The future of AI code tools

By Aatu Väisänen

November 20th, 2025

I might stir up the hornet's nest with this one... but here's my hot take: most of the new, shiny AI code editors and AI-to-app tools (a.k.a. vibe coding platforms) will lose relevance in a couple of years.

Why?

Well, let me explain by starting with the basics.

AI code editors... what are those?

Any regular code editor (for example VS Code or a full-fetched IDE like JetBrains) becomes an AI code editor the moment you glue AI (a large-language model, LLM) to it and ask the AI to write code for you automatically. In a nutshell, this can mean three things:

1. Suggesting to auto-complete the lines that you are typing

2. Profoundly explaining parts of the code project for you...

3. ... Or writing all of the code based on your prompts: this is the so called "agentic mode".

Popular examples of these purpose-built AI code editors are Cursor and Windsurf, which have gained a lot of traction in the past year or so. Both of them are VS Code forks...

Wait what, so they are just built on top of one of the most popular code editors in the world?

The bread and butter of vibe coders: AI-to-app tools

AI-to-app tools, like Lovable, v0, and Bolt.new are solely focused on agentic AI. In a nutshell, each of these tools are the same as the agentic mode in an AI code editor, the code side of things is just abstracted behind a fancy loading screen and a chat input. In fact, you could in theory accomplish most of what these tools do in a AI code editor like Cursor, given that you have crafted the prompts needed for this purpose.

There is nothing special about most them!

Basically anyone can create their own version of a AI-to-app tool. Heck, Vercel even encourages to build your own one using their v0 Platform API [1]. It's the same with AI code editors: with Microsoft announcing their latest progress on making VS Code the go-to open source AI code editor [2], the ground has been laid for anyone creating their own hyped fork of VS Code, and calling it the most revolutionary AI code editor ever.

And for the explanation...

So, what's my point with all of this?

Since the release of ChatGPT in November 2022, the AI hype train has been churning forward without signs of stopping. As we head toward 2026, the big tech AI infrastructure spending spree will pass a trillion dollars. All of the additional compute power will mean that AI will become even more commonplace than now, especially in the realm of coding. However, with the vast amount of new AI code generation tools popping regularly, I'd argue that the demand for them just isn't there yet.

Let's look at history for a reference. When the internet bubble was about to burst, there were a lot of startups doing basically the same things built around internet hype, just packaged in their own neat versions. Ultimately when the bubble burst, most of these startups vanished into thin air: only a few key players barely survived. The lesson from this? When a new revolutionary technology advancement happens, the transition towards fully harnessing the potential for said technology doesn't happen overnight. For internet it took over a decade before its true potential was unchained.

Conclusion. Is an AI code tool meltdown coming?

With the abysmal amount of money being spent in the hopes of making AI the biggest productivity leap in the 21st century, it's quite challenging for me to understand what the end-game of the hundreds of AI code startups will be. When the tech giants are essentially burning money in order to win the AI race, how could any of these AI startups compete? The startups are also just burning cash without any clear vision to turn profitable. It means trouble for these startups when venture capital turns of their money streams. Not saying it will happen anytime soon, but it can happen.

Let's take Cursor as an example: when it first came out, sure it was pretty cool and innovative, but it didn't last long until a bunch of other startups spinned up their own version of an AI code editor. Additionally VS Code (again, which Cursor was built on) has adopted most, if not all of the features that were new and fancy in Cursor originally.

Many of these tools will eventually be bought by bigger companies (like Windsurf was acquired by Cognition), but even still: I'm pretty sure that most of the currently available AI code tools / vibe code generators will eventually die out, and just the tools owned by big tech will prevail.

Sources:

[1] Build your own AI app builder with the v0 Platform API https://vercel.com/blog/build-your-own-ai-app-builder-with-the-v0-platform-api

[2] Open Source AI Editor: Second Milestone https://code.visualstudio.com/blogs/2025/11/04/openSourceAIEditorSecondMilestone