AI research startup Anthroponic aims to raise up to $5 billion in the next two years to take on rival OpenAI and enter more than a dozen major industries, according to company documents obtained by TechCrunch.
The pitch deck for the 2020 fundraising round for Anthropics SeriesC, founded by former OpenAI researchers, outlines these and other long-term goals.
On board, Anthropopic says it plans to build a “frontier model” — tentatively called “cloud-next” — that’s 10 times more capable than today’s most powerful AI, but that will cost a billion dollars over the next 18 months. .
Asked for comment, an Anthropologie spokesperson said: “We are planning additional product announcements and will be talking about them soon.”
The Information reported in early March that Anthropic was looking to raise $300 million at a $4.1 billion valuation, bringing the total raised to $1.3 billion. At the time of the document’s creation, only half was raised from a “mysterious investor,” but the deck confirms that target number.
Anthroponic describes the boundary model as a “next-gen algorithm for AI self-teaching,” a reference to an AI training technique called “constitutional AI.” At a high level, constitutional AI seeks to provide a way to adapt AI to human needs—having systems respond to questions and perform tasks using simple guiding principles.
Anthropoque’s approximate boundary model requires 10^25 FLOPs, or floating point operations – several orders of magnitude larger than even today’s largest models. Of course, ShThis translates into computation time, which depends on the speed and scale of the system doing the computation. Anthroponic points to clusters with “tens of thousands of GPUs” (in its fleet).
This boundary model can be used to build virtual assistants that answer emails, conduct research, and produce art, books, and more, some of which we’ve already discovered like GPT-4 and other large language models.
“These models can begin to automate a large part of the economy,” says Peach Dick. “We believe that the companies that train the best 2025/26 models will be far enough away that no one will succeed in the next cycles.”
The boundary model is a cloud-based successor to the anthroponic chatbot, which includes searching, summarizing, writing and coding on a variety of documents, and answering questions about specific topics. In these ways, it is similar to OpenAI’s ChatGPT. But it is the anthroponic cloud – thanks to constitutional AI – that makes it “much less likely to produce harmful effects,” “easier to communicate with” and “more in control.”
Anthroponic released the cloud commercially late last year following a closed beta process, allowing 15 partners initial access. Among its beta users and potential customers, it counts the following industries (with an asterisk indicating that the human model is in the planning stage):
- Legal Document Summary and Analysis*
- Medical patient records and analytics*
- Customer service emails and chat
- Coding for consumer and B2B models
- Productivity related search, document editing and content generation*
- Chatbot for public question and answer and advice
- Search using natural language responses
- HR functions such as job descriptions and interview analysis*
- Therapy and training
- Virtual Assistants*
- Education at all levels*
Dario Amodei, a former VP of research at OpenAI, launched Anthropoc in 2021 as a public interest corporation, with several OpenAI employees, including OpenAI’s former head of policy, Jack Clark. Amodei parted ways with OpenAI amid disagreements over the company’s direction, namely the startup’s growing commercial focus.
Anthroponic now competes with OpenAI as well as startups like Cohere and AI21 Labs, all of which are developing and producing their own text-generating — and in some cases image — AI systems.
“Anthroponic was focused on research for the first year and a half, but in September we were convinced of the importance of business, which we fully entered. [2022]” reads the pitch. “We have developed a go-to-market and initial product specialization strategy based on our core expertise, brand and where we see adoption occurring over the next 12 months.”
The deck revealed that Alameda Research Ventures, a sister firm of Sam Bankman-Fried’s collapsed cryptocurrency startup FTX, was a “silent investor” with “non-voting” shares that led Anthropics’ $580 million Series B round. Anthroponic expects Alameda’s shares to be liquidated through bankruptcy proceedings in the next few years.
Google is one of Anthropix’s investors and has pledged $300 million in Anthropix for a 10% stake in the startup. Under the deal, first reported by the Financial Times, Anthroponic has agreed to “co-develop” Google Cloud with the companies to make it a “preferred cloud provider.”[ing] AI Computing Systems.
Other Anthroposic backers include James McIlave, Facebook and Asana founder Dustin Moskowitz, former Google CEO Eric Schmidt, and Skype founder Jan Talin.
