How Bolt.new hit $40M ARR in 5 months
CEO Eric Simons on the unlikely journey from near-death to breakout hit
👋 Hi, it’s Kyle Poyar and welcome to Growth Unhinged, my weekly newsletter exploring the hidden playbooks behind the fastest-growing startups. A special welcome to the 728 new readers who subscribed since last week’s 🔥 post on AI agents for marketing.
The story of Bolt.new, the AI app builder, feels almost too good to be true.
Its maker, a company called StackBlitz, had spent the previous seven years struggling to stay alive. They had stalled out at $0.7M ARR and were weeks away from shutting down. Then they launched Bolt.new on October 3rd, 2024 with a single tweet. It took off instantly.
Bolt.new essentially became a second ChatGPT moment for its users. The product reached $4M ARR in only four weeks. It crossed $20M in three months and then $40M ARR in five months. Bolt.new now claims five million users and is still adding more than a million users each month. Oh, and the team remains hyper-lean; there are fewer than 40 employees including less than 10 in GTM roles.
I sat down with founder and CEO Eric Simons along with chief of staff Alex Berger on the company’s unlikely journey from near-death to breakout vibecoding hit.
Related resources: You should play with vibecoding for GTM, The surprising patterns behind viral AI products, The state of B2B monetization in 2025, More Zero to One growth case studies
Bolt.new was in the right place to ride the wave of improving LLMs.
You could read this story as an invitation to just keep building and eventually growth will materialize. Don’t.
StackBlitz found itself with the right technology at a time when the potential of the technology could finally be unlocked. The company started in 2017 with a vision to make it as easy to build full stack web applications as it was to use Figma. The original form factor was an online development environment (IDE) that operates within a web browser. While this proved popular with developers, few were willing to pay for the technology.
Eric had built a prototype for what’s now Bolt.new in February 2024. He didn’t think it would work because the AI models weren’t good enough. Then he got access to Anthropic’s Claude 3.5 Sonnet model in June; the agentic coding skills were leaps-and-bounds better. By July 1st Eric hard-pivoted back to the February prototype. Bolt.new was ready three months later.
“We had exactly the right technology to unlock the opportunity,” Eric told me. “Suddenly we arrived at a moment when a data center can magically stream code in a browser, and we had the tech to run it.”
They’ve kept the UX deceptively simple despite building a product that’s incredibly powerful.
Bolt.new famously launched with a single tweet. That tweet didn’t go viral by accident. Eric and team took the time to prepare a great first tweet with a slick 43 second demo video. They also created subsequent ones to lay out exactly what to do with the product including walking through tons of specific examples.
When people went to Bolt.new’s website, it was one of the simplest interfaces out there. There was no marketing jargon. No credit card requirement. No need to even log-in. “It was a prompt box and that’s it,” Eric recalled. “People could write what they want, hit enter and voila.”
The barrier to having a wow-moment was essentially zero. That meant people could easily build cool products, then they could share exactly what they built. “The word of mouth was built in because you want to share what you made. That’s the core growth loop.”
In my research I’ve found ungated product experiences like Bolt.new’s can 3x the share of website visitors who start using a product. While this additional usage means more AI and support costs, the improvements to activation and conversion often more than make up for it. (At Rows, profiled last year, moving to an ungated product experience 4x-ed the number of new activated users per week.)
The Bolt.new team hasn’t played around with growth hacks or much product experimentation. But I personally believe Bolt.new’s subtle UX choices have strongly contributed to its breakout growth.
Many AI products feel like a black box. Bolt.new breaks down the AI process step-by-step and in real time, which builds trust. From there the app makes it easy for users to refine prompts; there’s even a one click “enhance prompt” button. And Bolt.new has a preview mode, allowing users to test the AI-generated outputs before committing.
Product virality remains the company’s best growth loop. Shipping velocity and referrals accelerate virality.
Bolt.new’s core growth model is simple:
Attract builders, particularly early adopters
Educate and inspire folks to build their own apps
People build apps and Bolt.new makes it easy to create for free
Builders share what they’ve made, attracting more builders
Bolt.new’s growth strategy comes down to finding ways to make the flywheel spin faster and faster.
Their social media includes slick demo videos and curates community use cases. Eric also finds ways to demo the product in front of other folks’ audiences. A tutorial video with Greg Isenberg in November was particularly explosive in the early days of Bolt.new.
On the product side, Bolt.new treats their product velocity as part of the growth model. “We’re known for rapidly shipping cool capabilities and even fixing bugs quickly,” Eric underscored. “People get really excited about that and it reignites word-of-mouth. The faster we’re moving, the faster growth tends to be.”
