Recently, Amazon announced that it is betting $4 billion on Anthropic, a competitor of OpenAI and an artificial intelligence (AI) unicorn, to increase the layout of artificial intelligence cloud fields. The investment is Amazon's largest investment in generative AI to date, in response to two competitors, Google and Microsoft.
However, just days after Amazon announced the investment, news broke that Anthropic was refinancing. On October 4, it was reported that Anthropic was negotiating with Google and other investors to complete a new round of financing of at least US$2 billion (about 14.394 billion yuan), with a post-investment valuation of between US$20 billion and US$30 billion — more than five times the company's valuation of 4 billion yuan in March this year. For a while, this two-year-old artificial intelligence startup was in the limelight.
Anthropic: OpenAI Competitor?
It is reported that Anthropic AI was established in 2021 by Dario Amodei, former vice president of research at OpenAI, and former OpenAI employees Jared Kaplan, Sam McCandlish, Tom Brown and others who were involved in the development of GPT-2 and GPT-3. In January this year, Anthropic released Claude, an AI chatbot model product based on Anthropic LM v4-s3, which is highly recognized in the technical circle and is considered a strong competitor to ChatGPT.
Anthropic told investors that the company's annual revenue has been around $100 million, but due to the deep cooperation with Amazon, it is expected that its revenue will grow significantly in 2023, reaching $200 million, with monthly revenue close to $17 million; By the end of 2024, Anthropic's annual revenue will reach $500 million, which is 1/200th of Anthropic's valuation and much higher than OpenAI's valuation multiple.
Based on this transaction, Amazon AWS will become the primary cloud provider for Anthropic's mission-critical workloads and will enhance existing applications and intelligent experiences across businesses by incorporating Anthropic's generative AI capabilities. At the same time, Anthropic will also use the large number of new artificial intelligence chips purchased from Amazon to compete with Nvidia to train its future artificial intelligence models and accelerate its own artificial intelligence layout.
According to industry analysts, the ability and experienced R & D team, sufficient resources and financial capabilities, and the advance layout in the field of artificial intelligence - these three elements make Anthropic a new star in the field, and Microsoft's OpenAI and Google's DeepMind are listed as the top three companies in the current global AI field. What's more, Anthropic is also the only startup that is not deeply tied to the big factory. In addition to Amazon, Anthropic has also launched similar cooperation with Google Cloud, Zoom and other enterprises.
Amazon: Join the AI team?
Industry insiders pointed out that all this seems to be a competition between Anthropic and OpenAI, but behind it is the giant battle between Amazon and old rivals such as Microsoft and Google. Microsoft and OpenAI have joined forces to stir up the global artificial intelligence market and have a head start in it. Google brought its own large models and related products to a high-profile shot, quickly keeping up with the wave. As the wave of generative artificial intelligence intensified, Amazon, which was watching the fire from afar, finally couldn't sit still, chose to imitate the model of Microsoft and OpenAI, and bet heavily on Anthropic to join this AI team battle.
In order to save the situation of lagging behind in artificial intelligence, Amazon is trying to find a unique way and find a path with its own characteristics. On the one hand, unlike Microsoft and Google's approach of starting from consumer-grade products, Amazon uses the AWS platform to focus on enterprise users, covering IaaS instances, PaaS platforms, and SaaS software full-stack technology layers. For example, the large model Titan allows customers to customize the Titan model with their own data; Cloud service Bedrock allows users to use the underlying models offered by Amazon and other startups, including Anthropic; CodeWhisperer, a generative AI code development service, helps developers write, debug, optimize, and deploy code; The low-cost, high-performance generative AI instance EC2 Inf2 uses Amazon's self-developed Inferentia2 chip to provide high-performance and relatively low-cost AI inference.
On the other hand, Amazon is also shortening its own expansion cycle in the field of artificial intelligence through "buying, buying, buying". Mergers and acquisitions have played an important role in Amazon's expansion. From a simple online bookseller to the largest e-commerce platform, and then to achieve an international layout through inclusiveness in the middle and late stages, investment and mergers have left a strong mark in the history of Amazon's e-commerce expansion.
Whether it's buying instead of research and development or buying up competitors, Amazon is good at consolidating its dominance through mergers and acquisitions. A professional once commented: "If Amazon realizes that a company poses a threat to itself, it will often not continue to fight for the market, but will directly acquire competitors." ”
Compared with Microsoft's initial investment in OpenAI of only $1 billion, Amazon's sale of Anthropic is $4 billion, and the industry is looking forward to what kind of chain reaction will occur in the future.
AI Security: A New Outlet?
It is worth noting that since its inception, Anthropic's vision has been to be a company focused on AI safety research, hoping to develop controllable and explainable AI systems. In terms of AI safety, Anthropic pioneered the concept system of "constitutional AI", emphasizing the alignment of values in artificial intelligence systems. It is reported that Anthropic founder Dario Amodi and a group of colleagues decided to leave and found Anthropic because they had doubts about the ethical and safe way OpenAI handled technology during their tenure at OpenAI.
Previously, Dario Amoudi had said he would rather Claude (Anthropic's generative AI assistant) be boring than get dangerous.
This philosophy coincides with Amazon. Dai Wen, director of the solution architecture department of Amazon Web Services Greater China, said on the Amazon Web Services re:Inforce2023 China site held recently that generative AI applications are like icebergs on the surface of the sea, and to safely harness this new technology, you also need to pay attention to the glaciers under the surface. Security is an unavoidable and important issue in building generative AI, and enterprises can only better accelerate business innovation with AI if they protect data, models and applications during the AI journey.
In fact, Amazon Web Services has long proposed the "onion-type" security protection model, which divides security protection into multiple levels, each layer has different security policies and controls. "Security has always been a top priority for Amazon Web Services and is our top priority." Dai Wen said.
After the rise of large models, the direction of "AI Safety" has attracted the attention and discussion of many industry experts. The research team of Yoshua Bengio, a Turing Award winner and one of the three giants of deep learning, recently published in Nature mentioned that artificial intelligence systems can become an important assistant for scientists to discover new knowledge, but there are also potential security risks. For example, scientific data has problems such as incompleteness and bias, which need to be standardized; You also need to consider the accessibility of the data, privacy, etc. Standardization of models and data is also necessary.
Max Tegmark, a professor at the Massachusetts Institute of Technology and founder and president of the Future of Life Institute, also said that in conversations with AI researchers and CEOs of tech companies, it's clear that there is a general sense of anxiety about the risks posed by AI.
At the 2023 Bund Conference, the Greater China Region of the Cloud Security Alliance (CSA), the world's authoritative international industry organization, announced the establishment of the "AI Security Working Group", and more than 30 institutions including China Telecom, Ant Group, Huawei, Baidu, Volcano Engine, Xidian University, and the National Financial Evaluation Center became the first batch of initiators. The organization is committed to working together to solve the security challenges caused by the rapid development of AI technology.
It can be seen that AI security seems to be becoming a new outlet in the field of artificial intelligence. Amazon "got on the car" at this time, whether it can set off a new round of artificial intelligence climax, let's wait and see.
Display and interaction technology continues to break through, and the metaverse is becoming a reality
To boost the new energy vehicle industry, the policy combination is "powerful"
Author丨Song Jing
Editor丨Zhao ChenmeiEditor丨Maria Executive Producer丨Lian Xiaodong