Anthropic in Talks to Buy Decart for $6 Billion to Slash AI Infrastructure Costs

Reviewed byNidhi Govil

15 Sources

Share

Anthropic is negotiating to acquire Israeli AI startup Decart for approximately $6 billion, marking its largest deal yet. The acquisition targets Decart's chip optimization technology that could dramatically reduce Anthropic's ballooning compute expenses as it prepares for a public listing.

Anthropic Pursues Largest Acquisition to Control Spiraling Compute Costs

Anthropic is in talks to buy Decart AI for roughly $6 billion

1

2

, according to multiple reports, in what would be the Claude maker's largest AI acquisition to date. The deal has not been finalized and could still fall through

4

, but the target reveals Anthropic's strategic shift toward cost optimization as it prepares for a widely anticipated public listing.

Source: Finextra Research

Source: Finextra Research

Decart, founded in 2023 by Israeli brothers Dean and Orian Leitersdorf alongside Moshe Shalev

5

, builds software that squeezes more performance from AI chips, cutting the cost of AI training and inference. The Israeli AI startup was valued at nearly $4 billion in May after raising $300 million led by Radical Ventures

5

, with backing from Nvidia, Sequoia Capital, Atreides Management, Valor Equity Partners, and Adobe Ventures. A $6 billion price tag would represent a premium of roughly 50% over that recent valuation

2

.

Source: Jerusalem Post

Source: Jerusalem Post

Why Efficiency Matters More Than Flashy Features

Unlike OpenAI's $6.5 billion purchase of Jony Ive's io or SpaceX's $60 billion acquisition of Cursor

1

, which focused on attracting customers through gadgets and software, Anthropic's pursuit of Decart centers on cost optimization and AI infrastructure costs. The startup's Decart Optimization Stack (DOS) is marketed as a vertically integrated inference and training platform that maximizes chip utilization across hardware-aware model design, kernel tooling, proprietary compilers, and inference optimization

3

.

With Anthropic's revenue potentially reaching $100 billion this year according to investors cited by the Financial Times

1

, compute expenses could hit $56 billion. PitchBook analysts estimate Anthropic's gross profit margin reached 44% in the second quarter

1

. A 10% efficiency gain from AI training optimization would save approximately $5.6 billion annually

1

, making the $6 billion acquisition price potentially justifiable.

Decart's Technology Beyond Cost Savings

While efficiency drives the acquisition logic, Decart also develops consumer-facing AI models. Its Lucy model processes live video feeds to produce real-time footage showing people wearing clothing or accessories, addressing challenges in fashion e-commerce

5

. E-commerce platform eBay is both an investor and customer. Decart's Oasis model creates synthetic environments used in robotics and self-driving technology development, demonstrated through a playable Minecraft-style world model in 2024

3

.

Source: Gizmodo

Source: Gizmodo

These generative video and world models showcase Decart's technical breadth, though the core value for Anthropic lies in DOS's ability to reduce AI chip usage costs. Anthropic pays xAI $1.25 billion per month to train and run AI models in data centers

3

, making any technology that improves chip efficiency strategically valuable.

Competitive Pressure and Market Dynamics

Anthropic faces mounting pressure as cheaper Chinese models undercut Western labs and customers increasingly prioritize price over performance. Data from corporate expenses operator Ramp shows that Fable, Anthropic's most advanced model, accounts for just 11% of business spend on Claude products, with significantly more dollars flowing to cheaper chatbots

1

. As Meta and SpaceX push cheaper alternatives, margin improvement becomes critical for competing at the market's lower end.

Anthropic was not the only suitor. Nvidia, SpaceX, and Amazon reportedly showed interest in Decart

2

, signaling that efficiency has become a strategic asset across the AI industry. When a chipmaker, rocket company, and cloud giant all pursue the same AI startup acquisition, the market has clearly decided that cost optimization matters as much as raw capability.

Strategic Timing Before Public Listing

This would mark Anthropic's fifth acquisition of the year

2

and by far the largest. Anthropic confidentially filed an S-1 with the SEC in early June

5

, preparing for what could be the largest IPO in AI history. The timing suggests Anthropic wants to demonstrate fiscal discipline to prospective public market investors who will scrutinize compute expenses closely.

If the deal closes, Decart's team would join Anthropic's inference and performance organization

5

. Anthropic has been aggressively expanding chip and compute partnerships, including agreements to use AWS Trainium chips, Google tensor processing units, and SpaceX computing power

5

. The company has also discussed leasing computing capacity from Meta in an arrangement potentially worth $10 billion over two years

5

. Last week, Anthropic announced it was hiring engineers to co-design custom chips and AI models to make Claude run faster at lower cost

5

.

The Decart acquisition represents the other half of this equation: securing hardware capacity while simultaneously wringing more performance from each chip. For a company eyeing a public listing with investor scrutiny on profitability, buying efficiency technology sends a clear signal that Anthropic understands the AI race now rewards thrift as much as scale.

Today's Top Stories

© 2026 TheOutpost.AI All rights reserved