[AI Hanip News] Yesterday's Ally Becomes Today's Enemy
OpenAI Faces Off Directly with Sponsor and Partner Microsoft
Google Sets the Stage for Competitors... Nvidia Also Joins Forces
OpenAI Reaps Returns on Investment... Competition with MS
Sam Altman, CEO of OpenAI, is having a conversation at the 'K-Startups meet OpenAI' event hosted by the Ministry of SMEs and Startups on the 9th at 63 Square, Yeouido, Seoul. Photo by Jinhyung Kang aymsdream@
View original imageOpenAI, the developer of ChatGPT, has launched an enterprise AI chatbot called 'ChatGPT Enterprise.' It is up to twice as fast and can input data four times larger in capacity compared to the paid personal version, 'ChatGPT Plus.' Most importantly, there is no need to worry about security issues. Corporate data is never used for training and is encrypted. Although the price has not been disclosed, it is clear that monetization has begun in earnest. Since the revenue model for generative AI has not yet been established, the most promising customers are enterprises. There are predictions that OpenAI will exceed 1 trillion won in annual revenue this year, but profitability cannot yet be guaranteed. This is because a huge infrastructure cost is poured into running ChatGPT.
What draws attention is the complex and subtle relationship with Microsoft (MS). MS is OpenAI’s largest sponsor and partner. In July, MS also launched an enterprise chatbot called 'Bing Chat Enterprise' using OpenAI’s technology. In this situation, OpenAI targeted the same market with a similar service. In just six weeks, yesterday’s ally has become today’s rival.
Competitor or Partner... The Complex and Subtle AI Ecosystem
Google has declared a direct challenge to MS. It launched an enterprise AI service called 'Duet AI' that is functionally and price-wise identical to MS’s offering. It drafts emails and summarizes meeting notes, with features almost identical. The price is also the same at $30 (about 40,000 won) per employee.
On the other hand, yesterday’s rival can also become an ally. Google has enabled the use of over 100 AI models on its cloud. Companies can choose from various models such as Meta’s 'LLaMA 2' and Anthropic’s 'Claude 2,' in addition to their own AI models. This means Google has laid the groundwork for competitors.
Google has also partnered with NVIDIA. In the past, Google and NVIDIA had a rivalry over AI chip performance. Google developed its own AI chip to reduce dependence on NVIDIA and even claimed its chip was superior. However, this time, Google announced the launch of a supercomputer called 'A3VMs' equipped with NVIDIA’s flagship AI chip 'H100.' At the same time, Google introduced a new model 'TPU v5' with more than twice the performance of previous AI chips.
China’s Pursuit Has Begun
Chinese companies have unveiled a large number of AI chatbots. According to local media, 11 companies including Baidu, China’s largest search engine; ByteDance, the operator of TikTok; and AI startup SenseTime have received permission to offer AI chatbot services. In China, government approval is required to operate AI chatbot services. They are evaluated on whether the security level is adequate and whether socialist values are embedded. Accordingly, Baidu immediately launched the AI chatbot 'Ernie Bot,' while ByteDance and SenseTime released 'Duobao' and 'SenseChat,' respectively.
Hot Picks Today
If They Fail Next Year, Bonus Drops to 97 Million Won... A Closer Look at Samsung Electronics DS Division’s 600M vs 460M vs 160M Performance Bonuses
- "Transaction Fees 10 Times Higher": Key Precautions for ETF Investors
- Room Prices Soar from 60,000 to 760,000 Won and Sudden Cancellations: "We Won't Even Buy Water in Busan" — BTS Fans Outraged
- "Manhole Cover Blasts Open in 12 Seconds... The Reason Behind the 'Gangnam Flood Disaster' [Report]"
- "Who Is Visiting Japan These Days?" The Once-Crowded Tourist Spots Empty Out... What's Happening?
LLM Moving on Four Legs
Google has developed 'SayTap.' SayTap is a technology that uses large language models (LLM) to command a robot dog’s movements. Instead of inputting precise codes about where to place the feet or the angles of the legs, commands can be given in natural language. It responds not only to simple and clear commands like 'walk forward slowly' or 'lift your right foot,' but also to ambiguous phrases such as 'I have good news; we are going on a picnic this weekend' with a running movement.
© The Asia Business Daily(www.asiae.co.kr). All rights reserved.