NVIDIA’s new AI chip just changed the game! Here’s why Blackwell is the future of computing 🚀

In his highly anticipated keynote at COMPUTEX 2024, NVIDIA CEO Jensen Huang introduced groundbreaking advancements that promise to redefine the landscape of artificial intelligence (AI) and computing. Highlighting the historical shifts in computing over the past six decades, Huang emphasized that we are on the brink of another transformative period. This keynote revolved around NVIDIA’s new AI chip, the Blackwell platform, and its upcoming iterations, showcasing the company’s relentless pursuit of pushing technological boundaries.

NVIDIA’s new products include the DGX Blackwell with eight GPUs and an air-cooled system, and the MGX modular system, supporting both x86 and liquid cooling. Huang highlighted the advancements in GPU performance, power efficiency, and cost reduction. He also introduced Spectrum X800, a network solution designed for AI, promising high-speed connectivity for tens of thousands of GPUs.

Huang talked about the future of AI, emphasizing the role of generative AI and the upcoming physical AI era, involving robotic systems for factories, warehouses, and humanoid robots. NVIDIA’s comprehensive approach includes integrating AI, robotics, and simulation within the Omniverse platform.

The Evolution from Hopper to Blackwell

Huang started by acknowledging the success of the Hopper platform, which has been a monumental data center processor. However, he swiftly shifted focus to the new Blackwell platform, a comprehensive system encompassing CPUs, GPUs, NVLink, NICs, and switches. Each generation of NVIDIA’s platforms, he explained, is not merely about advancing the GPU but about building and integrating an entire platform into an AI factory supercomputer, which is then disaggregated and sold in parts.

Blackwell and Beyond

Blackwell represents a significant leap in performance and efficiency. Huang revealed that next year will see the launch of Blackwell Ultra, continuing NVIDIA’s tradition of annual technological advancements. Following Blackwell, the Rubin platform and its subsequent Ultra version are already in full development. These platforms are architecturally compatible and push every technological limit, from TSMC process technology to memory and optics technology.

Unveiling the Blackwell Platform

Huang proudly presented the Blackwell platform’s production board, describing it as the most complex and highest performance computer ever made. The Blackwell platform features the largest chip globally, with two Blackwell dies connected via a 10 terabyte per second link, delivering unparalleled performance. The AI flops have increased a thousandfold over eight years, and the energy efficiency has improved dramatically.

DGX and MGX Systems

The keynote also introduced the DGX Blackwell system, an air-cooled powerhouse with eight GPUs designed for integration into existing infrastructures. For those needing liquid cooling, the MGX system offers a modular solution with four Blackwell chips connected to create a giant 72-GPU unit. This setup increases bandwidth by 18 times and AI flops by 45 times while maintaining a power efficiency ten times greater than previous generations.

Advancements in Networking: Spectrum X

Huang addressed the critical role of networking in AI factories, introducing the Spectrum X800 Ethernet switch. Designed for tens of thousands of GPUs, Spectrum X800 Ultra and X1600 are set to follow, supporting hundreds of thousands to millions of GPUs. This innovation brings the capabilities of InfiniBand to Ethernet architecture, essential for deep learning and AI factories.

Addressing Ethernet Challenges

NVIDIA’s advancements include world-class RDMA for Ethernet, congestion control, adaptive routing, and noise isolation. These features ensure optimal performance by reducing training times and improving efficiency, making Ethernet a viable solution for AI data centers. The upcoming Spectrum X800 Ultra and X1600 switches are poised to handle the increasing demands of AI workloads, heralding a new era of data center scalability.

The Future of AI: Physical AI and Robotics

Huang’s vision extends beyond current advancements, forecasting the rise of physical AI—robots with cognitive capabilities and an understanding of physical laws. NVIDIA is developing platforms for various robotic systems, including robotic factories, warehouses, manipulators, mobile robots, and humanoids. These platforms integrate computers, acceleration libraries, and pre-trained models, all tested and trained in NVIDIA’s Omniverse.

Robotics and Industrial Automation

Highlighting partnerships with companies like Siemens, Huang detailed NVIDIA’s role in transforming industrial automation. Robotic factories will orchestrate robots to build robotic products, a future where NVIDIA’s platforms will be integral. Self-driving cars and humanoid robots are identified as high-volume robotics products that will benefit from these advancements.

Insights

1. NVIDIA’s Blackwell platform marks a significant advancement in AI chip technology.
2. The integration of various components into a single platform enhances AI factory capabilities.
3. Future AI developments will focus on generative AI and physical AI, including robotics.
4. NVIDIA’s Spectrum X800 addresses networking needs for large-scale AI operations.

