From the beginning, the goal of Generative Value has been to study the world’s most important companies and industries.
It doesn’t take much imagination to put OpenAI in that category.
They kicked off this whole shebang, they’re the highest-valued AI startup in the world, and they’ve got their hands in many cookie jars (Stargate, semiconductors, robotics, data centers, foundation models, agents).
If there’s one company to study to understand AI, it’s OpenAI.
It’s why this will be the most detailed deep dive in the history of GV and why it will be a multi-part series. Because this company needs it.
This will be the first in a multi-part series on OpenAI. This part will cover the context and history of OpenAI; the next parts will cover technology, business models, and my thoughts on the company's future.
To make this as evergreen as possible, I’m adding a new section to company deep dives on the lessons people should take with them. Some worldly wisdom if you will (As Charlie says, “Our job is to find a few intelligent things to do, not to keep up with every damn thing in the world.”)
Worldly Wisdom
1. “Right place, right time” -> “Right place, stay alive long enough to make it the right time.”
It took seven years from OpenAI’s founding to the ChatGPT moment, and even then, it took them by surprise.
The same thing happened with Nvidia building out CUDA for 15 years before AI really took off.
This story plays out over and over again (‘technology realization times’ seem nearly impossible to predict - the cloud took 30+ years, the idea of AI has been around nearly as long as computers, and robotics haven’t seen their “ChatGPT moment” yet).
The term “right place, right time” gets thrown around a lot. It’s more accurate to say resilient people stick around long enough for it to be their time.
2. Special People create Special Companies
They key term above is “resilient people.” The details of how OpenAI would succeed have never been clear.
What you could see in 2015, when OpenAI was founded, was that some of the world’s best innovators were hovering around this organization. Ilya, Greg, Sam, Elon, and some of the world’s great investors all were tied to this company.
3. Don’t underestimate the Power Law
Most thought the $20B valuation was expensive. Only 13 private unicorns (on CBInsights’ list) are worth $20B, after all. Well, OpenAI was valued at $157B recently and looks to be valued even higher shortly.
The key point is the idea of positive feedback: advantages compound. Once the compounding starts, it’s hard to stop it.
4. It’s Hard to Compete Against Free
If you’ve read my posts on the data industry, you’ve seen me talk about the catch-22 of open source. It’s wonderful for the industry: more control, cheaper services, more innovation. But it’s brutal for business models: both as a competitor and for companies trying to monetize it.
We’re seeing the same game play out with foundation models.
For all the success OpenAI has seen, the biggest question is how they create a sustainable business model, and open-source makes that a lot harder.
All is not lost, but it does mean OpenAI has to find other ways to build moats around leading-edge models, perhaps through direct sales relationships or only offering top models on ChatGPT.
5. If you zoom out far enough, only the big ideas matter.
I first heard this quote in the context of investing from Nick Sleep, “We keep our discussions to as high a level as we can manage in the belief that, in the long run, the high level is all that matters.”
OpenAI was a non-profit without a clear idea of what to build, that lost its main backer, has spent billions of dollars without a profit in sight, and lost its CEO for a brief time.
It’s still the highest-valued AI startup in the world.
The path was never clear. But if you zoom out, you had some of the world’s best people taking on one of the world’s most ambitious challenges.
With that, let’s get to the story of OpenAI.
The Vision for OpenAI
Since Alan Turing fathered computer science, technologists have been on one consistent mission to create “artificial intelligence.” Every technology, from calculators to mainframes to software, has been a form of AI.
In Turing’s words, “What we want is a machine that can learn from experience…[the] possibility of letting the machine alter its own instructions provides the mechanism for this.”
OpenAI was founded in the belief that we were on the doorstep for a major breakthrough in AI. It wouldn’t be a good thing if “big tech” monopolized that breakthrough. As Sam Altman described:
“Our grandparents – and the generations that came before them – built and achieved great things. They contributed to the scaffolding of human progress that we all benefit from. AI will give people tools to solve hard problems and help us add new struts to that scaffolding that we couldn’t have figured out on our own. The story of progress will continue, and our children will be able to do things we can’t…
How did we get to the doorstep of the next leap in prosperity?
