Amazon in talks to acquire Decart for $7B

The latest: talks are active, but no deal is signed

Amazon is in negotiations to acquire Decart, the Israeli-founded artificial intelligence startup, in a transaction that could be valued at around $7 billion, according to reports citing people familiar with the matter [1]. The talks could produce an agreement soon, though the same reports stress that they could still collapse or be disrupted by another bidder [1]. Reuters, which carried the report through multiple outlets, said it could not independently verify the Wall Street Journal’s account, and reported that Amazon and Decart did not provide immediate comment [2].

That combination matters. This is not a closed acquisition, not a regulatory filing, and not yet a formal announcement from Amazon. It is a live negotiation around a company that has rapidly become valuable because it sits at the intersection of three of the AI market’s most expensive problems: compute scarcity, chip dependence and synthetic environments for training advanced systems.

At the reported price, Decart would be far more than a small talent acquisition. The startup was valued at nearly $4 billion in May, according to PitchBook data cited in the reports, after raising more than $400 million from investors including Sequoia Capital, Benchmark and Radical Ventures [3]. Reuters also reported that Decart raised $300 million in May in a round led by Radical Ventures, with Nvidia joining as a new investor [2]. A $7 billion sale would therefore mark a sharp step-up from that recent private valuation and would reset expectations for founders building deeply technical AI infrastructure companies.

Why Amazon would want Decart

The strategic logic starts with chips. Decart has developed software that makes it easier for developers to run AI models across different processors, a capability that has become more valuable as companies compete for compute capacity [1]. Seoul Economic Daily, summarizing the reported talks, described Decart’s chip-portability work as a way to reduce the burden of adapting AI models that were originally built and tuned for specific chips [4].

For Amazon, that is directly connected to AWS. Amazon has been pushing its own Trainium AI chips as an alternative to Nvidia’s dominant GPUs, and reports say AWS executives were impressed after meeting Decart’s founders and gave the startup early access to Trainium3 in 2025 [4]. If Decart’s software helps models move more easily across chips, it could make Amazon’s in-house silicon more attractive to customers that do not want to be locked into one supplier or one hardware stack.

This is the central point of the proposed acquisition: Decart is not just a model company. It appears to be a portability layer for an AI industry that is hitting bottlenecks in hardware supply, training cost and deployment flexibility. In a market where access to Nvidia GPUs can dictate product road maps, any software that helps large customers shift workloads to alternative chips becomes strategically important. For Amazon, owning such a layer would support the broader AWS objective of making Trainium a credible option for advanced AI training and inference.

Decart is also a world-model company

Decart’s appeal is not limited to chip abstraction. The company is also associated with two world-model products, Lucy and Oasis, that broaden the story from infrastructure into AI-generated environments. Reuters reported that Lucy, backed by Nvidia, can edit live video in real time, while Oasis generates simulated environments for training and testing robotics and autonomous-driving systems [2]. The Straits Times, also carrying the Reuters report, said the same: Lucy is focused on live video editing, while Oasis is designed for robotics and self-driving test environments [5].

That matters because world models are becoming a major frontier in AI. Large language models predict text; world models attempt to learn and generate representations of physical or visual environments. If they work well, they can create synthetic spaces where robots, vehicles, games and creative tools can be trained or tested without depending only on real-world data collection. For a cloud provider, those workloads are compute-hungry, commercially valuable and strategically sticky.

Amazon already touches several markets where this could matter: cloud infrastructure, consumer devices, logistics, advertising, entertainment and robotics. A world-model startup could therefore be useful across AWS services, enterprise AI products, internal automation and possibly future Alexa or device experiences. That does not mean Amazon has publicly laid out such a plan; it means Decart’s product direction fits several Amazon priorities at once.

The valuation tells founders something important

The reported $7 billion price is remarkable because Decart was founded in 2023 by Israeli engineers Dean and Orian Leitersdorf, who are brothers, and Moshe Shalev [3]. A company that young reaching this scale of acquisition interest shows how AI infrastructure assets are being priced: not only on revenue, but on scarcity, technical differentiation and strategic relevance to hyperscalers.

