Ronen Chernin predicts explosive AI adoption growth from enterprises like Revolut, Shopify, Prosus, and Booking.com once they solve their initial 'cold start' problem of building internal AI developme...

According to Ronen Chernin, most enterprises are only utilizing AI in the first percent of their potential use cases and volume.

Ronen Chernin predicts that the world will need tens or hundreds of times more AI compute capacity than is currently available.

Ronen Chernin believes the biggest threat to Nebius is the potential for the AI market to become too consolidated among a few major players.

Nebius raised its prices by 30% a couple of months ago and continues to experience strong demand pressure on its supply.

The key to a successful partnership with NVIDIA is gaining the respect of their engineers, as NVIDIA is largely an engineer-driven company.

Nebius's future fourth product layer will focus on agentic workflows, where the platform will automatically select the best model to execute a given task.

Nebius has a CapEx program of $20-25 billion for the current year.

Nebius's managed inference product is called Nebius Token Factory and is targeted at vertical AI companies and enterprises.

Nebius customer Revolut initially had 99% of its inference budget allocated to closed models like OpenAI before shifting towards open-source alternatives.

Approximately 70-75% of the new capacity Nebius is building in the medium term is located in the United States.

Nebius differentiates from competitors like CoreWeave through its full-stack integration strategy, which includes building its own data centers, racks, and servers, as well as a comprehensive upstream...

The CapEx programs of Nebius's hyperscaler competitors are approximately 8 times larger than its own.

Nebius's product strategy consists of four layers: raw bare-metal infrastructure, a multi-tenant cloud, a managed inference platform, and a future layer for agentic workflows.

The shift to open-source models does not hurt frontier model providers like Anthropic and OpenAI because they continuously advance to new, more complex problems in a large, unaddressed market.

Following the release of a DeepSeek model about 15 months ago, Nebius's stock dropped 40% in one week, yet the company experienced its best sales week to date during that same period.

As customers scale their AI products, many shift from frontier closed models to tunable, open-source models to improve economics and quality for specific use cases.