The team invests pretty heavily in customer marketing including initiatives like a referral program (users refer folks and get free tokens), mini competitions on Twitter and most recently what Bolt.new calls The World’s Largest Hackathon. The hackathon brought 100,000 signups and Bolt.new is submitting it to the Guinness Book of World Records. (The marketing team even used Bolt.new to build and launch a full-stack site for people to register and get updates on the hackathon.)
As a side note, Bolt.new lets you start an AI video call to brainstorm ideas with Eric (this is powered by an app called Tavus). I hadn’t come across that before, but it makes total sense given the company’s growth model and community-focus. (Adam Robinson of RB2B has a similar AI clone for users.)
Bolt.new innovated on pricing, being an early adopter of AI credit models that have now been copied far and wide.
When Bolt.new launched, they only had a $9 per month plan. In fact, StackBlitz’s pricing was $9 per month going back to 2018. “The thought was that if people burn through the $9, we’ll deal with it later,” Eric told me.
Well, later came fast. “People started burning through all their credits within the first day.”
Eric decided to work through the weekend to figure out new pricing. He landed on adding a few plans for people to upgrade to, charging more if people needed more inferences. (Bolt.new’s starting price has since increased to $20 per month when billed monthly.)
This AI credit pricing seems pretty commonplace now. It wasn’t at the time. Eric reminded me that before this AI products were typically around $20 per month for ‘all you can eat’ access, although if folks used the product too much, they might get slowed down.
“We flipped the script on pricing. The more people invested in the tool, the more they could build. That led to people building real products. It turned out this is how everyone is re-orienting their pricing, too.” - Eric Simons, co-founder & CEO of Bolt.new
The first plans that Bolt.new rolled out only went up to $200 per month. They’ve since added Pro plans that scale up to $1,000 per month as well as Teams and Enterprise plans.
A new change that Bolt.new put in place on July 1, 2025 is credit rollovers: any unused tokens roll over to the next month. Bolt.new is the first-mover here, too (although Lovable has copied it); everyone else had been use it or lose it. The Bolt.new team is betting that this will be a retention mechanism (“you’ve paid for this, you can cancel but you’ll lose all these credits”) and help with the end user experience.
As a side note, Bolt.new dog-foods the product to make working prototypes of new pricing pages. Instead of writing up suggestions in a Google Doc, they feed the existing pricing page into Bolt.new, prompt what they want to adjust and then hand the prototype over to product and design.
Their velocity is made possible with a small team and maximizing context per head.
When Bolt.new launched, the team had fewer than 20 people. They’re still under 40, meaning that Bolt.new’s ARR per employee is well north of $1M.
Of the 40 people, about half sit in engineering. Another five are in marketing. The remainder are fairly evenly spread across sales, people operations, design, product management and other functions.
Eric said that having such a small team was a huge benefit in the early days. “You have to move so fast. There’s no way we could’ve kept up with demand with a larger team because we wouldn’t have been able to respond immediately to it.”
I asked if there was any technology or AI tools that made it possible for Bolt.new to be so lean. Only one really moved the needle. The biggest help came from an app called Parahelp, which does sophisticated customer support resolution using AI. Before Parahelp, Alex and Eric answered all support tickets for the first 30,000 active subscriptions. Parahelp initially resolved about 50% of tickets; now it’s up to 90% including more complex tickets like issuing refunds and helping people debug.
What Eric thinks matters most is maximizing the context-per-head at Bolt.new. This allows everyone to make independent decisions quickly. Bolt.new was in a fortunate position of having a core team that had worked together for 4+ years.
Alex, Eric’s chief of staff, joined the company in Fall 2020 as employee #7. Everyone from his cohort is still at the company, he said. Alex attributes that loyalty to hiring folks who are real, pragmatic, caring and direct.
Bolt.new doesn’t look at pedigree in hiring – in fact, many folks on the team didn’t graduate from college. Instead they gravitate toward a contract-to-hire model, which lets both sides try-before-you-buy. About 10 of the 40 team members at Bolt.new started as contract-to-hire before joining full-time.
The TL;DR
Bolt.new was weeks away from shutting down. Then a pivot catapulted the company to $40M ARR in only five months. My takeaways on Bolt.new’s unexpected growth journey:
Bolt.new was in the right place to ride the wave of improving LLMs.
They’ve kept the UX deceptively simple despite building a product that’s incredibly powerful.
Product virality remains the company’s best growth loop. Shipping velocity and referrals accelerate virality.
Bolt.new innovated on pricing, being an early adopter of AI credit models that have now been copied far and wide.
Their velocity is made possible with a small team and maximizing context per head.
Brilliant write up and the resilience of the team to work through and hit a hockey stick is inspiring. Thank you for sharing. Hope to join their ranks
Great read! Focus on removing frictions while your competitors add more gates. It's like saying no to 1000s of things so you can focus on what matters like Steve Jobs once said. Anyone else guilty of over engineering their signup flow??