Title

WordPress Loses 19% to AI, GitHub’s AI Code Surges Tenfold—The Rapid Devaluation of ‘Code Writing Jobs’

Body

“Website creation will cost 500,000 yen”—how many more years will this estimate hold?

WordPress has lost 19% of its share of the internet. On GitHub, the volume of AI-generated code has surged tenfold year-on-year. Linus Torvalds, the creator of Linux, declared, “This is the new normal.”

What’s happening is simple. The cost of writing code is plummeting.

This isn’t just a story for engineers at large corporations. When local small and medium-sized enterprises think, “What should we do about our website?” or “What should we do about our business systems?” the fundamental pricing of their options is changing.

—

WordPress Loses 19% Share—What Happened?

For years, WordPress has been a giant supporting 43% of the web. “Let’s just use WordPress” has been the mantra for web creation among small and medium-sized enterprises.

Now, WordPress has lost 19% of its share to emerging platforms leveraging AI. Tools like Wix AI, Framer, and v0.dev are making a world where “you just give text instructions and a site is created” a reality.

What does this mean?

Traditionally, outsourcing a corporate website on WordPress would cost between 300,000 to 1,500,000 yen. This includes theme selection, plugin configuration, design customization, and responsive adjustments, all of which incur labor costs.

However, by using AI tools, you can simply input, “We are a company that provides this service, and our target audience is this,” and a decent website can be generated in just a few minutes. While it may not be perfect, the era where you can get a 70-point website for under 50,000 yen has already arrived.

The question is this: Would you choose “70 points for 50,000 yen” or “90 points for 1,000,000 yen”?

For many small and medium-sized enterprises, the answer is clear. How many of them truly need a “90-point website”? Corporate sites that serve as business cards, recruitment pages, service introductions—spending 1,000,000 yen on these is rapidly losing its rationale.

—

GitHub’s AI Code Surges Tenfold—The Day ‘Human-Written Code’ Becomes a Minority

The data from GitHub is even more shocking. The volume of AI-generated code has surged tenfold year-on-year. The number of users for GitHub Copilot has skyrocketed, and AI-written code is already mixed with human-written code in many repositories.

Let’s think about the numbers.

Traditionally, outsourcing a simple customization of a business system—say, “I need a feature to automatically generate quotes”—would cost between 200,000 to 500,000 yen. This is based on engineers working for 2 to 5 days at daily rates of 30,000 to 100,000 yen.

What happens when you use AI code generation? By conveying requirements to models like Claude, GPT-4, or Gemini, functional code can emerge in just a few minutes. Of course, it can’t be deployed directly; testing and modifications are necessary. However, the process of “writing from scratch” disappears. This alone cuts the workload by more than half.

A 500,000 yen development can become 100,000 yen. A 200,000 yen modification can become 30,000 yen.

This is a threat to programmers, but a chance for small and medium-sized enterprises. The constraint of “not being able to systematize due to budget constraints” is lifted.

—

The Significance of Linus Torvalds Declaring ‘New Normal’

Linus Torvalds, the founder of Linux, described the rapid increase in AI-generated code as the “new normal.”

To understand the weight of this statement, one must know who Torvalds is. He is the quality manager of the Linux kernel, the software that supports servers, smartphones, and cloud infrastructure around the world. He is arguably the strictest person in the world regarding code quality.

His acknowledgment that “it will become normal for AI-generated code to mix in” effectively puts an end to the argument that “AI-generated code is unusable.”

What’s important is the structural change that follows.

When the cost of writing code decreases, what happens?

1. The value of “what to create” increases over “creating.” As coding becomes cheaper, the value of personnel who can define requirements and design business processes increases relatively.
2. The value structure of “maintenance” changes. If something can be cheaply created with AI, it can simply be rebuilt if it breaks. The rationale for continuously paying a monthly maintenance fee of 50,000 yen is questioned.
3. The barriers to “in-house production” drop dramatically. Employees use AI tools to create things themselves. Outsourcing becomes less necessary. This option becomes realistic.

For local small and medium-sized enterprises, the third point is the most significant.

What was once considered impossible due to the lack of engineers—business improvements—may now be possible even for office staff. If you ask AI to “write a macro for a spreadsheet that automatically generates monthly invoices,” it will produce something functional. Even if it’s not perfect, if something that used to take 2 hours of manual work can be done in 10 minutes, that alone saves 20 hours a month.

—

arXiv’s Restrictions on AI Papers—Trust Issues with ‘AI-Created Content’

On the other hand, the rapid increase in AI code and AI content has raised another issue.

The academic preprint server arXiv has tightened restrictions on inappropriate content generated by AI. This is due to the influx of low-quality papers produced en masse by AI, which has increased the burden of peer review.

The same thing is happening in the world of code. AI-written code may “work,” but it is not necessarily “correct.” There may be security holes. It may not consider edge cases. Who guarantees the quality of the code generated in bulk?

This is a point that small and medium-sized enterprises should be cautious about.

Jumping at the chance to use AI because it’s cheap and deploying it without quality checks is dangerous. This is especially true for systems that handle customer data, payment-related functions, and forms containing personal information. Human oversight is necessary here.

In other words, while the cost of writing code decreases, the “cost of judging code” does not. In fact, it may even increase.

—

So, What Should Small and Medium-Sized Enterprises Do?

Let’s move beyond abstract discussions and outline three concrete actions.

1. Question Web Development Estimates

Check whether the estimates you are currently receiving or will receive are calculated based on “pre-AI market rates.” If you receive an estimate of 1,000,000 yen, ask, “What would the labor cost be if we used AI tools?” Just asking this can lead to a reduction of 30% to 50%. If it doesn’t decrease, consider changing vendors.

2. Conduct One Experiment of ‘AI × Business Improvement’ Internally

No grand projects are necessary. Try asking ChatGPT or Claude to “write a script to automate our monthly task of ___.” It doesn’t have to work perfectly. Create one experience internally where “asking AI yields something useful.” This is the first step.

3. Cultivate People Who Can Articulate Business Needs Over ‘Code Writers’

In an era where AI writes code, the value will increase for those who can clearly articulate, “This part of the business is wasteful, and I want to change it.” It’s better to focus on inventorying business flows than sending people to programming schools. The most accurate instructions for AI come from individuals who understand the business.

—

Conclusion: What Lies Beyond the Devaluation of Costs

The loss of WordPress’s share, the tenfold increase in GitHub’s AI code, and Torvalds’ declaration of the “new normal” may seem like separate news items, but they point to a single structural change.

The price of the act of “writing code” is collapsing.

For small and medium-sized enterprises that have previously said, “We have no technology, so we have to outsource” or “We have no budget, so we can’t digitize,” this change is a tailwind.

What used to cost 1,000,000 yen can now be done for 100,000 yen. What used to cost 100,000 yen can now be done in-house.

However, only the “creation” part is becoming cheaper. The ability to judge “what to create,” “is it really necessary?” and “is the quality acceptable?” is now more crucial than ever.

The companies that will win in this era of cost collapse are those that can say, “We can experiment more because we can create cheaply.”

Large corporations move slowly due to approvals and deliberations. Small and medium-sized enterprises can act the next day if the CEO says, “Let’s try it.” This agility is the greatest weapon in the era of cost collapse.

Question estimates, conduct one experiment first, and articulate business needs. I hope you start with these three steps.