Four years after its inception, AI app hosting platform Runpod has reached an impressive $120 million annual revenue run rate, according to co-founders Zhen Lu and Pardeep Singh in an interview with TechCrunch.
Their entrepreneurial path serves as a remarkable illustration of how strong development and opportune timing can attract significant success.
This narrative highlights their ascent, from bootstrapping to over $1 million in revenue, to securing a $20 million seed investment after Dell Technologies Capital partner Radhika Malik discovered them via Reddit. Furthermore, they gained a crucial angel investor in Hugging Face co-founder Julien Chaumond, who, after using their product, contacted them through support chat, as reported by the founders to TechCrunch.
The genesis of their venture dates back to late 2021, when the two friends, colleagues as corporate developers at Comcast, grew tired of their existing hobby.
They had previously assembled specialized computer rigs in their New Jersey basements for Ethereum mining. Although they managed to mine some cryptocurrency, it proved insufficient to recoup their initial investment. Moreover, the impending network upgrade known as “The Merge” signaled the end of mining as they knew it.
Adding to this, Lu commented that the activity became “boring” after only a few months.
Having convinced their wives to allow a combined investment of roughly $50,000 into this hobby, Lu and Singh recognized that preserving domestic tranquility necessitated finding an alternative use for their graphics processing units (GPUs).
Given their involvement in machine learning projects at their jobs, the developers decided to transform their mining setups into AI servers. This pivot occurred prior to the emergence of both ChatGPT and DALL-E 2.
During the process of reconfiguring their rigs, Lu observed, “We realized how truly terrible the software stack was for managing these GPUs.” As experienced developers, they identified a significant problem ripe for a solution.
Runpod emerged, as Lu explained, “because we felt that the actual experience of developing software on top of GPUs was just hot garbage.”
By early 2022, after a few months, they were prepared to launch their creation. Runpod functions as a platform designed for hosting AI applications, prioritizing speed, adaptable hardware (featuring a serverless option for automated setup), and developer tools such as APIs, command-line interfaces, and various integrations.
In 2021, their platform offered only limited integrations, such as support for Jupyter notebooks. Their subsequent challenge was attracting beta testers.
Lu reminisced, “As first-time founders, we lacked knowledge in marketing or general business development. So I thought, ‘Alright, let’s just post on Reddit.’”
Consequently, they shared their offer across several AI-focused subreddits. The proposition was straightforward: complimentary access to their AI servers in return for valuable feedback. This strategy proved successful, attracting beta users who subsequently converted into paying clients. According to the founders, within nine months, they resigned from their jobs and achieved $1 million in revenue.
Scaling Through Self-Funding
However, this success introduced a new challenge. Lu recounted, “After six months, our business users expressed concerns, stating, ‘We want to operate genuine business applications on your platform, but we can’t do that with servers located in people’s basements.’”
The New Jersey-based founders initially hadn’t considered seeking venture capital. Instead, they opted for revenue-share agreements with data centers to expand their capacity. This approach, however, proved demanding, requiring them to constantly anticipate future needs.
Singh emphasized, “A lack of available GPUs shifts both market and user sentiment. If customers don’t perceive sufficient capacity, they will seek alternatives.”
Concurrently, their user community expanded significantly on Reddit and Discord, experiencing a particular surge following the introduction of ChatGPT.
Venture capitalists were actively scouting for promising investments. Malik discovered Runpod on Reddit and initiated contact, marking their inaugural call with a VC. Lu, however, was unfamiliar with the process of pitching to investors. He noted, “Radhika was exceptionally helpful, right from our initial discussion,” as she elucidated the VC mindset and assured him she would remain in contact.
During this period, Lu was focused on operating a self-sustaining business. He stated, “For nearly two years, we operated without any external funding.” Consequently, Runpod never provided a free service tier, requiring the platform to be at least self-sufficient, even if profits were modest. Unlike other AI cloud services that originated from crypto mining, these founders consciously avoided incurring debt.
By May 2024, amidst widespread enthusiasm for AI applications, their prescient choice to introduce AI hosting for developers two years prior began yielding substantial returns. Their platform had expanded to serve 100,000 developers, culminating in securing a $20 million seed investment, jointly led by the venture capital divisions of Dell and Intel, and featuring contributions from notable figures such as Nat Friedman and Chaumond.
While they have not pursued additional funding since then, they are now preparing to do so, confident that their robust business model warrants a substantial Series A round.
Presently, Runpod serves a customer base of 500,000 developers, encompassing everyone from individual users to Fortune 500 enterprise teams with multi-million dollar yearly expenditures, as stated by the founders.
The company’s cloud infrastructure extends across 31 global regions, catering to prominent clients including Replit, Cursor, OpenAI, Perplexity, Wix, and Zillow.
The competitive landscape is intense, with developers having access to all major cloud providers (AWS, Microsoft, Google), alongside numerous specialized alternatives such as CoreWeave and Core Scientific.
However, Runpod views its market position distinctively – as a platform primarily focused on developers. They anticipate that coding will not disappear but rather evolve, transforming programmers into creators and operators of AI agents.
Lu articulated their ambition: “Our objective is to become the foundational platform for the next generation of software developers.”