AI21 Raises $53 Million More for Funding, While OpenAI Faces Internal Chaos
AI21, a leading competitor to OpenAI in the enterprise AI market, has recently secured additional funding. This comes at a critical time for AI21’s rival, as OpenAI is currently experiencing internal turmoil and potential instability. AI21 has raised $53 million in its series C round of funding, bringing the total funding amount to $208 million and the company’s valuation to $1.4 billion. The company has now accumulated $336 million in total funding.
OpenAI is among a few companies, including Cohere and Anthropic, that offer large language model (LLM)-driven AI systems to enterprises, competing directly with OpenAI. However, unlike OpenAI, these companies do not provide open-source solutions. With OpenAI’s uncertain future, enterprise companies seeking a secure and reliable AI partner may consider exploring alternative options.
AI21 Offers Choice, Robustness, and Safety to Enterprise Companies
In an interview with VentureBeat, Yoav Shoham, co-founder and co-CEO of AI21, emphasized the importance of choice, robustness, and safety for enterprise companies. AI21 works closely with its enterprise clients, providing a personalized and high-quality service known as a “white-glove” approach. Shoham highlighted that this level of individualized attention is not something OpenAI can offer. Currently based in Tel Aviv, AI21 boasts a team of 250 employees.
Notably, Intel Capital and Comcast Ventures are among the latest investors to join AI21’s funding round, joining the ranks of other influential investors like Nvidia and Google.
“OpenAI is the main company AI21 sees when competing for deals,” Shoham acknowledged. However, he expressed confidence in AI21’s ability to perform well against OpenAI. He also stated that AI21 is actively securing its position in the market by appointing Tom Nides, the former US Ambassador to Israel and former leader of Morgan Stanley, as a board member.
AI21’s Approach: Task-specific AI Systems for Enterprises
In contrast to OpenAI’s general-purpose large language models like ChatGPT, AI21 focuses on developing task-specific AI systems for enterprise clients. According to Shoham, the chat interface, popularized by ChatGPT, may not always be the most efficient way to work with large language models. OpenAI’s strategy, in his opinion, relies on brute force and lacks the ability to excel at specific tasks that enterprise clients typically require.
Shoham pointed out that OpenAI’s business model is not centered around working closely with enterprise clients. Instead, OpenAI operates at scale and serves over 100 million users. Enterprise companies, on the other hand, seek robustly programmed applications tailored to their unique needs. Shoham sees scripting tools like Langchain as a step in the right direction for utilizing OpenAI’s large language models via an API but believes that true enterprise AI requires the development of full programs.
Looking ahead, Shoham predicts that large language models will evolve beyond their current form. He suggests that in the near future, discussions will revolve around “AI systems” instead of LLMs. OpenAI’s recent significant price drop for its services was seen by Shoham as a strategic move to capture market share, even if it means operating at a loss. In contrast, AI21 is focused on building both general-purpose LLMs and task-specific large language models, accompanied by robust algorithms tailored to specific use cases.
One concrete example of AI21’s success in delivering task-specific models is their summarization model, which outperformed GPT-4, ChatGPT, and Claude in a test conducted by a major financial institution. While the challenges of working with LLMs have slowed enterprise adoption, Shoham notes a recent surge in enterprise interest over the past six months.
“Companies have moved from sporadic experimentation to mass experimentation,” Shoham stated. He further added, “Within a year, you’re going to see real money flowing into this field.”
