In 2024, Nvidia acquired AI21 Labs, an Israeli company specializing in natural language processing (NLP), for approximately $1.5 billion. The acquisition was aimed at strengthening Nvidia's position in the enterprise AI market by incorporating AI21's advanced large language models and its AI orchestration technology. The deal underscored Nvidia's strategy to expand beyond hardware into software and AI services, leveraging AI21's expertise in building models that understand and generate natural language.
AI21 Labs, founded in November 2017 in Tel Aviv, Israel, by Yoav Shoham, Ori Goshen, and Amnon Shashua, had established itself as a key player in the generative AI space. Its products, including the Wordtune writing assistant and the Jurassic series of language models, had gained significant traction. The acquisition brought AI21's technology and team under Nvidia's umbrella, with the founders continuing to lead the company as a standalone unit within Nvidia.
Background of AI21 Labs
AI21 Labs was founded in November 2017 by Yoav Shoham, a professor emeritus of computer science at Stanford University, Ori Goshen, a serial entrepreneur, and Amnon Shashua, a professor at the Hebrew University of Jerusalem and co-founder of Mobileye. The company focused on developing AI systems that could understand and generate natural language, with an emphasis on making these technologies accessible to enterprises.
In January 2019, AI21 raised $9.5 million in a seed funding round. On October 27, 2020, it launched its first product, Wordtune, an AI-based writing assistant that could suggest paraphrases and rewrites in real time. Google later named Wordtune one of its favorite extensions of 2021. In August 2021, the company launched AI21 Studio, a platform for building custom language models, and introduced Jurassic-1, a natural language processing system with a token vocabulary of over 250,000.
Funding continued to grow: in November 2021, Walden Catalyst invested $20 million, followed by a $25 million Series A round led by Pitango First. In July 2022, AI21 raised $64 million in a Series B round led by Ahren, with participation from Shashua, Walden Catalyst, Pitango, TPY Capital, and Mark Leslie. On March 9, 2023, the company released Jurassic-2, claiming improved response times, broader language support, and advanced instruction following. On August 31, 2023, AI21 closed a $155 million Series C round, with new investors including Google and Nvidia.
The Acquisition Announcement
The acquisition was announced in 2024, though the exact date was not publicly disclosed. Nvidia, which had already invested in AI21's Series C round, moved to acquire the company outright for approximately $1.5 billion. The deal was part of Nvidia's broader push to offer end-to-end AI solutions, combining its graphics processing units (GPUs) with AI software and models.
At the time, Nvidia had reached a market capitalization of over $3 trillion, driven by surging demand for AI chips. The acquisition of AI21 Labs allowed Nvidia to integrate AI21's large language models, such as Jamba, into its enterprise offerings, potentially competing with other AI providers like OpenAI and Anthropic.
AI21's Technology and Products
AI21 Labs developed several notable AI technologies. Wordtune, its flagship writing assistant, used natural language processing to help users rephrase and improve text. In January 2023, the company launched Wordtune Spices, a generative AI tool that could generate a range of text options to enhance sentences.
In the large language model space, AI21 released Jurassic-1 in August 2021 and Jurassic-2 in March 2023. On March 29, 2024, the company launched Jamba, an open-weights model built on a hybrid Mamba state space model (SSM) and transformer architecture, using mixture of experts (MoE) to achieve context lengths up to 256,000 tokens. In September 2024, AI21 partnered with Amazon Web Services (AWS) to make its Jamba 1.5 family of models available on Amazon Bedrock, enabling enterprise integration.
In March 2025, AI21 released Jamba 1.6 for private enterprise deployment, claiming it outperformed other open models on multiple benchmarks. The same month, the company launched Maestro, an AI planning and orchestration system designed to improve the accuracy of models like GPT-4o and Claude 3.5 Sonnet in complex tasks.
Strategic Rationale
Nvidia's acquisition of AI21 Labs was driven by several strategic factors. First, it provided Nvidia with a portfolio of proprietary large language models and a team of researchers and engineers experienced in NLP. Second, it allowed Nvidia to offer AI software solutions that could run on its hardware, creating a more integrated ecosystem. Third, it positioned Nvidia to compete in the enterprise AI market, where companies increasingly sought customizable AI models for tasks like writing assistance, code generation, and data analysis.
The acquisition also reflected a broader trend of AI infrastructure companies moving into software. Nvidia had previously acquired Mellanox in 2020 for $7 billion and Run:ai in 2024, but the AI21 deal was notable for its focus on language models rather than hardware or infrastructure.
Leadership and Integration
Following the acquisition, AI21 Labs continued to operate under its existing leadership. Amnon Shashua remained chairman and co-founder, while Ori Goshen and Yoav Shoham served as co-CEOs. The company's advisors, including Dan Jurafsky, professor of linguistics and computer science at Stanford University, and Sebastian Thrun, adjunct professor at Stanford, remained involved in an advisory capacity.
AI21 Labs was integrated into Nvidia's software and services division, with its models made available through Nvidia's enterprise platforms. The founders expressed enthusiasm about the acquisition, noting that it would allow them to scale their technology with Nvidia's resources.
Impact on the AI Industry
The acquisition of AI21 Labs by Nvidia was seen as a significant consolidation in the AI industry. It highlighted the growing importance of proprietary language models and the competition among tech giants to control the AI stack, from chips to models to applications. The deal also underscored the strategic value of Israeli AI startups, which had become attractive targets for global tech companies.
For Nvidia, the acquisition helped diversify its revenue streams beyond hardware. While the company's GPUs remained the backbone of AI training and inference, the addition of AI21's models allowed Nvidia to offer a more complete solution to enterprises, potentially increasing customer lock-in.
The acquisition also raised questions about the future of open-source AI models. AI21 had released Jamba with open weights, but under Nvidia, the company might shift toward more proprietary offerings. However, as of the acquisition, no major changes to AI21's open-source commitments were announced.
See Also
- AI21 Labs
- Nvidia
- Large language model
- Generative AI
- Amazon Bedrock
- Amazon Web Services
- OpenAI
- Anthropic
- GPT-4o
- Claude 3.5 Sonnet
- Mamba
- Mixture of experts
- Transformer
- Natural language processing
- Wordtune
- AI21 Jurassic
- AI21 Jamba launch
- AI21 Jamba 1.5 launch
- Nvidia's 2024 market cap