"We went from $0.7 million in ARR to $20.7 million in like two months."

In a recent unfiltered conversation on the “Around the Prompt” podcast, Eric Simons, CEO of Bolt.new, delved into the dramatic story behind his company’s meteoric rise. What appears to be an overnight AI sensation is, in fact, what Simons calls an “overnight success seven years in the making”—a journey that took the company from the brink of shutdown to a $20 million ARR powerhouse in a matter of months.

From Near-Shutdown to Explosive Growth

Simons began by recounting the long and arduous journey of Stackblitz, the parent company of Bolt.new. The original mission, he explained, was to “make it as easy for people to build like full stack web applications as it is to use Canva for design.” However, despite building powerful underlying technology, the company struggled for years to find product-market fit.

The core challenge was persuading developers to abandon their local machines. “To use it for like meaningful work…getting people out of their local environments, just like every other cloud IDE has found out, is a huge challenge,” Simons admitted.

The situation became dire. By late last year, after years of fighting for every dollar of revenue, the company was preparing for the worst. “At the end of last year, we were getting ready to like spin down the company,” he revealed. At that point, after nearly five years of monetization efforts, their annual recurring revenue stood at a hard-fought but unsustainable “$600-700k or something.”

The turning point came with a pivot to generative AI. “Everything changed when we had the idea, hey, what if we kind of took this Frontier AI stuff and made a…text-to-app…sort of experience basically. And that’s what Bolt.new is.” The launch in October was transformative. In a stunning reversal of fortune, Simons stated, “We went from, you know, $0.7 million in ARR to, you know, $20.7 million in like two months.” With a touch of understatement, he added, “We felt it was probably good to keep the thing going at that point.”

The Secret Sauce: A Browser-Based Operating System

When asked if the previous seven years of struggle were necessary, Simons was unequivocal. The core technology developed during that time, a WebAssembly-based operating system called WebContainer, is precisely what makes Bolt’s seamless experience possible.

“We can…write an operating system from scratch in WebAssembly that can run inside of your browser tab and you know boot environments in like 100 milliseconds,” he explained. This architecture bypasses the costly and slow model used by competitors. “If you’re going to do cloud-based ID stuff, you have to spin up a VM for every single user, for every single project…you’re dealing with latency because every keystroke has to be synced to the server, synced back.”

This technical advantage has massive business implications. Because the compute happens on the user’s machine, the cost to the company is negligible. “It doesn’t cost us anything tangibly…what that means is like we can give this away like candy,” Simons said. This allows for a frictionless, no-credit-card-required user experience that provides instant value.

The New Creator Class: Democratizing Software

Perhaps the most surprising discovery for the Bolt.new team was who was using the product. Initially built as a developer tool, they found a completely new audience flocking to the platform.

“The majority of the people coming to the thing are not developers,” Simons stated, estimating they constitute “somewhere between like 60-70% of our audience. They are PMs, they are designers, they’re entrepreneurs.”

The team was initially confused by support tickets from non-technical users. “They’re asking a lot of ‘What is this error code?’ And we’re like, ‘Aren’t you a developer?’” This confusion quickly turned into a profound realization. “This is actually enabling an entirely new category of people to build real software…Previously, the only way for them to actually get those ideas into actual coded software is through the fingertips of a developer. And now with Bolt, they’re putting their own fingertips into a keyboard.”

For Simons, this shift represents a fundamental reorganization of the industry. He asserted, “We’re kind of at the ground zero of like how the world builds software is about to get completely reorganized.”

Product Vision: From Power Tool to Full-Stack Company Creation

Looking ahead, Simons sees Bolt.new evolving into a universally accessible “power tool” for both amateurs and professionals. He uses an analogy: “It’s like we’re building a power drill…my mom used one to like put up a baby gate…You go to a construction site and like they’ve got the same DeWalt power drill where they’re like, these are professionals using this.”

The company is already moving beyond just code generation to solve the next set of user problems. Recognizing that non-technical users often get stuck, they are launching a program called “Bolt Builders.” Simons described it as “the Genius Bar…but for Bolt,” a marketplace where users can connect with experts live to get unblocked on technical issues, marketing, or distribution.

He believes this move highlights a key truth in the age of AI: “Humans are kind of just more important than ever. We’re now just at a point where folks can like specialize on punching in on problems that are just actually intellectually difficult, right? That LLMs are not great at.”

The Lean Machine

Bolt.new’s explosive growth was managed by an incredibly small team. The company reached $20 million ARR with just 15 employees, a testament to a philosophy of lean operations that was forged in the early days of bootstrapping. Simons contrasted this with the venture capital “mania of 2020-2021.”

“It was very popular to go and like hire, you know, 50 people in a quarter or whatever. And we didn’t do that,” he said. “You want fewer people with more context per head largely because people have more agency to like go and do things.” This agility, he argued, was absolutely key to navigating the scaling challenges that came with their sudden success.

When asked about the future, Simons expressed hope for code generation models that are “two or three times as good,” but jokingly added, “I hope we don’t see something that’s like a thousand-X, though. I think that would…uh…that would be…maybe scare.”