Tony Robbins once said, “The secret to living is giving. The more you give, the more you receive.”
Adhik Joshi became a founder because he was obsessed with making products that made peoples’ lives easier and more enjoyable. He started with video game modifications as a child, worked his way up to travel booking applications, and today he helms a wildly successful, full-service AI business out of India.
But prior to his most recent success, Adhik failed almost every project he started. While he had many great ideas, he could never attract interested customers. After years of defeat, he retreated to his bedroom for a year to read and figure out where he went wrong.
Everything changed when Adhik started following an AI company called Stability AI on Twitter (X.com).
After Stability launched its AI art product, Stable Diffusion, for public use, Adhik’s friends asked him if he knew where they could find an API. Being helpful, Adhik built one and bought the domain name stablediffusionapi.com for $4, launching Stable Diffusion API.
In the next month, news outlets started reporting on AI imaging, and searches for tools like Adhik’s went wild. Over the course of two years, his product earned 350,000 signups, 30,000 paying customers, and $1.2 million in ARR.
Despite his success, Adhik has yet to cut himself a single paycheck, instead reinvesting all of the money into a rebrand, employees, and an entire data center with 200-plus GPUs.
Now, he’s one of India’s exciting new entrepreneurs quietly defining the AI industry from his home in Bengaluru. Here’s how he did it and what you can learn from his story about being a better entrepreneur.
Modding Cybercafe Games and Crafting University Apps
While Adhik’s first entrepreneurial success was relatively small scale, it showed him that passion can turn into profits.
From a young age, Adhik loved building things for the enjoyment of others. At 14, he’d modify video games like GTA Vice City with special missions and play them with friends. His games were so popular that he sold copies to an uncle who ran a small group of cyber cafes in 2009. Through his uncle’s network, Adhik’s games started showing up in local cybercafes.
With that early success in mind, Adhik studied software engineering in 2015. In his first years, he tried to recreate that high of building something his friends loved by building an application similar to Google Maps that helped people find local businesses.
At that time, Google Maps rarely listed mom-and-pop shops on the streets of India. With Adhik’s app, Shoppit, people manually listed these small shops and added a phone number so customers could call and order things.
Most people Adhik talked to about the application were interested in it, but it ultimately failed because Adhik had no idea how to get the word out about it.
“I learned you can’t build a venture without having a proper go-to-market strategy,” he says.
Adhik’s next project was in 2017. He built a free tool that teachers at his university could use to schedule classes and record lectures (similar to the popular US application, Canvas). After spending hours on the project, he placed recording technology into a few classrooms only to find that some people at the university were hostile towards this new classroom aide.
“Many in administration didn’t want to relinquish their control to some technology, and teachers didn’t like having their lectures recorded,” he said. “People are afraid of losing what they do. They are often lazy and need incentives to adopt new technology.”
Undeterred, Adhik next built an online travel agency letting users enter a destination to receive an entire travel plan including flights, hotels, and activities. This app failed because the customer acquisition cost was too high. Competing with giants like Expedia and Booking.com meant that he had to pay over $6 dollars per click to stay at the top of Google Ads.
But Adhik’s failures were his mentors. Because the cost of acquisition was so high for his travel app, he learned how to do SEO. Because he had so much trouble getting customers on board his first two apps, he learned go-to-market (GTM) strategies. All of these would come together for his first truly successful project during Covid lockdowns.
A Year of Pondering and Reading
During the lockdowns of 2020, Adhik shut down and canceled subscriptions for all of his old projects in frustration and decided to focus on self-improvement. He needed to understand exactly what was behind his failures.
He read hundreds of well-known books like Hobbes’s The Leviathan, Zero to One by Peter Thiel, and The Art of Seduction by Robert Green. In his reading process, Adhik aimed for efficiency, only reading sections that he deemed important rather than the entire book.
“I realized focusing too much on finishing is more virtue-signaling than useful,” he says.
Despite his philosophy, some books took much longer for Adhik. For example, Peter Thiel’s Zero to One.
“It took me three months to finish it because every time I read the parts on the importance of salespeople, it hit me really hard that I should have done more related to sales,” he says.
Finally, after a year of pondering what he did wrong, Adhik knew that his next project would be the one that finally brought him fame and fortune. Plus, he’d begun following a social media account that would change his life forever.
An Interesting Little Project on Twitter Now Worth Over $1 Billion
In 2020, a company called Stability AI began teasing a yet unreleased AI image generation project they’d created called Stable Diffusion AI.
At the time, there was little fanfare around AI art as the images were interesting but blurry and chaotic. Similar to the likes of a former Bootstrapper founder’s Nightcafe Studio circa 2022. But Adhik was captivated by AI art’s possibilities and eagerly awaited the full release.
