Sid Sheth, Founder, d-Matrix | theCUBE + NYSE Wired: AI Factories - Data Centers of the Future скачать в хорошем качестве

Sid Sheth, Founder, d-Matrix | theCUBE + NYSE Wired: AI Factories - Data Centers of the Future

In this theCUBE + NYSE Wired: AI Factories – Data Centers of the Future segment, theCUBE’s Dave Vellante sits down with Sid Sheth, Founder, President and CEO of d-Matrix, to unpack why 2025 marks the “age of inference.” Sheth explains how the DeepSeek moment (January 2025) catalyzed a shift from training to deployment at scale – where efficiency, performance per watt and cost now define the winners. He breaks down how AI factories are becoming the decision layer of enterprise infrastructure, with inferencing demand projected to outpace training by 10–100x. The discussion explores a likely bifurcation: a few mega “training factories” chasing AGI alongside a much broader, distributed footprint of inference factories across hyperscalers, neoclouds, sovereign deployments and the edge – reshaping data centers as a cornerstone of digital strategy. Sheth details d-Matrix’s approach with the Corsair platform (purpose-built for inference) highlighting in-memory compute, chiplet architecture and floating-point numerics to boost perf-per-dollar, perf-per-watt and, critically, low-latency user experiences. He also shares why latency grows even more important as agentic systems scale machine-to-machine interactions. The conversation touches on JetStream, an I/O accelerator introduced to remove compute and I/O bottlenecks and a pragmatic go-to-market: d-Matrix accelerator cards slot into customers’ existing fleets (e.g., servers from Dell, HPE, Lenovo, Supermicro) to avoid rip-and-replace. Sheth clarifies the company’s “vertical integration” applies at the silicon level – stacking memory atop compute (3D “skyscraper” capacity) – while preserving customer optionality at the server and rack level. For enterprises, the takeaway is clear: AI-scale infrastructure is moving from capex-heavy experiments to ROI-driven operations where efficiency, speed and flexible integration turn AI factories into the new unit of value. Find more SiliconANGLE news and analysis https://siliconangle.com/ Follow theCUBE's wall-to-wall event coverage https://siliconangle.com/events/ Learn about the latest theCUBE events https://www.thecube.net/ 00:00 - Embracing the Age of Inference: Transitioning from AI Training to Market Impact 02:59 - Building the Foundations: Powering AI with Infrastructure and Computing 05:58 - The Role of d-Matrix in AI Inference 08:23 - Key Metrics in AI Inference Performance 11:44 - d-Matrix's Technological Approach 14:12 - AI Inference as a Decision-Making Process 16:52 - Strategic Infrastructure: Balancing Integration and Future Growth #theCUBE #NYSEWired #theCUBEresearch #AIFactoriesDataCenters #dMatrix #Inference #DataCenters