Anthropic in Talks to Acquire Decart AI for $6 Billion
The deal would mark Anthropic's largest acquisition to date as it seeks to optimize inference efficiency and expand into real-time world models.
Anthropic PBC is reportedly in negotiations to acquire the Israeli AI startup Decart for approximately $6 billion. The move signals a strategic push by the AI safety-focused lab to secure critical infrastructure for model efficiency as it prepares for a potential initial public offering.
According to Bloomberg, the $6 billion price tag would represent the largest known acquisition in Anthropic's history. Decart has emerged as a high-value target due to its specialization in real-time world models and the development of the Decart Optimization Stack (DOS), a suite of tools designed to significantly improve AI inference efficiency. The startup's technical capabilities were highlighted by the launch of "Oasis," the first fully playable, real-time AI-generated open-world model, which attracted 1 million users within three days of its release.
The Race for Inference Efficiency
Founded in 2023 by Dean Leitersdorf and Moshe Shalev, Decart has scaled rapidly within the competitive AI landscape. The company has raised approximately $450 million in total funding, including a $300 million round completed in May 2026 that valued the firm at $4 billion. This rapid growth has attracted attention from several of the world's most powerful technology entities; reports indicate that Nvidia, Amazon, and SpaceX have also shown interest in acquiring the lab.
For Anthropic, the acquisition is less about adding a new model and more about the underlying plumbing of AI. Running massive large language models (LLMs) requires immense compute power and incurs staggering operational costs. By integrating Decart's optimization technology, Anthropic could drastically reduce the latency and cost associated with AI inference, allowing it to serve more users more efficiently on existing hardware.
Strategic Implications
This deal underscores a pivotal shift in the AI arms race. While the previous era focused primarily on the size of parameters and the volume of training data, the industry is now pivoting toward "inference efficiency" as a primary competitive moat. The ability to extract maximum performance from limited hardware is becoming as valuable as the models themselves, particularly for companies scaling toward commercial viability.
Beyond cost savings, Decart's work on real-time generative video and world models provides Anthropic with a potential bridge into multimodal experiences that go beyond text and static images. This would allow the company to compete more directly with other frontier labs developing interactive, simulated environments.
What's Next
While the talks are reportedly advanced, the deal remains unconfirmed by the parties involved. Market observers will be watching to see if the $6 billion valuation holds or if a bidding war erupts among the other interested suitors. If finalized, the acquisition will be a key indicator of Anthropic's financial strategy and technical roadmap leading up to its anticipated public debut.