$Microsoft(MSFT.US) I read that ghost story "2028 Global Intelligence Crisis" today. The idea that a single developer could use Claude Code to develop the core functionality of a medium-sized SaaS made me think seriously, and I don't think it's entirely valid.

First, it's an objective fact that one developer + Claude Code can indeed build the core functionality of a SaaS, and I believe Claude Code's capabilities will go far beyond that in the future.

But the question is, how much cheaper would the development and later maintenance costs of such an internal SaaS application really be compared to directly subscribing to a mature SaaS tool?

If we're only looking at it from a cost perspective, why aren't people using the many high-quality open-source SaaS projects on GitHub, instead of starting from scratch and handcrafting one with AI? Wouldn't that incur higher development time and human resource costs?

AI is a brain, not just a more efficient tool killing another tool. If it were merely a tool, then AI wouldn't have the value we imagine.

Therefore, I don't think AI will kill companies like Microsoft. On the contrary, it will make companies that already have strong moats and software ecosystems even stronger, with higher subscription rates for their seats.

What will definitely be affected are jobs like customer service and related SaaS. The characteristics of such SaaS are that the workflows are highly repetitive yet still rely on human labor, or those leaning towards automation and information aggregation (like Ctrip, Meituan) will face a fatal impact.