The Essence (80/20)The Origins and Evolution of the 80/20 Principle The Discovery by Vilfredo Pareto In 1897, Italian economist Vilfredo Pareto uncovered a striking pattern in his study of wealth and... More

• Core Topics:
  • Blackwell Platform: Revolutionary AI chip technology with integrated components for AI supercomputers.
  • Technological Push: Advancements in processing, packaging, memory, and networking technologies.
  • DGX and MGX Systems: High-performance, scalable AI systems with air-cooled and liquid-cooled options.
  • Spectrum X800: Networking solution designed for efficient AI data center operations.
  • Future AI Directions: Emphasis on generative AI and the development of physical AI for robotics in various industries.
• Detailed Descriptions:
  • Blackwell Platform: Combines CPU, GPU, NVLink, NIC, and switch, designed for AI factory supercomputers, offering high performance and efficiency.
  • Technological Push: Focus on maximizing the capabilities of various technologies to enhance AI performance and reduce costs.
  • DGX and MGX Systems: Provide robust AI computing power with innovative cooling solutions and modular design for scalability.
  • Spectrum X800: Ensures high-speed, efficient connectivity in AI data centers, supporting tens of thousands of GPUs.
  • Future AI Directions: NVIDIA’s roadmap includes advancements in generative AI for interactive applications and physical AI for robotic systems.

Action Plan For Data Centers

1. Adopt Blackwell Platform: Integrate NVIDIA’s Blackwell platform into data center operations for enhanced AI capabilities.
2. Leverage Technological Advancements: Utilize the latest advancements in processing, packaging, and memory technologies to improve AI performance.
3. Implement DGX and MGX Systems: Deploy these systems to achieve high-performance AI computing with scalable and efficient cooling solutions.
4. Upgrade Networking Infrastructure: Transition to Spectrum X800 for improved AI data center connectivity and efficiency.
5. Prepare for Future AI Trends: Invest in generative AI technologies and develop strategies for implementing physical AI in robotic systems.

Blind Spot

The potential challenges and costs associated with transitioning to the new Blackwell platform and integrating it with existing infrastructures may be underestimated. Ensuring compatibility and managing the transition smoothly will be crucial.

Nvidia (NVDA) Technical Analysis

Price Trend and Moving Averages:

• The stock price has been in a strong upward trend, especially evident in the recent months.
• The 50-day moving average (blue line) is at 98.97, and the 200-day moving average (red line) is at 68.82. Both moving averages are trending upwards, indicating a long-term bullish trend.
• The current stock price of 131.88 is significantly above both moving averages, reinforcing the bullish outlook.

Volume:

• Volume is relatively stable, with occasional spikes. Higher volume on up days compared to down days suggests accumulation and strong buying interest.

Relative Strength IndexIn the world of technical analysis, the Relative Strength Index (RSI) stands as a cornerstone tool for traders seeking insights into market momentum. Developed by J. Welles Wilder ... More (RSI):

• The RSI value is 80.81, which is above the overbought threshold of 70. This suggests that the stock might be overbought in the short term and could be due for a pullback or consolidation.

On Balance VolumeThe On Balance Volume indicator (OBV) is a technical analysis tool used to measure the flow of money into and out of a security over a specified period of time. It is a cumulative ... More (OBV):

• The OBV line is trending upwards, indicating that volume is higher on up days and lower on down days. This confirms the bullish trend and suggests that institutional money might be flowing into the stock.

Stochastic RSIIn the realm of technical analysis, the Stochastic RSI (StochRSI) emerges as a powerful tool for traders seeking to navigate market dynamics with precision. Developed by Tushar S. ... More:

• The Stochastic RSI is at 1.000, which is the maximum value, indicating extreme overbought conditions. This further supports the potential for a short-term pullback or consolidation.

Average Directional IndexThe Average Directional Index (ADX) stands as a cornerstone indicator in the toolkit of technical traders, offering insights into the strength of market trends. Developed by Welles... More (ADX):

• The ADX value is 43.44, which is above 25, indicating a strong trend. A rising ADX suggests that the strength of the current trend (upward) is increasing.

Chaikin OscillatorNamed after its creator Marc Chaikin, the Chaikin Oscillator stands as a formidable tool in the arsenal of technical analysts. This oscillator is designed to measure the accumulati... More:

• The Chaikin Oscillator is at 561.13M, which is positive, indicating accumulation and suggesting that the buying pressure is greater than the selling pressure.

Time-Frame Signals:

• 3-Month Horizon: Hold. While the long-term trend is bullish, the overbought conditions indicated by the RSI and Stochastic RSI suggest a potential short-term pullback. It would be prudent to wait for a better entry point after some consolidation.
• 6-Month Horizon: Buy. The strong upward trend, supported by moving averages and the OBV, indicates a continued bullish outlook. Any short-term pullback could provide a buying opportunity.
• 12-Month Horizon: Buy. The overall trend is strongly bullish, and the ADX suggests that the strength of this trend is increasing. Long-term investors could benefit from holding or adding to their positions on dips.

In summary, Nvidia Corp shows strong bullish characteristics in the long term, but short-term overbought indicators suggest a possible pullback. Long-term investors might consider buying on dips, while short-term traders should be cautious of a potential consolidation.

Remember, past performance is not an indication of future results. Always conduct your own research and consider consulting with a financial advisor before making any investment decisions. 🧡

Final Thoughts

Jensen Huang’s keynote at COMPUTEX 2024 encapsulated NVIDIA’s commitment to driving the next wave of computing transformations. With groundbreaking innovations in AI chips, comprehensive platform integration, advanced networking solutions, and a visionary approach to robotics, NVIDIA is poised to lead the industry into an era where AI and physical AI become ubiquitous. The introduction of Blackwell and the promise of future platforms like Rubin set the stage for unparalleled advancements in computing, ensuring that NVIDIA remains at the forefront of technological evolution.