Kristian Tanninen

Kristian Tanninen

Product CTO

Chat with AI
Kristian Tanninen

Builders' Report·

Where did the AI go? What founders build in November 2026

In 2025, 286 billion dollars of private money went into AI in the US. Yet three in four products founders set out to build right now are ordinary software. So where did the AI go?

3 in 4

products founders set out to build right now are ordinary software

25%

are built around an AI model, against 24% in the nine months before

+92%

more interest in AI prototypes, in one month

Our data · open in the Product CTO app

AI is the way to build tools for everything.

These numbers come from our own data. It is also open in the Product CTO app, for anyone who wants to dig further. They cover about 3,000 products that startup and bootstrapped founders set out to build between late August and mid-November 2026, compared with the nine months before.

First, the money. Stanford's AI Index 2026 puts it in one line:

“U.S. private AI investment reached $285.9 billion in 2025”

Now the products. B2B SaaS platforms, marketplaces, dashboards and mobile apps lead what founders set out to build. Products built around an AI model are a steady quarter of the list, and that share barely moved in a year.

The two ends meet in how the software gets made. Menlo Ventures tracked where company money for generative AI went in 2025:

“Coding is generative AI's first "killer use case," with $4B in enterprise spend”

Lovable is the clearest case. It writes an app from a plain description, and the business behind it grew fast:

“Lovable crossed $400 million in annual recurring revenue in February”

In Stack Overflow's 2025 survey, 84 percent of developers said they use or plan to use AI tools.

So AI did not replace the ordinary product. It became the way ordinary products get built. AI is not the tool for everything; it is the way to build tools for everything, and the money followed the builders.

That changes what a founder should build in 2027, and our market data already shows the move. Making a first version became cheap, so more people make one: interest in AI prototypes almost doubled this month, and fintech MVPs rose about as fast. Simple products a prompt now produces are fading: interest in ready-made marketing material and simple online booking systems fell by almost two thirds in a month.

What grows instead is everything after the prototype. The two needs that grew fastest without any AI in them are about the backend: the backend behind a mobile app concept, and a backend that holds up in production. Both grew by about two thirds in a month. The research on AI-written code shows why that part stays hard. Veracode had AI models write code and tested every sample:

“45% of code samples failed security tests”

Google's DORA research found the same pattern in delivery:

“an estimated reduction in delivery stability by 7.2%”

So 2027 moves from vibe-coded prototypes to AI-delivered products. The idea and the first version are the cheap part. The products that last are built with AI and built to hold up once people pay for them.

Sources

  1. Stanford HAI, AI Index Report 2026, 2026
  2. Menlo Ventures, 2025: The State of Generative AI in the Enterprise, Dec 2025
  3. TechCrunch, Lovable says it added $100M in revenue last month alone, with just 146 employees, Mar 2026
  4. Stack Overflow, Developer Survey 2025: AI, 2025
  5. Veracode, 2025 GenAI Code Security Report, Jul 2025
  6. Google Cloud, Announcing the 2024 DORA report, Oct 2024

Research and opinion from data and public sources. Not investment, legal or business advice.

Reel transcript · 1:35

Last year, almost three hundred billion dollars went into AI. Yet three in four new products are plain software. So the question is, where did the AI go.

Stanford HAI · AI Index 2026

First, the money. Stanford's AI Index has the number. Two hundred and eighty six billion dollars. Private AI investment in the US, in 2025.

Now the products, from our own data. Three in four products founders set out to build are ordinary software. SaaS platforms, marketplaces, apps. And the share built around AI barely moved in a year.

Menlo Ventures · State of Generative AI 2025

The two ends meet in how software gets made. Menlo Ventures puts it plainly. Coding is generative AI's first killer use case.

TechCrunch · March 2026

Look at Lovable. It writes an app from a plain description. In February, it passed four hundred million dollars in yearly revenue.

So AI is not the tool for everything. It is the way to build tools for everything. And the money followed the builders.

That moves the market for 2027. Prototypes got cheap. Interest in AI prototypes almost doubled in a month. Simple tools a prompt can make fell by two thirds.

Veracode · GenAI Code Security Report 2025

What grows is everything after the prototype. Veracode tested code that AI wrote. Almost half of the samples failed security tests.

So 2027 moves from vibe-coded prototypes to AI-delivered products. Build with AI. And build it to hold up once people pay for it. The data behind this report is open in the Product CTO members area.