Anurag Goel and
the trouble with
getting things online

Anurag Goel was trying to find something worth working on for years. He built applications, tried ideas, and put them online to see how they behaved outside his laptop. Again and again, getting them online became a project of its own. Eventually, the obstacle became more interesting than the experiments. Render grew out of that recurring irritation: a developer ought to be able to build a useful thing without acquiring a second occupation in cloud administration.

There is a pleasing inconvenience in this origin story. The infrastructure industry sells access to computing power; Goel kept finding that using it required a remarkable amount of additional work. For someone searching for a large problem, this was both an annoyance and an invitation. He had already spent nearly five years at Stripe. He could afford to choose carefully. That made the choice harder to hurry.

A customer with suggestions

Goel grew up in New Delhi. He remembers his first computer class being around sixth grade, though programming became a more substantial interest in college. He studied computer science at IIT Delhi and moved to the United States for a startup job. The startup was acquired, and he found himself working on enterprise software. Side projects supplied the entertainment his day job was failing to provide.

One of those projects needed to accept payments. In 2010, he tried the small service then known as /dev/payments. It would become Stripe, though its early customers were still dealing with a much younger product. Goel had suggestions. He sent them. A customer willing to explain what is awkward can be quite a useful nuisance.

Those conversations brought him into contact with Patrick Collison. When Collison asked whether he wanted to join the company, Goel initially declined. Later, with his wife preparing to attend graduate school on the East Coast, the prospect of continuing to work alone from home looked less appealing. He already liked the Stripe team. In July 2011, he joined as employee eight, ahead of the company’s public launch that September.

His work there moved across engineering, growth, and financial risk. By the end, he was running the risk team. It was an education in the many jobs a small company needs done, and in the attention developers give to the details of a product. Stripe’s customers had to trust what happened after they made an API call. An elegant introduction only got you so far.

Choosing something worth ten years

Goel left Stripe in February 2016. He wanted a problem that mattered to him enough to survive the novelty of starting. Buying a domain and feeling briefly delighted by an idea would be insufficient preparation for the years that followed. His method was practical: build an application in a field, run it, and see whether the interest held.

He even briefly considered politics, then decided it was a poor fit for his personality. The alternative involved rather more software. At South Park Commons, an early community of people exploring what to do next, he could test projects around other builders and hear where their experiments got stuck.

Crestle was one result. While taking a fast.ai deep-learning course, Goel noticed people struggling to run Jupyter notebooks on AWS. He built a service that supplied a GPU-backed notebook with a click. People used it, and he considered making a larger company of it. But he could not see himself wanting to spend the next ten years on that particular problem. Crestle was acquired in 2018.

The work still paid an intellectual dividend. Building it helped him understand Kubernetes and Docker, tools for managing and packaging applications. He had previously looked at the broader deployment problem and thought it too large to tackle. Now he could see a possible way through.

“I wanted to solve a big, hairy problem that I could also be excited about for decades.”

Anurag Goel · The Kubelist Podcast, June 2026

His earliest Render work, begun in July 2017, offered a Python coding environment in the browser. That approach made quick experiments convenient, but restricted the applications people could bring. He changed direction, developing a platform that could run existing Python, Node, and Ruby applications and deploy updates from GitHub. The customer’s code had a vote in the design.

The people already in the room

South Park Commons gave this idea something more useful than an audience: people who needed it. Its website became the first product hosted on Render. Members’ projects, and later their companies, supplied early customers. Ruchi Sanghvi helped Goel start the fundraising process; General Catalyst’s Steve Herrod led the seed round.

The appeal was easy to demonstrate to a room of builders. They had applications they wanted other people to use. They also had finite hours. Render could take over work that otherwise demanded specialist attention before the application reached its audience. The community was close enough to make complaints useful and concrete.

Goel raised money after he had a working version. The early team comprised a designer and three engineers, including him. They spent well over a year building before the Disrupt appearance that brought wider attention. This was a small group accepting a considerable maintenance obligation: customers would expect their applications to keep running long after the first successful demonstration.

Invite-only access began in 2018. Public signups opened in April 2019. That sequence matters. Render’s beginnings were a succession of prototypes, users, and decisions, spread across several years. An origin date makes a tidy company biography; the product itself needed several attempts.

After the confetti

In October 2019, Render won TechCrunch Disrupt’s Startup Battlefield. Goel later recalled the giant check, the confetti, and the sensation of momentarily inhabiting HBO’s Silicon Valley. The television comparison is understandable. A large ceremonial check is an unusual accessory for people whose ordinary work involves making applications stay up.

The win brought visibility. Goel has described the following years as a period of building reliability rather than pursuing growth at any cost. The distinction has practical consequences for a cloud company. A customer can admire your launch and still need reassurance that their business will be working tomorrow. Confetti offers limited assistance with that question.

An investor conversation in February 2020 made the economics plain. Lee Fixel asked what a typical Render customer paid, then what a typical DevOps engineer earned. The comparison put a price on the engineering work around the hosting bill. Goel’s pitch was about giving teams time back as well as providing computing resources.

By November 2021, Render had raised a $20 million Series A. In June 2023, Goel announced a $50 million Series B and reported more than half a million developers and teams using the platform. These were steps in a business whose central promise remained fairly consistent: let the people building applications spend more of their working lives on those applications.

January 2025 brought an $80 million Series C. In February 2026, Render announced another $100 million in a Series C extension, at a $1.5 billion company valuation. It reported more than 4.5 million developers. The growing use of AI coding tools supplied a new setting for an old concern: creating software was getting faster, while running it still demanded care.

The engineer who had to step back

Inside the company, Goel faced another version of the same question about time. He has written about his habit of diving into problems and doing the work himself. Engineering skill had helped him build Render. Continuing to exercise it at every opportunity could leave the organization dependent on him.

His account of becoming a CEO centers on hiring capable people and making it possible for them to thrive. That meant resisting the urge to write the feature or specification personally, even when he believed he could finish faster. The immediate saving could prevent the team from developing the ability to work independently. A founder can become a queue without noticing the line forming.

The restraint extended to customers. Render initially concentrated on early startups that urgently needed to deploy production applications. Goel has described the danger of adding segments and features before the original group has a product it would seriously miss. Each expansion makes a claim on the same limited engineering attention.

His discussions of product design return to what users are trying to accomplish. More capability need not mean more decisions on the first screen. Complex features can appear when they become useful. Support questions and confusing error messages also belong to the product experience. Every repeated explanation suggests an opportunity to make the next person’s attempt less difficult.

A conference, at last

On June 18, 2026, Goel opened Render’s first user conference, localhost, in San Francisco. The event arrived eight years into the company’s life. He explained that Render had only begun building a marketing team about a year earlier. Asking a hundred engineers to organize a conference had apparently failed to become an urgent product requirement.

The work under discussion included AI applications and agents. These can run for long periods and spend unpredictable stretches waiting for responses. Goel had been describing the problem in interviews: assumptions that suit a quick request and reply become awkward when software waits on a language model. Easier code creation leaves a considerable amount of work for the people operating the result.

Asked, in effect, why an early Stripe employee keeps choosing this schedule over a beach, Goel has offered a lightly unenthusiastic assessment of beaches and mai tais. The fuller answer is that making the cloud easier to use gives him a sense of purpose. Improvements to Render can help other people bring their own projects into the world.

That ambition returns him to the side projects. A person has an idea, writes some code, and wants someone else to try it. Goel has spent years examining everything that happens between those steps. He seems content to leave the holiday itinerary unfinished while he works on shortening that distance.