AI21 Labs, an innovative startup situated in Tel Aviv and dedicated to the advancement of an array of text-generating AI tools, has successfully secured an impressive sum of $155 million through a Series C funding round led by esteemed contributors including Walden Catalyst, Pitango, SCB10X, b2venture, Samsung Next, and the distinguished Amnon Shashua, who not only serves as the founder of Intel-acquired Mobileye but also as a co-founder of AI21 Labs. Notably, tech giants Google and Nvidia have also participated in this funding endeavor. As a result of this funding round, AI21 Labs’ cumulative funding reaches an impressive $283 million, thereby establishing the company’s valuation at an elevated $1.4 billion—an increment from the previously reported figure of $1.2 billion by The Information.
Intriguingly, the saga of AI21 Labs traces back to October 2020, when the company emerged from the shroud of secrecy to unveil its inaugural creation, the AI writing tool known as Wordtune. Yoav Shoham, one of AI21 Labs’ co-CEOs, shared with TechCrunch via electronic correspondence, “AI21 Labs came out of stealth in October 2020 with our first launch of [AI writing tool] Wordtune.” In the interval since that seminal launch, AI21 Labs has demonstrated impressive growth, effectively doubling its workforce while harboring ambitions to further expand its human resources.
AI21 Labs found its genesis in the year 2017, meticulously crafted by the visionary trio comprising Shashua, Shoham, and Ori Goshen—the latter of whom shares the mantle of co-CEO. Of noteworthy mention is Shoham’s distinguished background as a professor emeritus at Stanford and his prior success in vending two companies to the behemoth Google. His ventures encompass the realm of time management, courtesy of Timeful, and the orchestration of social networks through the Katango enterprise. Goshen, meanwhile, bolsters the leadership framework with his own entrepreneurial legacy, having steered the helm of numerous Israel-based tech ventures, including the telecommunications analytics enterprise, Crowdx.
AI21 Labs’ pièce de résistance takes the form of AI21 Studio, an ingenious developer platform adopting a pay-as-you-go model, geared toward crafting tailor-made text-centric business applications on the bedrock of AI21’s exclusive text-generating AI models. Of particular prominence is the cutting-edge Jurassic-2 model, emblematic of AI21’s relentless pursuit of innovation. Complementing this cornerstone offering is the widely-regarded Wordtune—an AI assistant proficient in multilingual reading and writing, akin in essence to the widely embraced Grammarly.
Clients find themselves empowered to harness the potential of AI21 Labs’ platform through APIs, thereby affording them precise access to generative AI use cases. These encompass the spheres of summarization, paraphrasing, grammar refinement, and spelling correction. Noteworthy is the diverse linguistic spectrum supported by AI21 Labs’ models, which encompasses languages as diverse as Spanish, German, Italian, and Dutch.
Standing as a stalwart contender within the scorching domain of generative AI, AI21 Labs finds itself amidst a proliferating array of both nascent startups and established entities. Industry titans such as Google, AWS, and Microsoft proffer toolsets bearing semblance to AI21 Studio. Moreover, the competitive landscape includes upstarts like Cohere, OpenAI, and Anthropic, with a subdued inclusion of marketing-focused vendors such as Jasper, Regie, and Typeface. Despite being at a funding juncture that is relatively modest compared to some peers, Yoav Shoham maintains a steadfast assertion that AI21 Labs’ solutions hold a distinct edge across several dimensions.
Despite potential superficial resemblances, AI21 Labs’ offerings are meticulously fashioned upon some of the most expansive and intricate large language models the world has witnessed. A point of differentiation lies in the nuanced mastery they afford, transcending the capabilities of several generative AI applications prevalent in the market. This prowess is buttressed by the veracity of their training data, which remains aligned with contemporaneous events—an attribute in stark contrast to text-generating models anchored in antiquated data repositories, rendering them inadequate in addressing inquiries pertinent to present-day occurrences.