The biggest news from OpenAI since ChatGPT’s introduction was revealed on Monday: ChatGPT Enterprise, the AI chatbot’s commercial layer, will be made available on that day.
According to OpenAI COO Brad Lightcap, the tool has been in development for “under a year” with the assistance of more than 20 businesses of various sizes and sectors. Without usage restrictions, up to two times quicker performance than earlier versions, and API credits are all included with ChatGPT Enterprise. According to Lightcap, the price would not be disclosed publicly, and “it will depend, for us, on every company’s use cases and size.” Block, Canva, and The Estée Lauder Cos. were beta testers.
According to PitchBook, Microsoft’s $10 billion increase in its investment in OpenAI earlier this year made it the largest AI investment of the year. In April, the startup apparently closed a $300 million share transaction at a valuation of $27 billion to $29 billion, with contributions from companies like Sequoia Capital and Andreessen Horowitz. ChatGPT reached 100 million monthly active users two months after its November launch, setting records as the fastest-growing customer application in history: A research vice president at Gartner, Brian Burke, described the uptake as “a phenomenal uptake — we’ve honestly never seen something like it, and the curiosity has grown ever since,” to CNBC in May.
Lightcap and OpenAI found that teams at more than 80% of Fortune 500 organizations were using ChatGPT.
One significant distinction between ChatGPT Enterprise & the consumer-facing version is that, even though some of these features aren’t yet accessible in the Monday launch, ChatGPT Enterprise will enable clients to input company data to train and adapt ChatGPT for their particular industries and use cases. The business also intends to roll out ChatGPT Business, a different usage tier for smaller teams, though it did not provide a timeframe.
According to Lightcap, launching the enterprise version first and delaying the business tier “gives us a little more of a way to interact with teams in a hands-on way to learn what the deployment process looks like until we fully open it up,”
In a blog post, OpenAI stated “We do not train on your company’s information or conversations, and our models don’t learn from your usage,” adding that clients’ conversation data will be encrypted both in transit and at rest. However, the firm does record aggregate information about how the product is used, such as performance metadata and more, as is largely customary, according to Lightcap.
The launch of ChatGPT Enterprise coincides with the intensification of the AI arms race among the top chatbot developers, including OpenAI, Microsoft, Google, and Anthropic. Tech firms are rushing to release new chatbot apps as well as new features in an effort to persuade users to include generative AI into their daily lives. OpenAI released its iOS app in May, and then in July it released its Android version. Google develops its Bard chatbot on a regular basis, and Microsoft updates Bing with new capabilities like visual search. A new AI chatbot, Claude 2, was introduced by Anthropic, the AI business created by former OpenAI employees, in July, months following raising $750 million over two investment rounds.
Like many big language models, ChatGPT is expensive to run; according to a tweet from CEO Sam Altman in December, each chat probably costs OpenAI “single-digit cents,” suggesting that running the service for 100 million users a month may cost millions of dollars.
Choosing which features to emphasize was the largest challenge for ChatGPT Enterprise’s development, Lightcap told CNBC.
He stated that, out of everything launching in the following few months, “the prioritization of how you pulled ahead those things based on how consumers are using the product — and what consumers really want and what’s empowering — was the subject of a lot of debate, I would say, on the team.”
One specific instance is Advanced Data Analysis, a function of ChatGPT Plus that was formerly known as Code Interpreter. The feature “sat stack-ranked in a list with plenty of other things that we believe are kind of equally or more exciting,” according to Lightcap, but companies’ feedback prompted them to prioritize making it available sooner rather than later.
According to the company’s blog post, OpenAI intends to sign up “as many enterprises as we can during the next few weeks.”