Amazon is negotiating a roughly $7 billion acquisition of AI infrastructure startup Decart, signaling an intensifying strategic battle among tech giants for foundational compute software. Decart's core asset is its "Dynamic Operating System" (DOS) optimization stack, which claims to enable seamless switching across heterogeneous hardware like Nvidia, Google TPU, and Amazon Trainium, achieving up to 8x model inference speedups. This move directly supports AWS's goal of reducing per-token costs for its custom Trainium chips while strengthening its position in real-time world models and physical AI simulation.
Cross-verified sources indicate the proposed price significantly exceeds Decart's recent $4 billion valuation and surpasses Anthropic's earlier $6 billion offer. Notably, AWS granted Decart early access to Trainium3 as early as 2025, establishing deep technical compatibility. Conversely, Anthropic abandoned its bid in September due to due diligence concerns, highlighting Amazon's decisive execution in integrating vertical-specific software solutions.
This event profoundly impacts the AI industry by exposing "compute fragmentation" as a critical bottleneck for large-scale deployment. By acquiring cross-chip adaptation capabilities, Amazon aims to break vendor lock-in risks and build more resilient cloud-native AI infrastructure. This offers developers lower inference cost options and suggests that future AI competition will shift from mere hardware stacking to deep architectural co-optimization.