Way back in March of 2019, this weird thing happened where a relatively insignificant tech nonprofit called OpenAI became a “capped” for-profit company—whatever that is. The month earlier, OpenAI had announced the creation of an uncanny, über-powerful language model called GPT-2 that was supposedly just too dangerous to release. Then in November, OpenAI seemingly changed its mind and GPT-2 was released after all.
OpenAI said in the blog post about the release that it saw, “no strong evidence of misuse so far,” but added that it was impossible to “be aware of all threats.” Most people never used GPT-2, because OpenAI never injected it into a viral chatbot.
As someone who wrote about this at the time, it was puzzling to watch it all play out. OpenAI seemed like small potatoes, but it was also building creepy AI tech, and shifting in public image from being a do-gooder computer lab advertising its trepidation about harming a hair on anyone’s head to an enterprise that needed to ship something asap because it was clearly promising someone, somewhere, that they were going to get rich.
Document discovery from Elon Musk’s lawsuit against OpenAI and Microsoft has provided a tiny window into what was actually happening inside Microsoft during this bizarre time for this bizarre company, and how the transition may have turned OpenAI into the money-hungry beast it is today, with revenues growing tenfold between 2023 and 2025.
GeekWire’s Todd Bishop dug through the cache of emails, memos, texts and the like from Microsoft and OpenAI, and what he found was revealing. Microsoft, and CEO Satya Nadella in particular, had invested heavily in OpenAI by then, and were not quiet during OpenAI’s uneasy transition to for-profit status. Nor were they shy about the need to make money as soon as possible. Absolutely none of this should come as a surprise, but it makes for fascinating reading anyway.
During that gap where GPT-2 was sitting there unreleased and OpenAI had recently become a capped nonprofit, Microsoft’s chief financial officer, Amy Hood, weighed in about the company’s concerns about that “capped” part. She wrote in a July 14 email to a group including Nadella, “Given the cap is actually larger than 90% of public companies, I am not sure it is terribly constraining nor terribly altruistic but that is Sam’s call on his cap.”
GPT-3, which was even more exciting than GPT-2 was released in 2020, and the first version of OpenAI’s language model, Dall-E was released in January 2021. The next month, Microsoft and OpenAI were negotiating an additional injection of money from Microsoft, and Sam Altman wrote an email to Microsoft, saying “We want to do everything we can to make you all commercially successful and are happy to move significantly from the term sheet,” and he added that he wanted “to make you all a bunch of money as quickly as we can and for you to be enthusiastic about making this additional investment soon.”
In November of 2022, ChatGPT was released, and as you know, all hell broke loose. In January of 2023, Nadella sent a text message to Altman, saying “when do you think you will activate your paid subscription for ChatGPT?”
Altman said he was “hoping to be ready by end of jan, but we can be flexible beyond that. the only real reason for rushing it is we are just so out of capacity and delivering a bad user experience,” and asked “any preference on when we do it?”
“Let me think about it and weigh in. Overall getting this in place sooner is best,” Nadella replied. Two weeks later, he followed up and asked “how many subs have you guys added to ChatGPT?”
Three days later, the paid version of ChatGPT launched.