In three words: deep learning worked.”
The following sections came from a variety of sources, including Acquired’s Nvidia podcast and Bloomberg’s Foundering Series. I also don’t discuss Elon’s departure and Sam’s departure (and ensuing return) here; we don’t know exactly what happened, and I’m an investor, not an investigative journalist.
Pre-OpenAI (pre-2015)
Three important variables laid the groundwork for OpenAI:
1. The Rise of Deep Learning
In 2012, a team of Ilya Sutskever (former Chief Scientist at OpenAI, current founder of Safe Superintelligence), Alex Krizhevsky, and Geoff Hinton (the "Godfather of AI") would crush prior benchmarks in the competition:
This was the “first widely acknowledged, successful application of deep learning.”
Bonus: improving GPU hardware laid the foundation for that deep learning breakthrough to occur.
2. The Google/Meta AI Duopoly
After the ImageNet competition, Facebook and Google formed a near duopoly on AI talent. Culminating with Google’s acquisition of DeepMind, this worried people across Silicon Valley.
Two ambitious people in particular didn’t like that idea, and wanted to form an alternative, a research lab dedicated to non-profit AI research.
Those two people? Elon Musk and Sam Altman.
3. The Rise of Sam Altman
Sam became the CEO of OpenAI and has been its most public figure since the ChatGPT moment.
Leading up to this, he had become the president of YC and a particular favorite of Paul Graham:
He built the relationships and credibility to be among Silicon Valley’s elite.
These three variables provided a technological breakthrough, a reason for an alternative to Google/Facebook to exist, and a man to help lead this new venture.
It culminated in the now-famous meeting at the Rosewood Hotel on Sandhill Road.
OpenAI is Founded
At this dinner, Elon and Sam propose the idea of OpenAI to a group of leading AI researchers. One man in particular is drawn to the idea: Ilya Sustekever (described as the “AI Genius” of OpenAI). Along with Greg Brockman (described as the “Workhorse” of OpenAI), they became the two leaders of OpenAI with the backing of $1B from Elon Musk, Peter Thiel, Reid Hoffman, and others.
Elon and Sam would be the co-chairs of the board.
The core idea of OpenAI was clear: AI could completely change the world; in that event, it’s really important for it to be developed by a company free of financial incentives.
From their original announcement in 2015: “Our goal is to advance digital intelligence in the way that is most likely to benefit humanity as a whole, unconstrained by a need to generate financial return.”
“Because of AI’s surprising history, it’s hard to predict when human-level AI might come within reach. When it does, it’ll be important to have a leading research institution which can prioritize a good outcome for all over its own self-interest.
We’re hoping to grow OpenAI into such an institution.”
…
OpenAI’s early years were defined by experimentation, releasing OpenAI Gym (AI toolkit), OpenAI Universe (creating AI for digital worlds, like video games), OpenAI Five (agents to play Dota 2), and OpenAI Dactyl (human-like robotic hand).
The Transformer
In 2017, Google released the “Attention is All You Need Paper”, unveiling the Transfomer architecture. Ilya knew the importance of this breakthrough, as described: “Ilya’s reaction was pretty affirmative right away.”
I’ll give a more in-depth breakdown in the next article. For context (ha), I’ll give a brief description of them here:
The real breakthrough of the transformer is the ability for it to incorporate the context of surrounding words into the output of its model. Transformers do this by measuring the “relevance” of surrounding words, their positions in a phrase, and storing that with the original word.
The goal is to take an initial set of data and create a series of vectors that incorporate the meaning of that data. They then predict “the probability” of the next word based on that meaning.
This laid the groundwork for OpenAI’s breakthrough with GPT-1 in a paper titled "Improving Language Understanding by Generative Pre-Training.”
