Why a $1 Billion Startup Value Says Little About Lasting AI Business

A startup valued at $1 billion is not necessarily a lasting business. That is the argument of Ilman Shazhaev, founder and CEO of Dizzaract, in a guest piece for Crunchbase News. His test is simple: will customers keep paying once the early excitement fades?

Why the label falls short

A startup worth $1 billion on paper is called a unicorn. In the first half of 2026, 250 companies joined the Crunchbase Unicorn Board. Shazhaev says that pace raises a question about how many of them can hold on to customers. He would rather see $1 billion in yearly revenue from customers who keep paying.

He is not against fast sales. He says turning pilots into paying deployments shows a real problem is being solved. His concern is revenue built on discounts, experimental budgets or heavy compute subsidies, which gives little proof that customers will stay.

Cheap AI changes the picture

He points out that the cost of reaching a fixed level of AI performance has fallen thirteenfold every year since 2023. Startups that only resell access to AI models now face tough price competition. His example is OpenRouter, which raised a $113 million Series B. Developers can already switch between models through one shared interface.

Software built into a company's approvals, records and daily decisions is a different case. He says it can be much harder to replace than the AI model doing the work underneath.

The numbers he would check

• Whether customers renew at full price and expand their contracts.
• Whether margins stay healthy after the cost of running the AI, known as inference costs.
• Net revenue retention, which tracks how much revenue is kept from existing customers.
• How many deployments are live in real production use.

Where the money comes from

Shazhaev also says founders who know who keeps buying should raise money across regions. A local backer can open doors, because enterprise sales still depend on local budgets, buying procedures and relationships. He cites 1001 AI, based in London and the Gulf, which raised a $30 million Series A. Lux Capital led it, with Sanabil Investments, which is backed by Saudi Arabia's PIF. He notes the deal cannot prove future retention.

In his experience, venture firms ask how fast adoption can become repeatable sales. Sovereign-backed investors ask where user data is processed and which technical skills stay in the country. The second path can bring longer diligence, localization costs and contract terms that limit expansion elsewhere. He says neither pattern is universal.

What to do

This is an opinion piece, and the author runs his own company. It gives no retention figures for any named startup. Still, the questions are practical for anyone buying or backing AI tools: do customers pay full price again, and does the vendor still make money after delivering the service?