My First Million Wiki

OpenAI

8 min read

OpenAI

The most valuable startup in history began as a nonprofit. That sentence alone tells you something important about how the biggest opportunities hide inside structures that look wrong.

OpenAI was founded in 2015 with a mission to ensure artificial general intelligence benefits all of humanity. Ten years later, it is a company valued north of $150 billion, its product ChatGPT is used by hundreds of millions of people, and the nonprofit wrapper has become the subject of lawsuits and governance debates that will likely be studied in business schools for decades. On My First Million, OpenAI is discussed less as a company to admire and more as a platform to build on — the way people talked about the App Store in 2009 or Facebook’s developer platform in 2012.

The Dharmesh Shah Preview

One of the more remarkable stories in the MFM archive comes from Dharmesh Shah, the co-founder of HubSpot. Two years before ChatGPT launched and changed the world, Dharmesh built essentially the same thing. “I actually have known Sam before he started OpenAI, and I got access to the GPT API,” Dharmesh explained on the podcast. “I built this little chat application that used the API, and so I could have a conversation with it. I had the full transcript two years before ChatGPT came out” (Talking to a Billionaire About AI).

Dharmesh went on to accidentally compete with OpenAI twice — once with his chat application and later with a domain purchase that put him directly in their path. The story matters not because Dharmesh was prescient (though he was) but because it illustrates how obvious the opportunity was to anyone paying close attention. The GPT API was available. The capability was there. What ChatGPT added was not technology but packaging — a simple interface that made the technology accessible to people who would never call an API.

This is a recurring MFM thesis: the money is often not in the invention but in the interface.

The App Store Moment

When OpenAI launched its GPT Store — essentially an app store for custom AI agents — Greg Isenberg came on the podcast and called it a “mini-internet” shift. He and Sam brainstormed five specific businesses that could be built immediately: tax automation, AI life advice, credit repair, health guidance, and more (5 App Ideas for ChatGPT’s New Store).

The framing was deliberate. Isenberg was not talking about OpenAI’s technology. He was talking about the entrepreneurial surface area it created. Every new platform generates a class of entrepreneurs who build on top of it. The iOS App Store created a generation of mobile developers. Shopify created a generation of e-commerce operators. The ChatGPT Store, in Isenberg’s analysis, was creating a generation of AI application builders.

Sam and Shaan’s instinct was characteristically practical. They were not interested in debating whether AI would change the world. They were interested in identifying which specific businesses you could start this week using the tools OpenAI had just released. The show’s relationship with OpenAI is not philosophical. It is transactional in the best sense — what can you build on this, and how quickly can you build it.

The Billionaire Test

There is a telling scene from We Hosted a Slumber Party With Billionaires that captures where AI adoption actually stands. Sam describes a billionaire at a private gathering who was effusive about ChatGPT: “Man, AI’s just going to change the world. I use it every day.” When asked how, “He pulls out his phone and talks to ChatGPT.”

The moment is funny because the use case was trivial. But it is significant because it demonstrates something about ChatGPT’s actual distribution advantage: it is simple enough that a non-technical billionaire uses it daily. Most enterprise software never achieves that level of adoption. ChatGPT did it by being a text box that talks back — arguably the most intuitive interface since the search bar.

Sora and the Slop Question

When OpenAI released Sora, its AI video generation tool, the reaction from the MFM community was instructive. Sam described it plainly: “OpenAI have released a new app called Sora. And when you hear it, the first reaction is a big eye roll. It’s a feed like TikTok, but all the videos are AI. The initial reaction from the smart guy community was like, ‘Oh good, pure AI slop’” (The App That’ll Be Bigger Than TikTok).

The “AI slop” concern represents the single biggest risk to OpenAI’s long-term brand. If the tools become associated primarily with low-quality content — the AI equivalent of spam — the platform loses its appeal to serious builders. The tension between democratizing creation and flooding the internet with mediocre content is unresolved, and MFM treats it as an open question rather than a settled debate.

The Infrastructure Layer

Jason Lemkin, the SaaS veteran, offered a more structural view of OpenAI’s position during his appearance: “OpenAI and Claude can pull from other apps like HubSpot and Notion and just pull the data out” (How the Smartest Founders Are Using AI). This is the infrastructure thesis — OpenAI is not just a chatbot but a platform layer that connects to every other application.

Brian Halligan, Lemkin’s fellow HubSpot alumnus, extended the analysis: “The hardware companies are kind of set, the model companies are kind of set, all that infrastructure layer is set. The app level is really starting to really fly. Perplexity’s the really obvious one, but Harvey in legal, Rogo for investment banking. Like every kind of job now has an app company that’s doing pretty well” (7 Brutal Questions for a $20B Founder).

This is the view MFM ultimately lands on. OpenAI itself is interesting, but the opportunity for MFM’s audience — entrepreneurs with 10 million in revenue — is not in competing with OpenAI. It is in building applications on top of OpenAI’s infrastructure for specific industries and use cases. The AI equivalent of building a Shopify store rather than building Shopify.

The Fastest-Growing Everything

The numbers around ChatGPT’s growth have been discussed on the show with a mixture of awe and practical analysis. ChatGPT reached 100 million users faster than any consumer application in history. It did so without a traditional marketing budget, without a social graph to leverage, and without the kind of viral mechanics that typically drive consumer adoption.

Siqi Chen, discussing the broader AI landscape, noted that “there’s that graph for YC and the fastest growing SaaS companies. I think Cursor is up there. I am fairly convinced that ElevenLabs is actually faster than all of them” (Silicon Valley OG Shares Crazy Stories). The observation matters because it suggests that OpenAI did not just create a product. It created a market velocity — a speed of adoption that is now the benchmark every AI company measures against.

For the MFM audience, this velocity is both exciting and terrifying. It means that AI businesses can scale faster than anything in the history of software. It also means that your competitive window is shorter than it has ever been. Whatever you build on top of ChatGPT, someone else can build a version of it next month.

What OpenAI Means for MFM Entrepreneurs

The show’s practical stance on OpenAI distills to three principles:

  1. Build on the platform, not against it. Dharmesh accidentally competed with OpenAI and learned the hard way. The opportunity is in vertical applications, not horizontal ones.

  2. The interface is the product. ChatGPT succeeded not because GPT-4 was dramatically better than what came before but because the chat interface made the technology usable. Packaging matters more than capability.

  3. Speed beats perfection. The ChatGPT Store, Sora, and every subsequent OpenAI launch create temporary windows where early movers can establish positions. Those windows close fast.

Sources

See Also

Graph View

Backlinks