Things are changing quickly at OpenAI; which is becoming par for the course in the software world.
November 21, 2023
Its been an eventful and an emotional week for everyone at OpenAI.
The generative AI pioneer fired its co-founder Sam Altman on Friday. Within 24 hours, its Chairman and top lieutenants had all quit. By the end of the weekend, more than 90% of its employees have threatened to do the same. Now Altman’s been hired by Microsoft (Nasdaq: MSFT) and he might take those key employees with him. Meanwhile, OpenAI’s Board is in chaos and is now begging Sam to rejoin them again.
That sequence of events will make your head spin. Talent migrates across the software world at a breakneck speed, and the chess board can be rearranged extremely quickly.
To illustrate further, just take a look at this timeline — which shows how long it took for the software world’s greatest disruptors to reach 1 million users:
Netflix (Nasdaq: NFLX) is the OG of disruption. Nearly a quarter century ago, it offered a paid subscription service that would mail DVDs directly to your door (and then later stream them directly into your living room). Customers swapped out their Blockbuster memberships for the Netflix subscription, and the company reached one million paying users around 3 1/2 years later.
The aftermath of the dot-com bubble was terrible for the Nasdaq but was incredible for digital technologies. Mobile device adoption and cloud computing made it much easier for software companies to launch new apps and immediately get them into the hands of customers. The time required to reach one million users fell from years to months.
And then last year, something incredible happened. OpenAI unleashed its AI-powered chatbot, “ChatGPT”, to the world. It spread like digital wildfire. After launch in November 2022, it had amassed one million users in a period of only 5 days. Just one year later, it now has more than 180 million unique active users all around the world.
GPT raised the bar on how quickly new users could be acquired. And as you’d expect, its competitors frantically scrambled to respond.
Alphabet had already been working on AI for decade, though it quickly incorporated its own generative AI to supplement its existing search platform. Meta Platforms (Nasdaq: META) introduced LLaMA as an open-source option for developers who wanted to build on a smaller and less expensive model.
It suddenly felt like everyone had their own LLM. And as end users, we had a plethora of options to choose from.
But in the business world, there’s also no free lunch. Each query asked to ChatGPT cost around three cents for it to respond. 180 million users asking questions daily adds up quickly, and OpenAI turned to Microsoft to fund its ambitious growth.
Microsoft hosted GPT on its Azure cloud infrastructure, charging OpenAI for the computing that was required for each query. It also committed up to $10 billion of an equity investment into OpenAI, in order to progress the fundamental research and also to incorporate the models as a “Copilot” addition to its enterprise products like Office and Outlook.
OpenAI’s growth and its popularity skyrocketed, and its nearly-infinite potential wasn’t overlooked by investors. In its most recent funding round, the private company was worth an estimated $86 billion.
Yet despite OpenAI’s accelerating growth and growing popularity, there was a very problematic elephant in the room.
The company was originally founded eight years prior, as a non-profit whose mission was to develop a safe and efficient artificial general intelligence.
It wanted to serve the interests of humanity, and had already pivoted its strategy multiple times — from gaming to robots and now to large language models.
Several employees, including its co-founder and chief scientist Ilya Sutskever, believed it was growing much too quickly. Rather than plowing full-steam ahead into enterprise and global deployment, perhaps it was more important to double-down on the “safe” part of its stated mission. After all, the dangers of AI’s biases could cause harm if they weren’t accounted for — and this attracting an increasing level of regularity scrutiny.
That rift between those wanting slow and steady growth and breakneck-speed global growth grew wider over time. It became clear that the divisions of the company were not, in fact, being worked out constructively.
Which is what led us to the events of this following week. Ilya called Sam to an impromptu meeting and then fired him. Sam brought his team and his talents to Microsoft, who is now a bit less confident in its relationship with OpenAI. The first domino fell, and it caused a chain reaction that resulted in a nuclear-scale implosion.
This is a story that will be talked about for years, involving dozens of key stakeholders and a ton of moving parts.
Perhaps the most important reminder is that OpenAI is worth $86 billion, yet it only has 770 employees. The company grew faster than its internal controls and leadership strategy could keep up with.
We’ve seen implosions like this before. The bankruptcies of WeWork and Sam Bankman-Fried’s FTX immediately come to mind.
And while software is nearly-infinitely scalable and represents one of the most exciting industries to invest in, its fortunes can be lost just as quickly as they’re made.
Investors traveling at light speed should expect a bumpy ride. Make sure your spacecraft has a pretty secure seatbelt.
Already a 7investing member? Log in here.