Daytona: From a Linkedin Comment to Unicorn

Daytona just raised $60M at a $1B valuation. We quadrupled down in this round as well but my favorite part of the story happened looong before.

Seven months before our investment, Ivan randomly commented on my post about devtools. We did not know each other. He said it was great that we were investing in devtools and added that he would make sure we did more.

Ivan is the strongest founder I know on social media and nothing he does there is random. He chose us as investors before we even knew it.

Most founders use social media as a distribution channel. They announce product launches, hiring news and financing rounds etc.

‍

Ivan uses it as a company building tool.

He attracted customers, recruited employees and shaped Daytona’s messaging, all publicly on social media. He did not wait until he was fundraising to connect.

He publicly told us that we were going to invest.

The final part of the story came later.

Daytona started growing and we didn’t want to wait for the next fundraising process. We preempted the seed round.

Ivan publicly told that story too. The existing investor that preempted Daytona’s seed round was us.

Daytona’s revenue has grown 25x since January

Anthropic, OpenAI, NVIDIA, Cursor and Cognition are all customers.

We saw Daytona as a devtools investment but I think that description is too small today. The unit of cloud demand is changing.

Humans and applications initiated most computing workloads. AI agents can create thousands of autonomous workloads at once. They execute code, browse, interact with internal systems, pause, restart and branch into new tasks.

Every one of those agents needs a computer that is isolated, secure, and able to preserve state, while the compute underneath it expands and contracts with the task.

Traditional clouds were not designed for this.

Daytona is increasingly building a cloud for agents.

Daytona pivoted to a narrow wedge: giving developers better environments to run code. As agents became capable of doing more actual work, premise became much larger.

Agents don’t just need tools. They need actual computers, and thousands of them all at once. That is how a devtools product starts looking like infrastructure.

‍