AI ends the traditional software industry, just as the internet ended traditional media

AI programming tools are driving the software industry towards a "peak moment."

Recently, Doug O'Laughlin, founder of the chip financial analysis company Fabricated Knowledge, published an article stating that generative AI is reshaping the software industry in a way similar to how the internet disrupted traditional media.

The article points out that AI tools, such as Cursor and Claude Code, are significantly reducing the cost and time of coding, with development tasks that used to cost thousands of dollars per token now potentially costing only a few cents.

Doug believes that this will not only change the rules of the software development game but will also completely disrupt the industry's profit models and competitive landscape, much like YouTube's impact on traditional television.

By comparing the changes in the media industry during the internet era, Doug suggests that the software industry is facing a similar "peak moment." The best perspective to understand this change is to reference the impact of YouTube's rise on traditional linear media.

How Traditional Media Declined

To better understand the changes the software industry is about to face, the author first reviews the development trajectory of the traditional media industry.

Before the advent of the internet and YouTube, content consumption was primarily conducted through cable TV bundles, allowing users to access thousands of channels with a single subscription. This model persisted from the 1980s to the mid-2010s, with the penetration rate of cable TV bundles peaking at 87%.

However, the rise of YouTube changed everything.

Data shows that the S-curve of YouTube user growth almost perfectly coincides with the peak time of cable TV bundles.

The article states that in the 2010s, when YouTube experienced explosive growth, cable TV bundles reached their peak and then began to decline. Today, the penetration rate of traditional television has dropped to around 40%.

The Significant Lowering of Content Creation Barriers

What led to the decline of traditional media? The author believes that the key lies not only in the internet changing the distribution channels but also in the fundamental changes in media creation itself.

In the era of traditional media, content creators were relatively few (thousands of channels), whereas now there are approximately 113.9 million channels on YouTube, with over 32,300 channels having more than one million subscribers Taking MrBeast as an example, this content creator now has 400 million subscribers, and the article points out that this is more than the entire population of the United States, far exceeding the scale that cable television once achieved.

Importantly, MrBeast did not enter this field with a large amount of funding and distribution channels, but rather rose to prominence due to the significant decrease in the entry costs of media creation.

The author estimates that launching a television program in 2000 required about $250,000, while starting a YouTube channel now only costs about $3,000. This trend of declining costs is remarkably similar to the development trajectory of current programming assistant tools.

The "Cliff-like" Decline in Software Development Costs

The same logic applies to the software industry under the impact of AI. The article emphasizes that the cost of software development today is just a fraction of what it used to be.

Doug states that in the past, writing hundreds of lines of code would cost hundreds of dollars in labor time, but now spending hundreds of dollars a day can produce millions of lines of code.

Skeptical viewpoints argue that service quality, product-market fit, and distribution channels (i.e., sales) are the real differentiating factors. They point out that unlike traditional media, where both distribution and content costs have changed, the software industry is only seeing a decrease in creation costs, while distribution costs remain unchanged.

However, Doug believes this viewpoint is too narrow.

He states that while SoR (System of Record, the authoritative data source in large data systems) is unlikely to change, new business opportunities do exist. If the context window becomes infinite, allowing for endless invocation and manipulation of data and information, then what significance does a specific solution have compared to others?

Doug further points out that traditional software industry gross margins exceed 90%, but if everyone can develop similar products at low cost, then the days of high profits may be gone forever.

Although generative AI is costly, it essentially reduces the net costs of traditional software solutions for product creation.

The Arrival of the "Peak Software" Era

Doug boldly predicts in the article that the era of making huge profits through software is over.

The article states that the past game plan was to hoard a relatively small number of excellent programmers to increase internal development speed and refuse to provide this supply to competitors. But AI programming agents mean that software supply will grow exponentially.

Today, the software industry is undergoing an accelerated disruption, with a surge in supply that will overwhelm traditional software manufacturers, and niche solutions will ultimately surpass existing giants.

Similar to the media industry, the article points out that the SaaS (Software as a Service) model in the software industry is at a similar peak, and the proliferation of AI tools will accelerate this decline process. The surge in supply means that the high-profit "moat" of traditional software companies will gradually lose effectiveness, and sales and marketing costs will become the focus of competition, but this resembles a race of "downward competition."

The article predicts that traditional software companies still have opportunities to make money, but this will be achieved through a wave of consolidation. Today's Microsoft, Salesforce, ServiceNow, and Adobe may become tomorrow's Disney and Fox. **

The Essence of Software: Just a "Local Minimum" of Hardware?

Doug also proposed a more philosophical viewpoint: software itself may just be a "local minimum" in technological advancement, with the true core always being hardware.

He explained that, looking back at the history of computing, most software was initially sold to users in bundled packages. For example, IBM's early products.

The concept of software being charged as independent intellectual property actually emerged only during a specific historical stage after the popularization of hardware.

In a world of infinite software generation, what is the significance of software? The article points out that it is merely an output of hardware.

Hardware is the intersection of information and the physical world, and it is the fundamental driving force behind technological advancement. As AI tools make software generation infinitely possible, the value of software may return to the hardware itself, and future devices may come with built-in customized software, just like the early Cray computers.

In this model, hardware will become the new scarce resource, and value will be refocused on infrastructure such as chips and computing power.