NVIDIA (NVDA) Goldman Sachs Communacopia + Technology Conference 2024 summary

Evolution and strategy

• Accelerated computing was envisioned as a complement to general-purpose computing, initially targeting computer graphics and expanding into scientific computing and other domains.

• Maintaining architectural compatibility and a large installed base has been a core priority, enabling software investment protection and ecosystem growth.

• Expansion into new markets required developing domain-specific libraries and convincing industries to adopt them, leading to a broad ecosystem across gaming, AI, robotics, and more.

• Accelerated computing now underpins major advances in data processing, machine learning, and generative AI, driving efficiency and new capabilities.

Data center transformation and AI revolution

• General-purpose computing performance growth has slowed, ending the era of rapid CPU improvements and leading to computation inflation.

• Modernization of $1 trillion in data centers toward accelerated computing is underway, focusing on densification, energy efficiency, and cost savings.

• Generative AI is creating new digital skills, expanding beyond traditional IT into areas like digital employees and autonomous systems.

• Accelerated computing delivers significant ROI, with examples like Spark workloads achieving up to 20x speedups and 10x cost savings.

Innovation, competitive advantages, and supply chain

• AI infrastructure is more than chips; it requires integrated systems, bespoke software, and optimized architectures for performance and efficiency.

• The installed base and architectural consistency across clouds and devices create a strong competitive moat.

• Rapid innovation cycles are achieved by developing multiple chips and systems in parallel, enabling frequent performance leaps and customer value.

• Supply chain agility, especially in Asia and with TSMC, supports scaling production, with contingency plans for redundancy and IP flexibility.