If the transaction happens, it would also sit among Amazon’s largest acquisitions in recent years. PYMNTS noted that a Decart deal at roughly $7 billion would surpass Amazon’s $6.5 billion purchase of MGM in 2022 [6]. That comparison is useful because MGM was a media-library and streaming strategy acquisition; Decart would be a bet on core AI infrastructure and talent. The direction of spending tells the story: the biggest platforms are buying capabilities that could shape the cost and control of future AI workloads.

The number also has a signaling effect for other founders. Startups that solve unglamorous but essential AI problems — deployment, chip utilization, synthetic data, model portability, training environments — may command strategic premiums if they reduce dependence on a dominant supplier or unlock a hyperscaler’s proprietary hardware. In that sense, Decart’s reported valuation is not only a Decart story. It is a pricing event for the AI infrastructure stack.

The Anthropic angle

The reports also put Amazon’s interest in a competitive context. Bloomberg Law reported that Anthropic, the maker of Claude and a company closely tied to Amazon through major cloud and investment relationships, decided in September not to acquire Decart after exploring a deal worth about $6 billion and conducting due diligence [3]. Reuters similarly reported that Anthropic had explored an acquisition in August before deciding not to pursue it [2].

That detail is important for two reasons. First, it suggests Decart had already attracted serious strategic attention before Amazon emerged as the potential buyer. Second, it shows how valuable this layer of AI infrastructure has become to companies building large models and deploying them at scale. Whether the buyer is a frontier-model lab, a cloud provider or another compute-heavy platform, the motivation is similar: reduce friction between models and hardware, and secure scarce technical talent before rivals do.

What could still derail the deal

There are several obvious risks. The first is negotiation risk: the current reports explicitly say a deal may not happen if talks break down or another bidder appears [1]. The second is valuation risk: moving from a roughly $4 billion private valuation in May to a potential $7 billion exit only months later creates pressure on the buyer to justify the premium [3]. The third is integration risk. AI startups often move quickly because of small teams, narrow focus and founder-led execution; large-company integration can slow that down.

Regulatory scrutiny is another variable, though the current reports have not described a formal antitrust review because there is no announced deal yet. If Amazon does sign an agreement, regulators could examine how the acquisition affects cloud competition, AI compute markets and access to chip-portability technology. The more Decart is viewed as infrastructure rather than an isolated product, the more strategically sensitive the deal could become.

The bottom line

As of October 12, 2026, the Decart story is best read as an active, high-stakes negotiation, not a completed takeover. The reported $7 billion price reflects how much value hyperscalers now place on software that can make AI models less dependent on any single chip platform and on teams that can build world models for simulation, video and robotics. For Amazon, Decart could strengthen AWS’s Trainium strategy and deepen its advanced-AI portfolio. For founders, it is another sign that the fastest route to a massive exit may be solving the infrastructure problems hidden underneath the AI boom.

Even Alexa, it seems, may need a very expensive upgrade patch.

Sources from the last 72 hours

1. [1]Report: Amazon in talks with AI startup Decart over possible $7B acquisitionOct 12, 2026, 2:31 AM
2. [2]Amazon in talks to buy AI startup Decart for around $7 billion, WSJ reportsOct 12, 2026, 4:05 AM
3. [3]Amazon Is in Talks to Buy AI Startup Decart, WSJ ReportsOct 11, 2026, 10:46 PM
4. [4]Amazon in Talks to Buy AI Chip-Portability Startup Decart for $7 BillionOct 12, 2026, 4:18 AM
5. [5]Amazon in talks to buy AI start-up Decart for around US$7 billion: ReportOct 12, 2026, 1:26 AM
6. [6]Amazon Weighs $7 Billion Purchase of AI Startup DecartOct 11, 2026, 2:00 AM

AI-generated article based on recent web research, then preserved as a dated editorial snapshot.