all categorized under the term “AI utilization,” the realities are worlds apart.

When we line these three up, a certain structure emerges: the boundary between the “areas where AI costs disappear” and “areas where enormous amounts of money are still required.” Misjudging this boundary could lead small businesses to make unnecessary investments or, conversely, to overlook free resources. Both scenarios can be fatal.

—

Cost-Free Area: Voice Dictation

To get straight to the point: Voice input is now free.

Privatewhisper.ai is a local voice dictation tool based on OpenAI’s Whisper model. It operates in the browser, and voice data is not sent to servers, ensuring privacy. The price is zero yen.

You might think, “Voice input has existed for a while.” Indeed, it has. But the accuracy is different. Whisper-based models have reached practical levels of recognition accuracy for Japanese, including automatic punctuation insertion. The quality that used to require a monthly subscription of several thousand to tens of thousands of yen for voice recognition SaaS is now available for free.

Consider what this means.

In a local small business, a craftsman can speak their work report into a smartphone, and it turns into text, creating a daily report. A veteran employee in their 60s, who struggles with typing, can finish their report in five minutes. The digitalization bottleneck that previously halted progress due to “being bad at IT” can now be eliminated with voice input.

Implementation Cost: 0 yen. Training Cost: Almost 0. Effect: Daily report preparation time reduced from 30 minutes to 5 minutes.

Finding such “areas where costs have disappeared” and making full use of them is the correct approach for small businesses to utilize AI.

—

Billion-Yen Area: AI-Driven Cancer Vaccine Manufacturing

On the other hand, the AI cancer vaccine manufacturing line announced by China is an entirely different story.

This manufacturing line, set to be established in Beijing, designs and produces mRNA-based personalized cancer vaccines using AI. It analyzes the genetic mutations of each patient’s tumor and designs a vaccine specifically for them. The design process, which traditionally took several months, is shortened to a matter of days or weeks with AI.

It’s an impressive technology. However, building this manufacturing line is estimated to require an investment in the billions of yen. Clean rooms, GMP (Good Manufacturing Practice) compliant facilities, bioreactors, and quality control systems—the costs of the “physical infrastructure” needed to operate the AI model are astronomically high.

What’s noteworthy here is that AI reduces “design costs” and “time,” but not the “cost of manufacturing infrastructure.”

While the intellectual labor costs for vaccine design have dramatically decreased due to AI, the facilities required to actually produce the vaccine, the systems ensuring quality, and the processes to meet regulations still require significant investment.

This structure is crucial to remember: the “costs reduced by AI” and the “costs that do not decrease with AI” are different. In the software realm, costs can plummet, but when it comes to hardware, regulatory compliance, and physical infrastructure, costs remain high.

—

National Budget Area: NASA’s Space AI Medicine

NASA’s AI medical support system presents an even more extreme example.

When astronauts head to Mars, the communication delay with Earth can be as much as 24 minutes. They cannot consult ground-based doctors for urgent medical decisions. Therefore, AI is being developed to monitor astronauts’ health, assist in diagnoses, and, in some cases, guide treatment procedures.

The budget for this project is publicly estimated to be in the range of several million to tens of millions of dollars (hundreds of millions to billions of yen). Moreover, the technical hurdles are immense, including radiation-hardened hardware that operates in space, edge AI processing under extreme communication constraints, and ensuring the reliability of life-or-death decisions.

While there is no direct relevance for small businesses, it is essential to observe where this technology will eventually land.

If a level of “AI medical care that operates even when communication is lost” is achieved in space, it could be repurposed for remote medical care. AI could assist in primary diagnoses in clinics on remote islands or mountainous areas where specialists are unavailable. Such a future could be realized through technologies developed with NASA’s budget making their way into the private sector.

Technologies developed with national budgets will become affordable for small businesses in five years. Whether one can anticipate this time lag will determine their ability to capture first-mover advantages.

—

Boundary Map: Where Should Small Businesses Compete?

When we organize these three examples, the AI cost boundary map looks like this:

The important takeaway from this map is that the top two areas are rapidly heading towards “zero costs.”

An AI transcription service that cost 20,000 yen per month a year ago can now be replaced with free open-source alternatives. An AI chatbot that cost 50,000 yen per month six months ago can now be built for several thousand yen using the ChatGPT API.

The faster the costs decrease in a given area, the more advantageous it is for early adopters. This is because while competitors are still “waiting and seeing,” they can integrate AI into their business processes and systematize them. The tools are available for anyone to use. The differentiator will be the accumulation of operational know-how on “how to integrate it into one’s business.”

—

Three Actions Small Businesses Should Take Today

1. Implement voice input for all employees (Cost: 0 yen)

Whether using Privatewhisper.ai or the standard voice input on iOS/Android, make “speaking to input” an option for all employees. The benefits will be greatest for employees who struggle with typing. Test it in all scenarios where text generation is required: daily reports, meeting minutes, email drafts.

2. Try three AI tools under 10,000 yen per month (Cost: under 30,000 yen per month)

Start with tools like ChatGPT Plus (3,000 yen per month), Canva AI (free to 1,500 yen per month), and Google NotebookLM (free). If there’s no effect after three months, you can stop using them. You can conduct an experiment for 30,000 yen over three months. Even if you fail, it’s just the cost of two nights out at an izakaya.

3. Avoid the “tens of millions of yen” area (Savings: tens of millions of yen)

“AI-equipped manufacturing lines” and “fully customized AI system development” are the domains of large corporations. If small businesses dive into these, it could take ten years to recoup their investments. Instead, look for SaaS services that offer monthly subscriptions. There are plenty of options that can be started with zero initial investment.

—

The Real Question is Not Whether to Use AI

Voice input has become free. Chat AI has dropped to 3,000 yen per month. Image generation is available for several thousand yen per month.

In this situation, companies debating whether to implement AI are already falling behind.

The real question is, “What will you change in your business in the areas where costs have disappeared?”

China has compressed the design time for cancer vaccines using AI. NASA is looking to delegate medical decisions in space to AI. They are not debating whether to use AI; they are considering “What constraints can we remove with AI?”

Small businesses should do the same. They can remove the constraint of “employees who cannot type” with voice input. They can eliminate the constraint of “documents that cannot be written without specialized knowledge” with AI chat. They can bypass the constraint of “promotional materials that cannot be created without outsourcing to a designer” with image generation.

Beyond the constraints removed, there lies a nimble winning strategy that large corporations cannot replicate.

Large corporations take three months for approvals. Small businesses can start tomorrow. Large corporations spend a year selecting a unified tool for the entire company. Small businesses can begin using tools with just the CEO’s approval today.

In areas where AI costs have disappeared, “size” does not confer an advantage. “Speed of action” becomes the advantage.

For small businesses, now is the best time. Identify the areas where costs have disappeared and take action starting tomorrow.