[Core Summary]

The Wall Street Journal reported on October 11 that Amazon is in advanced talks to acquire generative-AI startup Decart for around 7 billion dollars. Decart specializes in real-time long-video generation, low-latency streaming inference, and diffusion-model architectures optimized for consumer hardware. If completed, the deal would be AWS’s largest single move into the generative-video market and would reshape the competitive landscape currently dominated by Google DeepMind and OpenAI.

[Event Details]

  • Deal structure: Sources familiar with the negotiations say Amazon and Decart have held multiple rounds of talks on acquisition terms, with the headline valuation at roughly 7 billion dollars; a definitive agreement could be signed before year-end, though a deal collapse remains possible.
  • Decart’s business: Founded in 2024, Decart focuses on real-time long-video generation, low-latency streaming inference, and diffusion-model architectures optimized for consumer-grade hardware. Industry observers describe its core technology as a generational leap in video-generation efficiency.
  • Strategic intent: AWS aims to use Decart’s technology to strengthen its Bedrock generative-AI platform and the Nova foundation-model family, particularly to close gaps in video generation, video content understanding, and multimodal inference.
  • Financial backdrop: Decart had previously raised roughly 650 million dollars in venture funding and was valued at about 3 billion dollars in late 2025; the reported acquisition price more than doubles that mark in less than twelve months.
  • Competitive landscape: Both Google DeepMind and OpenAI are accelerating their proprietary video-generation models, and industry sources suggest Amazon’s urgency is driven by pressure on both foundation-model quality and cloud-services differentiation.
  • Regulatory angle: The deal’s scale and Decart’s market position in generative video may attract scrutiny from US and EU antitrust authorities, though no regulator has publicly commented.

[Full Perspective]

If completed, this acquisition would mark a strategic inflection point for hyperscale cloud providers in generative AI, shifting the playbook from in-house foundation-model development to a dual build-and-buy model. Over the past two years AWS has rolled out its own Titan and Nova model families and invested externally in leading AI labs such as Anthropic, but in the high-difficulty video-generation and streaming-inference segments, its in-house cadence has visibly lagged the iteration speed of Google Veo and OpenAI Sora. Buying Decart would effectively compress a 12-to-18-month R&D gap into a single transaction.

The deeper implication lies in how AI compute economics get redefined. Generative video requires dramatically more compute than text or image generation, with a single video inference typically consuming 50 to 100 times the GPU resources of a text inference. Decart’s core technical advantage is its co-designed model architecture and inference engine, which compresses video-generation cost to roughly one-tenth of conventional approaches. If AWS can scale that cost advantage across its Bedrock customer base, it would meaningfully lower the marginal cost of generative video across advertising, e-commerce, entertainment, and gaming, forcing competitors to absorb simultaneous pressure on price and performance.

From an industry-dynamics view, the deal may accelerate a layering of the generative-AI market: hyperscalers control the compute and foundation layers, mid-cap specialists become acquisition targets, and long-tail application developers build ecosystems on top of the platforms. Decart’s reported willingness to accept a buyout rather than pursue an independent listing reflects how, in the current funding environment, being absorbed by a hyperscaler has become an attractive exit for early-stage AI companies. This dynamic is likely to draw higher-quality targets into early discussions with the major platforms, lifting AI M&A valuations across the board.

[Multiple Perspectives]

  • Amazon: Has not publicly commented on the potential deal but emphasizes that AWS continues to invest in generative AI through a mix of in-house and external paths, with M&A as one of several options.
  • Decart management: Reportedly favors the Amazon offer, citing access to compute, customers, and distribution as accelerants, while acknowledging concern that the independent product roadmap will be wound down.
  • Wall Street analysts: Largely view the deal positively, arguing it strengthens AWS’s enterprise generative-AI positioning, though they caution that the 7-billion-dollar valuation requires a clearly articulated commercialization plan.
  • Competitors: Google DeepMind and OpenAI are reportedly preparing responses, likely accelerating video-model iteration, refining API performance, and offering developer-ecosystem subsidies to retain share.
  • Regulators and policymakers: Some US and EU lawmakers have raised concerns about hyperscaler consolidation in AI, calling for closer scrutiny of generative-AI M&A activity.
  • Open-source community: Expresses concern about whether Decart’s core models will remain open source after the deal, with worries that research outputs could be folded into AWS proprietary systems.

Editor: GoodInfo Global Newsroom