On the afternoon of September 25, the generative AI platform Anthropic announced on its official website that it had received a US$4 billion (approximately 29.2 billion yuan) investment from Amazon, becoming one of the highest financing levels in the global generative AI track.
After this round, Anthropic’s total financing reached US$5.6 billion, making it one of ChatGPT’s largest competitors. Anthropic has only been established for more than 2 years, which fully proves the huge commercial potential of generative AI. It is worth mentioning that Anthropic just completed a US$100 million financing exclusively invested by South Korea’s SKT on August 13.
According to authoritative media analysis from Reuters, Financial Times, Bloomberg, and Wall Street Journal, one of the main reasons for Amazon’s heavy investment in Anthropic is to catch up with Microsoft and Google and compete for the cloud computing market share of generative AI. The investment agreement between the two parties also Proved this.
With the support of ChatGPT, Microsoft's Azure cloud market share has been growing exponentially, and the market share gap with Amazon is getting smaller and smaller. As the absolute overlord of global cloud computing, Amazon is anxious to see this and hopes to achieve overtaking in a corner by investing in Anthropic.
According to the investment agreement announced by Anthropic: 1) Anthropic will provide generative AI services to its customers through Amazon’s cloud computing services. At the same time, Anthropic will obtain the right to use AWS Trainium and Inferentia chips for training, deployment and fine-tuning AI models.
2) Amazon initially invested US$1.25 billion to purchase a small stake in Anthropic, while retaining the right to increase the investment amount to US$4 billion.
3) Anthropic provides comprehensive technical support for Amazon’s Bedrock, including model customization, data fine-tuning, safety guardrails, etc.
4) Jointly promote generative AI security, reduce the output of illegal content such as fraud, discrimination, and prejudice, and build a transparent and stable AI development and application ecosystem.
Public information shows that Anthropic was founded in 2021 and is headquartered in San Francisco, California, USA. It was established by a number of core R&D personnel from the former OpenAI’s ChatGPT research team.
In terms of products, Anthropic mainly provides two ChatGPT products, Claude and Claude Pro.
Claude is a free product that can generate various text content through natural text question and answer. Highlights include: being able to upload 5 documents in PDF, txt, and Csv formats at one time, summarizing the core content, supporting Chinese, interpreting codes, etc. (Free trial address: https://claude.ai/chats)
Claude Pro is Anthropic's latest paid product for $20 per month, mainly targeting OpenAI's ChatGPT Plus. Claude Pro's dialogue text input is five times that of Claude, and enjoys priority use during high load peaks. It supports functions such as paper summary, data query, and code generation. (Experience address: https://claude.ai/settings/billing)
However, in order to ensure the performance and stability of Claude Pro, Anthropic can only send 20 times every 8 hours. If the content sent by the user is relatively short, about 200 English sentences, each sentence has about 15-20 words, then every 8 hours Can be sent 100 times.
The explanation given by Anthropic is that whenever a user sends a large text, Claude Pro will consume a lot of computing power for contextual reading and understanding of the entire content, and frequent sending will affect the stability of Claude Pro.
According to Anthropic’s case introduction, Bridgewater Associates, a well-known asset management company, is fine-tuning data through its model and creating an AI assistant for automatically generating charts, financial data, and summarizing financial summaries;
The famous travel publisher Lonely Planet has also created a travel AI assistant through data fine-tuning, providing users with a variety of travel suggestions and reducing travel costs by nearly 80%.
In terms of product safety, Anthropic has signed an AI security agreement with official organizations, voluntarily committing to manage various risks brought by AI, and abide by the three principles of safety, security, and trust in future product release and use.
The material of this article comes from Anthropic official website. If there is any infringement, please contact us to delete it.