What made Anthropic walk away from the $6 billion Decart deal?
The Israeli AI startup was deep into due diligence with Anthropic when the $6 billion-$7 billion acquisition suddenly collapsed, raising questions about its technology, scalability and business strategy. The fallout could also complicate Decart’s search for another buyer and deal a blow to Israel’s ambitions to establish itself as a global AI powerhouse.
What did Anthropic discover about Israeli AI startup Decart that caused it to pull out of an almost-signed $6 billion-$7 billion deal, and what will the consequences be for Decart?
Canceling a deal at such an advanced stage leaves Decart’s founders, Dean Leitersdorf and Moshe Shalev, and their investors in an uncomfortable position. The companies were already deep into due diligence when Anthropic suddenly discovered something it did not like. What could it have been?
It is unlikely to have been a financial issue. There is only so much room for creative accounting at a three-year-old company with around 100 employees and annual revenue running at several tens of millions of dollars. The much bigger question is whether Anthropic found something disturbing or disappointing about Decart’s technology, which is designed to dramatically reduce the cost of running AI by making more efficient use of computing chips.
A key attraction of Decart’s technology is that it is supposed to be chip-agnostic. In other words, it should work not only with Nvidia’s GPUs but also with chips developed by companies such as Google and Amazon. Did Anthropic discover that the technology did not work as well as expected? Did the efficiency gains weaken at very large scales, suggesting that the technology was not sufficiently scalable?
When a deal collapses at the last minute, more speculative explanations inevitably emerge. One possibility is that Anthropic, after gaining a detailed understanding of Decart’s technology during due diligence, decided it could develop something similar itself. Another is that geopolitical considerations played a role, including questions surrounding Qatar’s significant investment in Anthropic and the possibility of a Qatari connection to the acquisition of an Israeli company.
Whatever the explanation for Anthropic’s decision to walk away in the 90th minute, the consequences for Decart could be significant.
Few potential buyers would be eager to acquire a company that Anthropic examined so closely and ultimately decided not to buy. Before Anthropic, Decart had held talks with other companies, with Nvidia reportedly the most advanced. Nvidia is also an investor in the Israeli startup and was reportedly prepared to pay more than Anthropic.
Yet Leitersdorf and Shalev were attracted to Anthropic’s offer and were willing to accept a lower price, much of it in Anthropic shares.
For now, however, Nvidia is occupied with another major transaction: its $12.9 billion acquisition of Hugging Face, which it officially announced last week.
Anthropic, meanwhile, remains a private company and is preparing for an IPO, although it has not yet published a prospectus. The offering is expected within the next two months. That makes the decision to pursue a major acquisition even more significant. Even Elon Musk, who has shown little hesitation about breaking established conventions, waited until SpaceX’s IPO was completed before acquiring the AI company Cursor for $60 billion.
There is another question worth asking: Did Anthropic conclude that Leitersdorf and Shalev wanted the deal for reasons that did not align with its own interests?
On paper, Nvidia may be a more natural buyer for Decart. It could incorporate the startup’s technology into a broader offering for customers looking to extract more performance from their AI chips. Such a deal could also prevent future competitors from gaining access to Decart’s technology.
For Anthropic, whose core business is building and selling AI models and related services, the strategic fit is less obvious.
But those who know Leitersdorf could understand why Anthropic’s offer was so attractive to him. The 27-year-old entrepreneur, a prodigy who completed a doctorate in computer science at 24, has said from the beginning that he wants to build something enormous, a company on the scale of Google or Meta.
Selling a company just three years after its founding for a single-digit sum of billions of dollars therefore could not simply be “another deal.” It would have to be something much bigger. Bringing Anthropic to Israel and having the company establish a development center here would certainly fit that vision.
It is possible that Anthropic’s executives recognized this dynamic as well. Did they worry that Leitersdorf and his team would not remain sufficiently committed to Decart after the acquisition?
Leitersdorf is still only 27 and remains the technological brain behind a company that has yet to settle on a clearly defined business model and has so far focused heavily on projects. Along the way, Decart has moved between different products and directions, including an AI model for video aimed at the gaming industry.
Anyone considering an acquisition of Decart can see the strategic logic. More efficient use of computing resources is likely to become increasingly important as AI models grow more demanding and the cost of computing becomes a greater constraint. But that does not mean the technology is a finished product. It still requires significant investment, development and commercialization.
If Anthropic concluded that Leitersdorf was more interested in the acquisition and its shares than in continuing to build Decart inside Anthropic, that alone could potentially have been enough to kill the deal.
Beyond Decart itself, which now has to decide where to go from here, there is another casualty of the collapse: Israel’s ambition to establish itself as a major center of the global AI industry.
For a country whose technology sector still relies heavily on cybersecurity, a deal of the scale and nature of the proposed Decart-Anthropic transaction would have carried symbolic weight far beyond its value. It would have brought one of the world’s leading AI companies to Israel, created a major local development center and, perhaps most importantly, provided a powerful vote of confidence in the depth of Israeli AI talent.
For a brief moment, it was possible to imagine an Anthropic sign hanging from an office tower in Tel Aviv. That vision now appears to be receding.
The Decart deal may yet be revived in another form, and the startup still has other potential paths forward. But for now, the most important question is not only why Anthropic walked away. It is whether another buyer will look at Decart and see what Anthropic once saw, and decide to go all the way.
