Snowflake и OpenAI: союз за 200 миллионов долларов (и при чем тут Anthropic)

Snowflake decided that having just one favorite in the artificial intelligence world wasn't enough. After the company issued a check for $200 million to the Anthropic startup in December, it was OpenAI's turn. The sum is the same, the ambitions even higher.

While investors are selling software company stocks in panic, Snowflake's CEO demonstrates Olympic calm and continues to spend money. And this is not just corporate shopping, but a strategic attempt to survive in a world where simply storing data is no longer enough. Why would a cloud giant need to duplicate partnerships?

The answer lies in a strategy of technological agnosticism. Enterprise customers are a capricious and cautious bunch. One wants Claude models for their ethics and long context window, another wants GPT-4 for its sheer power and developed ecosystem.

By providing access to both giants through its platform, Snowflake transforms from an ordinary data warehouse into a supermarket of neural network solutions. You're not just storing your tables with them, you're bringing them to life with the best tools on the market, without leaving your security perimeter. By the way, about security and convenience.

Along with the deal, Snowflake rolled out Cortex Code—its answer to GitHub Copilot. This is an AI assistant for code writing that lives directly inside the platform. The point is for analysts and developers to be able to write SQL queries and Python scripts by simply describing the task in words.

It's a logical step: if you control the data, you should control the tools for working with it. Snowflake is trying to eliminate the middleman so users don't have to copy data to external services for analysis. But why is the market so nervous?

The large-scale sell-off of software sector stocks, which Sina Finance mentions, suggests that investors have begun to doubt the "AI bubble." Will these crazy expenses on API contracts pay off? Snowflake spends hundreds of millions just to have the ability to resell other people's brains.

For now, this looks like an arms race where everyone is afraid of missing out on the division of the pie, even if the pie itself hasn't been baked yet. In this game, Snowflake is betting that companies will be too lazy to move their petabytes of data to Microsoft or Google clouds for their built-in models. It's easier to pay Snowflake extra and get the same models "in place."

This is a classic lock-in strategy wrapped in the beautiful packaging of innovation. If you're already "sitting" on Snowflake, you now have one less reason to look at competitors like Databricks. The main thing: Will Snowflake become a universal Swiss Army knife for corporate AI or turn into a simple and very expensive middleman between OpenAI and the end user?