On August 22, 2022, Stability AI released their flagship product, Stable Diffusion. It was a detailed image generator that could create art based on text descriptions. Unlike Midjourney and DALL-E, its codebase was publicly available to use on personal servers rather than on the cloud. It could also run on devices with modest GPUs.
One post on the Stability AI Twitter page leading up to the release of Stable Diffusion.
Adhik immediately wanted to get involved with Stable Diffusion. Friends had already started asking him how they could make it run and if it was possible to get an API for the product. He saw another opportunity to make a helpful and potentially profitable project. AI art at that moment seemed to have all of the traits his previous ventures lacked: general public interest, lack of competition, and eager leads.
Adhik bought the domain stablediffusionapi.com for about four dollars. It was cheap because the name was long and about three people had searched for it in a month.
“A lot of people used to say domain names don’t matter, but they do for SEO,” says Adhik. “In my case, the name itself said it all.”
After spending three days learning Python, Adhik built an MVP for his Stable Diffusion API. He describes it as very scrappy. “There were a lot of spelling mistakes. For example, I wrote massage instead of message.”
Even with a poorly made website, once mainstream news media began reporting on the new AI image-generating trends, everything changed for Adhik. In September of 2022, one million people searched for the keyword “Stable Diffusion API”. This time around, Adhik wasn’t hopelessly drifting off the back of illusory tech waves, he was surfing on one of the largest tech tsunamis of the decade.
“People were coming onto the site, asking questions, they were paying, the energy was different,” he says. “I was spending day and night on it. Once you hit that energy level you want to drill down on it.”
In October 2022, Stability AI secured a $100 million-plus investment raising its valuation to $1 billion. On the back of Stable Diffusion’s success, Adhik crossed $700,000 in sales. In December of 2022, he hit one million dollars in ARR. Almost overnight, he was officially a successful bootstrapped founder.
How Adhik Drives a Company Growing at Breakneck Speed
Adhik has been on the gas pedal of his company since those fateful first months of 2022. He says he works twenty-four hours a day, seven days a week, and doesn’t take weekends. He loves (almost) every moment of it.
“I love solving problems and fixing bugs,” he says. “When I’m really tired, I sleep and go out with friends.”
However, he also spends a lot of time worrying about maintaining an AI product in the ever-changing AI landscape.
“Every single day, I feel that my product is going to fail, so I do stuff about it. For example, Stable Diffusion used to be the biggest AI image generator and now there’s Flux. You can see Google is now competing heavily with OpenAI. I don’t think you can sit back and relax ever in this industry.”
Over the course of the previous year, he’s also grown the Stable Diffusion API team to 35 people entirely sourced from LinkedIn and Twitter.
Adhik is also unique in that he does not draw a salary from Stablediffusionapi. Instead, he lives off income from side projects and reinvests all of the revenue into employee salaries and growth.
“If you’re in India, you don’t need to spend an absurd amount on rent so your burn rate is low,” he says. “If I wanted to sell on Acquire.com, I could make a lot of money, but for now, my family is doing well and I’m doing well.”
In 2023, Adhik was doing so well that he rebranded his business into ModelsLab, an end-to-end AI solution. Now, it handles language models, scaling models, video generation, and more. In fact, he’s now building a solar-powered data center that currently has over 200 GPUs and counting.
“GPUs are costly and talent in that space is scarce,” he says. “However, the cost of our GPUs goes down to zero when we own them and resell our servers to other businesses. We’ll then be able to open source our stuff at a much lower cost.”
Despite the pivot, Adhik has no intention of becoming the next big shark in AI computing. He wants to aid other founders in the AI space so they don’t have to go through the same hardships that he did.
“I have been through ten years of trial and error, and if I can help someone cut those years into six to ten months, it’s beneficial for everyone,” he says. “I’ve earned a lot so I want to pay it forward.”
As you might guess, venture capital groups regularly call Adhik to ask to invest in the project. So far, he’s rejected all offers.
“People have their own things they want to execute when they join a project,” he adds. “They say they’ll invest only if I don’t create a certain product that competes with something they’ve invested in. It’s better to keep your equity and build what you want.”
What Doesn’t Work, And Adhik’s Advice to Other Founders
While Stablediffusionapi saw tremendous growth from search engines, Adhik says he’s been unable to drive value from online ads. Part of it might be because he’s already done so well in SEO.
“I spent fifty thousand dollars and it didn’t work. We switched off ads to see if anything changed and the signups stayed the same because we had tons of backlinks and a great domain.”
Reflecting on where he came from, Adhik says one of the biggest myths about entrepreneurship is that there is only one shot at success. He thinks even if he’d missed the AI wave, another swell would have carried him eventually.
“Entrepreneurs like Bill Gates and Steve Jobs are the one percent of successful people,” he says. “For most people, it takes multiple trials and errors to reach a certain level.”
