On March 18, 2024, NVIDIA introduced the Blackwell GPU. Boasting 208 billion transistors, NVIDIA’s latest creation processes trillion-parameter AI models 30X faster than previous technologies, all at reduced power.  

This development prompts one key question: Has NVIDIA transcended Moore’s Law?  

Moore’s Law predicts that the number of transistors on a chip should double approximately every two years, maintaining a low cost. Traditionally, this doubling was achieved by miniaturising transistors.  

From 2012 to 2021 (Exhibit 1), the transistors and computational power growth generally aligned with Moore’s Law, doubling roughly every two years. Regardless, NVIDIA’s Blackwell GPU has shattered expectations, increasing computational speed by over a thousand times in just 8 years, vastly outpacing the gradual growth predicted by Moore’s Law.

Blackwell: Revolutionizing AI Processing

NVIDIA’s Blackwell platform marks a significant leap in AI technology, particularly in handling generative AI at an unprecedented scale. Its B200 and GB200 models demonstrate remarkable efficiency and power, significantly reducing resources and time needed for large-scale AI model training, like GPT-4. Compared to competitors like Intel’s Ponte Vecchio and AMD’s MI300 series, Blackwell stands out for its superior performance and efficiency, highlighting NVIDIA’s leadership in meeting the demands of advanced AI processing. ‍

The Journey of Chipset Development

Developing chipsets is a complex challenge, constrained by technological and raw material limitations. Overcoming these challenges requires sustained innovation across four critical areas:

1. ‍Manufacturing Technology: Pushing the boundaries of fabrication technology to create ever-smaller and more efficient circuits. ‍
2. Material Science: Identifying and utilising new materials that enhance the performance and efficiency of semiconductors while ensuring sustainability of production. ‍
3. Circuit Design: Refining the design of chip components to boost performance, efficiency, and density. ‍
4. Packaging Technology: Advancing the development of materials used to house chips and integration of material components into cohesive modules that meet safety, economic and performance criteria.

Strategic Priorities for Leaders

In light of NVIDIA’s advancements, leaders must proactively engage in several strategic activities:

1. ‍Strategic Partnerships: Collaborating with industry leaders like NVIDIA can provide access to cutting-edge technologies and insights. For example,  Indosat Ooredoo Hutchison (IOH) is looking to integrate the NVIDIA Blackwell Platform into its AI-Cloud Infrastructure. ‍
2. Investing in Advanced Technology: Committing resources to adopt and integrate the latest technologies, like the Blackwell chipset, ensures competitive advantage and harnesses the benefits of improved performance. ‍
3. Market Adaptation: Keeping pace with rapid technological advancements that enables development of new products and solutions to meet evolving market demands is crucial for maintaining relevance and competitiveness in the industry.

In summary, as the tech landscape continues to evolve rapidly, driven by advancements like NVIDIA’s Blackwell, organisations must remain agile, innovative, and forward-thinking to thrive.

References

1. Nvidianews.com. (2024, March 26). Nvidia News Blackwell Platform  
2. Our World in Data (2023, March 23).  What is Moore’s Law? https://ourworldindata.org/moores-law
3. Forbes.com (2022, October 7). The True Nature of Moore’s Law