Anthropic Reportedly in Talks to Buy Decart for $6B: A Bet on Efficiency Before IPO

Anthropic is reportedly in talks to acquire Decart, an Israeli AI startup focused on chip efficiency, for about $6 billion. If it closes, it would be the largest deal in Anthropic's history and a clear signal that the company is betting on cost control, not just model scale, as it prepares for a public listing.

The $6 Billion Bet on Chip Efficiency

Decart does not build a rival chatbot. It builds software that makes chips more efficient, cutting the cost of training and running AI models. The startup's most public-facing demo, the Oasis generative-video world model, grabbed attention in October 2024, but the core value is the unglamorous work of lowering compute bills. The proposed price reflects a rapid re-rating: Decart was valued at $3.1 billion in August 2025 and around $4 billion in May 2026 after a $300 million round led by Radical Ventures. The $6 billion figure marks roughly a 50% premium over that last mark.

Why Anthropic Is Buying an Efficiency Company Before Its IPO

Anthropic's own valuation has climbed to roughly $965 billion after its most recent mega-round, and the company confidentially filed an S-1 with the SEC in early June. With an IPO in view, the number that public-market investors will scrutinize most closely is compute cost. Buying Decart is a way to wring more from the hardware Anthropic is already securing through a multi-year GPU deal with CoreWeave to run Claude at production scale. The message is that thrift, not sheer size, is what the market now wants to see.

Anthropic was not the only suitor circling. Nvidia, SpaceX, and Amazon were all reportedly interested in Decart, underscoring that efficiency has become a strategic asset in its own right.

What It Means for AI Builders

If the acquisition closes, expect Decart's chip-efficiency software to be integrated into Anthropic's Claude infrastructure. That could lower per-chip costs, potentially affecting API pricing, capacity, or the ability to handle longer inference runs. The move also signals that even the most capital-intensive frontier labs are pivoting toward cost discipline. For builders, this could mean more stable pricing and better capacity guarantees from Anthropic, but the deal is still unconfirmed. The broader trend is clear: efficiency is now a first-class concern in the AI arms race.

Caveats

The talks are not final, both companies have declined to comment, and the deal could fall apart. Valuation and timing could shift as due diligence continues. Market conditions for AI compute remain volatile, and no public announcement has been made.