Source | AI Frontline ID | ai-front
Altman once said optimistically, "After the software reaches a certain level of capability, it will be like pressing a button, and profits will be poured in. In the end, the benefits will far cover the costs."
On August 30, The Information reported that as large companies continue to increase their spending on artificial intelligence, ChatGPT developer OpenAI is expected to generate more than $1 billion in revenue in the next 12 months by selling artificial intelligence software and its computing power. .
The revenue figure of $1 billion far exceeded the expectations the company had previously announced to shareholders. The report also said that the Microsoft-backed company now generates more than $80 million in monthly revenue.
Last year, before OpenAI started charging for its artificial intelligence chatbot ChatGPT, it only generated a total of $28 million in revenue. When investors bought shares from existing shareholders earlier this year, OpenAI was valued at $27 billion on paper.
The rapid growth in revenue shows that more app developers and large companies, including secretive firms like Wall Street firm Jane Street, are leveraging OpenAI's chatbot technology to find ways to make money or save costs. Microsoft, Google, and countless other companies looking to profit from artificial intelligence technology are paying close attention to OpenAI's development.
According to previous reports, although OpenAI currently has high costs and relatively low revenue, more investors believe that ChatGPT is hugely transformative, which will allow OpenAI to rely on ultra-high profit margins to become a trillionaire like Google and Amazon. USD market cap companies.
OpenAI money making strategy
OpenAI has completed a US$10.3 billion round of financing in 2023. According to US technology media reports in April, venture capital companies participating in this round of financing include Tiger Global Management, Sequoia Capital, California Anderson Hotz Fund, New York Thrive and K2 Global. In mid-January 2023, Microsoft invested US$10 billion in OpenAI, accounting for 49% of the shares.
The development and maintenance of large models require huge resource investment. OpenAI is facing cost pressure. While financing, it is also constantly monetizing ChatGPT through some methods.
Harvest C-side traffic
In February this year, OpenAI launched the subscription plan ChatGPT Plus, which is priced at US$20 per month. The value-added services provided by the paid version of ChatGPT include: skipping queues during peak hours, quick response, and priority access to new features and improvements. OpenAI also retains free services.
In March, OpenAI announced on its official blog that it would open the model APIs of ChatGPT and Whisper. Supporting ChatGPT and Whisper API calls are the gpt-3.5-turbo model and the Whisper large-v2 speech-to-text model respectively. According to OpenAI, gpt-3.5-turbo is currently the most powerful GPT-3.5 model, reducing the optimized chat cost to one-tenth of the previous GPT-3.5 model.
OpenAI priced the ChatGPT API at $0.002 per 1,000 tokens at the time, which is about 750 words. Users need to pay according to the total number of input and output tokens. At the same time, OpenAI provides dedicated instance options for developers who need to run large amounts of data through the API and who want to control specific model versions and system performance. These users will pay based on time periods to share the cost of reserved computing infrastructure. Whisper is an open source speech-to-text model announced by OpenAI as early as September 2022. It can realize transcription in multiple languages, and its API is priced at $0.006 per minute.
The user experience of ChatGPT far exceeds that of the previous generation, giving users a great sense of freshness, and users on the ToC side have brought a considerable traffic explosion. According to data from SimilarWeb, since ChatGPT was launched in the United States in November 2022, the number of active users has exceeded the 100 million mark in just 2 months. ChatGPT costs "single digit cents" per chat, according to a tweet from OpenAI CEO Sam Altman in December, suggesting it could cost millions of dollars to provide the service to 100 million people each month. Facing hundreds of millions of user visits, OpenAI’s operating strategy is to continuously optimize costs. According to previous Reuters reports, OpenAI may have continuously reduced the computing power used in ChatGPT inference in the first half of this year, which also led to a decline in the quality of users' somatosensory answers.
On the other hand, after about half a year, the popularity of ChatGPT began to decline as the novelty decreased. According to SimilarWeb data, global user visits in June were -9.7% month-on-month. Under this situation, OpenAI's focus began to shift to the ToB side.
OpenAI enters the B-side
Yesterday (Monday, August 28, local time), OpenAI launched the ChatGPT Enterprise version (ChatGPT Enterprise) for large enterprises . Officials claim that this is the most powerful version of ChatGPT to date.
Netizens commented: "Before companies adopt it, people simply don't understand how big this market will be."
According to OpenAI, in addition to performing tasks such as writing emails, drafting documents, and debugging computer code like ordinary ChatGPT, the enterprise version can also provide enterprise-level security and privacy, as well as advanced data analysis functions. OpenAI COO Brad Lightcap declined to reveal specific details on ChatGPT Enterprise pricing, noting that prices may vary based on each enterprise's needs. Lightcap said OpenAI "can work with each business to find the best plan for them."
Some media pointed out that compared with ChatGPT for ordinary consumers, an important difference of ChatGPT Enterprise Edition is that the Enterprise Edition allows customers to input company data to train and customize ChatGPT for the industry and use cases of the enterprise. ChatGPT Enterprise Edition is driven by GPT-4. Enterprise users can have priority access to GPT-4 without usage restrictions, and the access speed is twice that of ordinary GPT-4. Moreover, the enterprise version has more input content, and the context window has been expanded to 32,000 tokens and about 25,000 words, which is four times the input content of the ordinary version.
In addition, the development of ChatGPT Enterprise took less than a year and received help from more than 20 companies of various sizes from different industries. Beta users include Block, Canva and Estee Lauder. OpenAI said that enterprises are very interested in ChatGPT for enterprises, and claimed that as one of the fastest growing consumer apps in history, ChatGPT has been adopted by more than 80% of Fortune 500 company teams.
After launching ChatGPT Enterprise for large enterprises, OpenAI will also launch ChatGPT Business for smaller enterprises in the future, "giving us more ways to have substantive contact with the team and understand deployment actions before fully opening up."
A shot in the arm for everyone
The admission fee for large models is extremely high. In May this year, The Information revealed that due to the development of ChatGPT last year and the recruitment of key employees from Google, OpenAI's losses roughly doubled to about $540 million.
The gap of US$540 million will deter many start-up companies with weak resource capabilities from entering the AI field. In the past two weeks, an Indian media also reported that OpenAI spent too much money and might go bankrupt in 2024 due to declining traffic. In the article, it determined that "annual revenue is expected to reach US$200 million in 2023, and is expected to reach US$200 million in annual revenue." Reaching $1 billion in 2024 seems unlikely."
There is no doubt that the research and development of artificial intelligence requires a lot of capital. Sam Altman himself has said that OpenAI "will become the most capital-intensive startup in the history of Silicon Valley."
But even if returns on AI investments have been elusive, the ChatGPT craze has forced many companies to engage in top-down licensing of large-scale model-related development, thus changing the focus of countless enterprise technology spending. Especially cloud vendors led by Amazon, Microsoft, Google, Alibaba, Baidu and other companies. Amazon CEO previously said that large language models and generative AI are transformative. In the face of the economic downturn, Amazon will continue to invest heavily and join the generative AI competition. Large language models and generative AI will enable "Amazon to grow in the next few decades." Years can be at the heart of innovation in every business area.”
Today, OpenAI is performing slightly better than expected, giving its countless followers a bit of confidence.
Reference links:
https://www.reddit.com/r/singularity/comments/164xncx/openai_passes_1_billion_revenue_pace_as_big/
https://stock.jrj.com.cn/2023/07/07080837675456.shtml
https://wallstreetcn.com/articles/3696594
https://twitter.com/EMostaque/status/1696663084050403336