Amazon is in talks to acquire AI startup Decart for roughly $7 billion, according to a Quartz report, in a deal that would rank among the largest AI-focused takeovers by a hyperscale cloud provider.

The reported price, which neither Amazon nor Decart has confirmed, has not been tied to a closing date or definitive agreement. Quartz did not disclose Decart's prior funding history, headcount, or product focus, leaving the strategic rationale to be inferred from the buyer's existing AI footprint.

Why the price matters at $7 billion

$7 billion would exceed the value Amazon assigned to its 2023 commitment to Anthropic, in which the cloud provider took a minority stake and pledged up to $4 billion in compute and cash. An outright purchase of Decart at the reported price would shift Amazon's AI posture from partner-funded model access to full ownership of an independent AI asset.

The gap between a minority investment and a full acquisition matters commercially. Ownership gives the acquirer control over roadmap, pricing, and integration with proprietary infrastructure — areas where hyperscalers have been careful to preserve optionality.

Where Decart would fit in Amazon's stack

Amazon Web Services has assembled a multi-layer AI portfolio over the past three years. At the silicon layer, the company ships Trainium2 for training and Inferentia2 for inference, both designed to reduce dependence on Nvidia GPUs. At the model layer, AWS distributes third-party foundation models — including Anthropic's Claude family — through its Bedrock marketplace.

Decart's product specialization is not described in the Quartz report. Whether the startup works on model training efficiency, inference optimization, agentic systems, or applied tooling will determine which of those layers the acquisition would most directly reinforce.

The competitive backdrop

The three largest U.S. hyperscalers have each pursued a different route to AI capacity. Microsoft deepened its OpenAI relationship and committed multi-billion-dollar compute spending to host GPT-family models. Google paired its Gemini line with custom TPU silicon. Meta distributed Llama model weights as open source while building GPU clusters at record scale.

An Amazon deal at the reported price would place the company's M&A spend in the same conversation, though still below the cumulative capital Microsoft has routed into OpenAI through compute and equity combined.

What remains unverified

Three points should temper any reading of the report. First, the $7 billion figure traces to a single outlet. Second, "in talks" typically sits below a signed term sheet; deals at that stage fall apart routinely. Third, Decart's technology stack, customer base, and revenue are not disclosed, making valuation multiples impossible to compute.

What to watch

The first hard signal will be either a joint announcement, an Amazon 8-K filing, or a Decart funding disclosure that pre-dates the reported talks. Until then, the $7 billion number is a negotiating position, not a transaction price.