What Decart Does

Decart is a Tel Aviv-based AI research lab founded in 2023 that specializes in two related areas:

• Real-time generative video and world models — The company builds systems that generate and edit video in milliseconds rather than minutes. Its products include MirageLSD (Live Stream Diffusion), the first real-time zero-latency generative AI video model, and Lucy, a world-editing model for live video applications like virtual try-on and advertising .
• AI chip optimization infrastructure (DOS) — Decart's core product is DOS (Decart Operating System), a GPU-optimization layer that dramatically accelerates inference. DOS 2.0 delivers over 1,600 tokens per second for agentic inference — roughly eight times the industry average — and runs across Nvidia, AWS Trainium, and Google TPU hardware .

The company calls itself a "vertically integrated AI research lab" operating across the full stack from hardware optimization to world models .

Recent Funding and Valuation Trajectory

Decart raised a $300 million Series B round in May 2026, led by Radical Ventures, at a valuation of nearly $4 billion . Just three months earlier, it had raised capital from Nvidia at that same $4 billion valuation . The reported $6 billion acquisition price represents roughly a 50% premium over the May round .

The Competitive Bidding Process

Before Anthropic entered talks, Decart had already attracted interest from multiple major tech players:

• Nvidia was in advanced negotiations to acquire Decart first, but those talks shifted as larger bids emerged .
• SpaceX, Amazon, and Nebius were all reportedly linked to the bidding process, with Decart initially seeking a sale price in the $6–$7 billion range .
• Anthropic emerged as the leading bidder in mid-August, with talks becoming the most advanced and public .

The competition illustrates how scarce Decart's combination of chip-efficiency software and real-time video technology had become among AI infrastructure targets .

How It Fits Into Anthropic's Strategy

Pre-IPO compute buildup. Anthropic is preparing for a highly anticipated IPO and has raised $65 billion in total financing . The Decart deal is intended to help Anthropic handle surging compute demand by making its existing chip fleet run far more efficiently . Bloomberg Intelligence analysts noted Decart's "world models" could reduce the cost of training AI, directly improving Anthropic's gross margins, which the company is targeting at 77% ahead of its public listing .

2026 acquisition spree. The Decart talks follow a pattern of Anthropic aggressively buying technical capabilities in 2026 rather than building them organically. The company has been exploring multiple acquisitions to close capability gaps before going public .

Competitive positioning vs. OpenAI and Google. The acquisition would help Anthropic address two key competitive disadvantages:

• Chip efficiency — OpenAI and Google have massive dedicated compute fleets. Decart's DOS software can run across Nvidia, AWS, and Google TPU hardware, meaning Anthropic could get more throughput from the same chips than rivals .
• Real-time video and world models — OpenAI has Sora (video generation) and Google has Veo, but Decart's real-time, low-latency approach offers a different architecture. Acquiring Decart would give Anthropic its own video-generation and world-simulation capabilities, closing a product gap with competitors .

If completed, the deal would be Anthropic's largest acquisition by far and signal to public investors that the company is willing to spend aggressively on technical infrastructure to compete with better-capitalized rivals .

What Decart Does

Decart is a Tel Aviv-based AI research lab founded in 2023 that specializes in two related areas:

• Real-time generative video and world models — The company builds systems that generate and edit video in milliseconds rather than minutes. Its products include MirageLSD (Live Stream Diffusion), the first real-time zero-latency generative AI video model, and Lucy, a world-editing model for live video applications like virtual try-on and advertising .
• AI chip optimization infrastructure (DOS) — Decart's core product is DOS (Decart Operating System), a GPU-optimization layer that dramatically accelerates inference. DOS 2.0 delivers over 1,600 tokens per second for agentic inference — roughly eight times the industry average — and runs across Nvidia, AWS Trainium, and Google TPU hardware .

The company calls itself a "vertically integrated AI research lab" operating across the full stack from hardware optimization to world models .

Recent Funding and Valuation Trajectory

Decart raised a $300 million Series B round in May 2026, led by Radical Ventures, at a valuation of nearly $4 billion . Just three months earlier, it had raised capital from Nvidia at that same $4 billion valuation . The reported $6 billion acquisition price represents roughly a 50% premium over the May round .

The Competitive Bidding Process

Before Anthropic entered talks, Decart had already attracted interest from multiple major tech players:

• Nvidia was in advanced negotiations to acquire Decart first, but those talks shifted as larger bids emerged .
• SpaceX, Amazon, and Nebius were all reportedly linked to the bidding process, with Decart initially seeking a sale price in the $6–$7 billion range .
• Anthropic emerged as the leading bidder in mid-August, with talks becoming the most advanced and public .

The competition illustrates how scarce Decart's combination of chip-efficiency software and real-time video technology had become among AI infrastructure targets .

How It Fits Into Anthropic's Strategy

Pre-IPO compute buildup. Anthropic is preparing for a highly anticipated IPO and has raised $65 billion in total financing . The Decart deal is intended to help Anthropic handle surging compute demand by making its existing chip fleet run far more efficiently . Bloomberg Intelligence analysts noted Decart's "world models" could reduce the cost of training AI, directly improving Anthropic's gross margins, which the company is targeting at 77% ahead of its public listing .

2026 acquisition spree. The Decart talks follow a pattern of Anthropic aggressively buying technical capabilities in 2026 rather than building them organically. The company has been exploring multiple acquisitions to close capability gaps before going public .

Competitive positioning vs. OpenAI and Google. The acquisition would help Anthropic address two key competitive disadvantages:

• Chip efficiency — OpenAI and Google have massive dedicated compute fleets. Decart's DOS software can run across Nvidia, AWS, and Google TPU hardware, meaning Anthropic could get more throughput from the same chips than rivals .
• Real-time video and world models — OpenAI has Sora (video generation) and Google has Veo, but Decart's real-time, low-latency approach offers a different architecture. Acquiring Decart would give Anthropic its own video-generation and world-simulation capabilities, closing a product gap with competitors .

If completed, the deal would be Anthropic's largest acquisition by far and signal to public investors that the company is willing to spend aggressively on technical infrastructure to compete with better-capitalized rivals .