Nvidia in advanced talks to acquire AI21 Labs for up to $3 billion in talent-focused deal

Nvidia Pursues AI21 Labs in Major AI Acquisition Push

Nvidia is in advanced discussions to acquire AI21 Labs, an Israeli AI startup, for between $2 billion and $3 billion, according to reports from Calcalist 1

1

. The potential AI acquisition would represent more than double AI21's last known valuation of $1.4 billion from a 2023 funding round that included participation from both Nvidia and Google 2

2

. While Nvidia declined to comment on the reports, sources indicate the talks have advanced significantly in recent weeks, reaching the company's most senior levels 4

4

.

The timing of this deal reflects Nvidia's aggressive strategy to secure AI talent as competition intensifies in the artificial intelligence sector. Founded in 2017 by Amnon Shashua—who also leads Mobileye, the self-driving car technology company—AI21 Labs has quietly raised an additional $300 million earlier this year, though it never officially confirmed the funding round 2

2

. The Israeli AI startup currently generates approximately $50 million in annualized revenue, a modest figure compared to competitors like OpenAI but one that demonstrates traction in the enterprise market 3

3

.

Highly Skilled Workforce Drives Acquisition Strategy

Nvidia's primary interest in AI21 Labs appears to be its highly skilled workforce of roughly 200 employees, most of whom hold advanced academic degrees and possess specialized expertise in AI development 1

1

. This talent-focused approach mirrors Nvidia's recent $20 billion acquisition of Groq, where the transfer of high-level AI talent played a central role 3

3

. Jonathan Ross, Groq's founder and original architect of Google's TPU program, joined Nvidia along with other key team members as part of that deal.

The emphasis on acquiring expertise in machine learning and neural network architecture signals Nvidia's recognition that human capital remains as valuable as technology itself in the rapidly evolving AI landscape. Industry analysts have noted that AI21 had been considered "on the shelf" for several years, with Google previously showing interest in the company 3

3

. However, recent talks with Nvidia have progressed beyond earlier discussions with other tech giants.

Large Language Models and Enterprise Solutions at the Core

AI21 Labs offers the Jamba series of open-source LLMs, which the company claims can process long prompts up to 2.5 times faster than competing algorithms 2

2

. The Jamba architecture combines elements of a transformer model with Mamba, a neural network design that replaces traditional attention mechanisms with state space models. This mathematical structure, originally invented to study physical objects, enables Mamba models to process lengthy prompts using significantly less memory than transformers, boosting inference speeds considerably.

The company generates revenue through its Maestro software platform, which includes AI agent development tools that help developers build and manage artificial intelligence agents 2

2

. Maestro promises to improve model accuracy by 50% and can harmonize documents stored in different file formats, remove unnecessary information that might confuse AI agents, and apply various optimizations before datasets are processed 3

3

. After an AI agent generates a response, Maestro validates output accuracy and allows developers to customize processing results in specific formats such as reports.

Nvidia may integrate Maestro into Nvidia AI Enterprise, the AI software suite it ships with its graphics cards, which already includes pre-packaged large language models and development tools for building custom models 2

2

.

Strategic Shift Toward Enterprise Solutions

AI21 Labs recently changed its business focus, discontinuing Wordtune—its consumer-facing AI reading and writing assistant—to prioritize enterprise solutions 3

3

. This strategic pivot reflects broader industry trends as AI companies seek more sustainable revenue models through business-to-business offerings rather than consumer applications. The decision to pause Wordtune development came as AI21 struggled to keep pace with advances from major artificial intelligence players 4

4

.

The shift toward enterprise solutions aligns well with Nvidia's existing business model, which focuses heavily on providing AI infrastructure and tools to corporations rather than direct-to-consumer products. This compatibility may accelerate integration efforts should the acquisition close.

Nvidia Expansion in Israel Accelerates

If finalized, this deal would become Nvidia's fourth major acquisition in Israel, reinforcing the country's position as a critical hub for the company's AI development efforts 3

3

. Jensen Huang, Nvidia's CEO, has repeatedly described Israel as the company's "second home" 1

1

. The chipmaker is planning a massive R&D campus in Kiryat Tivon, just south of Haifa, that could eventually house up to 10,000 employees by 2031 5

5

.

The planned campus will include up to 160,000 square meters of office space, parks, and common areas across 90 dunams, inspired by Nvidia's Santa Clara, California headquarters 1

1

. Construction is expected to begin in 2027, with initial occupancy planned for 2031. This substantial investment in physical infrastructure demonstrates Nvidia's long-term commitment to maintaining and expanding its presence in Israel's thriving tech ecosystem.

The potential acquisition comes as Big Tech companies close out the year with multiple deals designed to push AI deeper into products, infrastructures, and monetization strategies 4

4

. Nvidia, which has become the most valuable company in history at more than $4 trillion in market capitalization, recently finalized a $5 billion investment in Intel shares as part of a larger tech partnership 4

4

.