The breakthrough, in Sam’s words, was that “humanity discovered an algorithm that could really, truly learn any distribution of data (or really, the underlying “rules” that produce any distribution of data).”
This “attention mechanism” happens to be extremely compute-intensive, which leads to the “discovery” of scaling: these systems get predictably better with scale.
The Inevitable Shift from Non-Profit to Profit (2019-2022)
Rightfully, there’s been a lot of debate around OpenAI’s slow transformation from a non-profit to a for profit company. This is what led to Elon and OpenAI parting ways.
But the discovery of scaling painted OpenAI into a corner. Scaling meant they had to apply huge financial resources to their systems, and the only way to attract huge financial resources was to provide huge financial incentives. And that’s what becoming a for-profit organization did.
So, in 2019, OpenAI transitioned to a “capped-profit” company and raised $1B from Microsoft and some other investors. This creates a “capped-profit” company managed by the non-profit OpenAI organization:
Bolstered by the new funding, OpenAI starts to take off. In 2020, they launched GPT-3. In 2021, they launched Codex (the model behind GitHub Copilot). That same year, they launched Dall-E. They raise another $1B from Microsoft as well.
It all leads up to the moment we know so well.
“Chat With GPT-3.5”
“In 2022, OpenAI was a quiet research lab working on something temporarily called “Chat With GPT-3.5”...We always knew, abstractly, that at some point we would hit a tipping point and the AI revolution would get kicked off. But we didn’t know what the moment would be. To our surprise, it turned out to be this.” - Sam Altman
To Microsoft’s credit, they saw this opportunity masterfully, and invested TEN BILLION DOLLARS into OpenAI. They moved, as fast as they could, integrating OpenAI into their products, all of them. It became THE priority for the company.
Since then, the story’s been well told:
OpenAI has become the defining AI company of this movement.
Sam Altman departed and returned to OpenAI.
They’ve seen increasing competition from Anthropic, Meta, xAI, Google, and recently DeepSeek.
They raised billions of dollars and became the 3rd highest-valued private company in the world.
They released reasoning models like o3, creating a new scaling vector.
In December 2024, they passed 300M users.
They’re reportedly at ~$4B revenue run rate.
This brings us to the current state of the company:
On the cusp of general-purpose agents.
Facing increasing competition, especially from open-source models.
Figuring out how to manage their governance structure, create a sustainable business model, and develop “superintelligence.”
As Sam puts it, “We are now confident we know how to build AGI as we have traditionally understood it. We believe that, in 2025, we may see the first AI agents “join the workforce” and materially change the output of companies. We continue to believe that iteratively putting great tools in the hands of people leads to great, broadly-distributed outcomes.”
In my next article, I’ll cover OpenAI’s business and technology, before discussing what the future of OpenAI looks like after that.
As always, thanks for reading!
Disclaimer: The information contained in this article is not investment advice and should not be used as such. Investors should do their own due diligence before investing in any securities discussed in this article. While I strive for accuracy, I can’t guarantee the accuracy or reliability of this information. This article is based on my opinions and should be considered as such, not a point of fact. Views expressed in posts and other content linked on this website or posted to social media and other platforms are my own and are not the views of Felicis Ventures Management Company, LLC.
Thanks a lot for the time that you invest to create and share your ideas. As deepseek is open source and its license allows one to use it for commercial purposes would that not reduce the valuation of OpenAI. OpenAI had a very high barrier to entry- infrastructure and API based monetisation. Thats gone with deepseek If businesses can fine-tune or deploy DeepSeek models without OpenAI’s constraints, they may opt for a self-hosted solution rather than paying for OpenAI’s services.
Wonderful work! please prioritise PALANTIR deepdive to tell and decode LLMs' crutial role on real world industries and business' data and business plan. it deserves an multipart approach as well ı guess (the most mystified corporate on the earth maybe) ... many thanks!