Amazon.com, Inc. (Nasdaq: AMZN) is reportedly in discussions to acquire artificial-intelligence startup Decart for around $7bn, a transaction that could give Amazon ownership of technology spanning low-latency AI infrastructure, real-time generative video and world models designed for physical AI. The Wall Street Journal first reported the talks on October 11, while Reuters said it had not independently verified the potential transaction.
No acquisition agreement has been announced, and the discussions could change, end without a deal or face competition from another bidder. The roughly $7bn figure should therefore be treated as a reported potential valuation rather than committed Amazon consideration, booked acquisition spending or a transaction scheduled to close.
Decart has expanded extraordinarily quickly. The company raised $300m in May in a round led by Radical Ventures, bringing total funding above $450m, while reporting around the financing placed its valuation near $4bn. A transaction near $7bn only months later would therefore represent a substantial increase in the price attached to a business that remains at an early stage of commercial development.
Why could Decart be strategically valuable to Amazon Web Services?
Decart operates at two levels of the AI stack that matter to Amazon. Its DOS platform is designed to optimise inference and training across AI hardware, while its Lucy and Oasis models target real-time immersive experiences and physical-world simulation.
The infrastructure layer is particularly interesting because Amazon is trying to make its internally designed Trainium processors a credible alternative to dominant third-party accelerators. A software company capable of moving demanding AI workloads efficiently across different chips can lower one barrier to customers adopting Amazon’s silicon.
Decart already has direct experience with Trainium. AWS says early testing of Decart’s real-time video models on Trainium produced up to four times higher frame throughput and twice the cost efficiency of the GPU systems used for the comparison, while reducing latency from 40 milliseconds to 10 milliseconds. Those are AWS and customer benchmark claims rather than universal performance guarantees, but they demonstrate a meaningful existing technical relationship.
Ownership could allow Amazon to integrate that knowledge more deeply into AWS hardware, Bedrock and future AI services. The risk is that acquiring a company currently designed to work across hardware platforms could make Decart appear less neutral to customers who want infrastructure independence.
What are Lucy and Oasis, and why are world models attracting so much capital?
Decart describes Lucy as a world model for immersive experiences and Oasis as a world model focused on physical AI. Lucy 2.5 can generate and transform video live at 30 frames per second, while Oasis is intended to create interactive simulated environments relevant to robotics and other systems operating in the physical world.
The strategic idea differs from conventional chatbots. A language model primarily predicts and produces sequences of text or other tokens, while a world model attempts to represent how environments behave over time and how objects respond to actions.
That capability could eventually become important in robotics, gaming, autonomous systems, design and simulation. Machines operating in physical environments need some ability to anticipate movement and consequences rather than merely produce a fluent textual response.
Commercial maturity remains much less certain than the technological ambition. Producing visually impressive demonstrations does not establish reliable robotics performance or prove that enterprises will pay enough for world-model services to justify multi-billion-dollar valuations.
A potential Amazon acquisition would therefore be a bet on where AI workloads are moving rather than simply the purchase of an established revenue stream.
Why does Decart’s $300m funding round make a reported $7bn price more notable?
Decart announced the $300m financing in May, bringing total capital raised above $450m. That round was associated with a valuation close to $4bn, meaning the reported Amazon discussions are occurring after an unusually rapid increase in the company’s implied strategic value.
Acquisition prices can exceed venture funding valuations because a buyer may value control, integration benefits and technology differently from minority investors. A strategic acquirer can also justify paying for capabilities that produce returns elsewhere in its organisation rather than only through the target’s standalone revenue.
Amazon has a particularly broad set of potential integration points. Decart technology could influence AWS compute, Trainium adoption, Bedrock services, consumer experiences, robotics or video-related products if a transaction were completed.
The danger is paying too early for optionality. Frontier AI technologies evolve quickly, making it difficult to know whether today’s architecture becomes a standard, is superseded by another approach or is reproduced internally by larger competitors.
That makes technical due diligence and employee retention as important as the reported purchase price. Much of the value in an early AI company resides in researchers, software and expertise rather than hard assets that can be resold if the strategic thesis changes.
How would a $7bn Decart deal fit Amazon’s enormous AI capital programme?
Amazon is already spending at a scale that makes even a multi-billion-dollar acquisition only one component of its broader AI strategy. Second-quarter 2026 net sales reached $200.6bn, while AWS revenue accelerated sharply as customers expanded cloud and artificial-intelligence workloads.
That growth is requiring extraordinary physical investment. Amazon has increased expected 2026 capital expenditure to approximately $220bn, with AI infrastructure accounting for a substantial portion of the increase, while trailing free cash flow has been pressured as property and equipment spending surged.
A Decart transaction would represent a different form of capital allocation. Data centres and processors expand physical capacity, whereas acquiring an AI research and software company purchases technology, talent and intellectual property intended to increase what Amazon can do with that infrastructure.
The two investments can reinforce one another if Decart makes Trainium or AWS services more useful. They can also compound execution risk if Amazon simultaneously spends heavily on hardware, frontier models, software companies and data-centre capacity before the resulting revenue streams mature.
That is why the reported $7bn value should be assessed against strategic integration rather than Amazon’s ability to finance the cheque. Amazon can afford a transaction of this size; the more difficult question is whether ownership produces more value than continuing to partner with Decart as an independent customer.
What must happen before this becomes an Amazon acquisition story rather than takeover speculation?
The first requirement is an actual agreement. Until Amazon or Decart announces signed terms, there is no confirmed acquisition price, structure, regulatory timetable or completion date.
Even an agreement would not guarantee closing. A transaction involving a strategically important AI company could attract regulatory review, while competing bids or negotiation over employee incentives and governance could still influence final terms.
Investors should also watch whether Decart remains closely tied to multiple chip platforms. Its ability to optimise AI workloads across hardware is potentially valuable precisely because the computing market is becoming more heterogeneous.
For now, the development is best understood as evidence of Amazon’s interest in moving deeper into real-time and physical AI. If the reported talks become a signed $7bn acquisition, the next question will not be whether Amazon bought another AI company, but whether Decart can make Amazon’s own computing infrastructure materially more competitive.
Discover more from Business-News-Today.com
Subscribe to get the latest posts sent to your email.
