Key takeaways where NVidia’s accelerated computing platform is driving the next wave in AI, digital twins, cloud technologies and sustainable computing.

• The Blackwell Chip that has AWS, Google, Oracle and Microsoft gearing up for it.
• The future is Generative, moving from retrieval to ultimately have inference engines to analyze and make sense real time.
• AI Security: A trusted environment that encrypts data at rest, in transit and while being computed.
• AI Factory to generate intelligence for the next industrial revolution.
• Manufacture digitally first: utilize digital twin to build the AI factory using Omniverse to reduce time and build a perfect assembly plant in one take.
• AI Foundry, NeMo Retriever, Chat with PDF: LLM in a box to aggregate knowledge in a Vector Database and enable it to be available to the enterprise to extract their “goldmines” and convert them to copilots.
• Physical AI would need 3 computers: Watcher and Context creator, Onmiverse Compute engine and a Robotic Engine to do.

Attention is all you need (aka The Transformer Architecture), and Conversation, Reasoning and Communication is all you want! Welcome to the world of Generative AI.

Unveiled at the 2024 Spring GTC event, NVIDIA‘s Blackwell superchip and LATTE3D are compelling additions to its portfolio and will certainly be the tools for this next industrial revolution we are embarking on – one that is governed by tools running AI in every aspect of our society.

Grant Michel: As we continue to explore the possibilities of AI, it’s exciting to think about the potential when it’s combined with our own reasoning and logical abilities. The result? An unprecedented level of efficiency and productivity. In fact, I have a theory that communication and English logic classes in schools will become increasingly popular as we seek to optimize our creative output. Let’s embrace the power of technology and education to unlock our full potential. Link: https://www.creativebloq.com/news/nvidia-latte3d-text-to-3d-generative-ai

As aptly put by Grant Michel from Würth Additive Group with whom we are partnering on engineering our products and solutions, tomorrow’s world will be governed by:
1. prompt engineering using LLMs to enable conversations
2. simulation, computation and analytics to enable reasoning, and
3. inference to enable communication.

For this reason, we at Numorpho Cybernetic Systems (NUMO) are building our process engineering platform, Mantra M5 to have a mix architecture consisting of GPUs, CPUS and TPUs/NPUs to enable the 3 constructs of converse, compute and infer to seamlessly pull-push information to enable data engineering the future.

As Grant puts it “…. communication and logic classes in schools will become increasingly popular as we seek to optimize our creative output. Let’s embrace the power of technology and education to unlock our full potential.”

At Numorpho Cybernetic Systems (NUMO), we are fully committed to utilizing Nvidia’s Omniverse, Metropolis, it DGX platform, OpenUSD object languate (along with ChatUSD) and its bits, bytes and AI underpinnings to depict our World Model to enable simulation, emulation, interaction and calibration of physical entities using digital twins.

We are part of their Inception program and plan to utilize it for enablement of process engineering in our Mantra M5 platform to enable the activities of make, manage, move, market and maintain. It would include parametric modeling, generative design, smart manufacturing and actionable intelligence utilizing conversation (prompt engineering), computation and inference engines to convert meaning to motion.

The Physical AI system defined by Jensen matches with our Lacanian triad: Symbolic, Imaginary and Real to converse, simulate and enable real world scenarios that defines the core framework of Mantra M5 using #gpus #cpus #npus and #tpus.

Summarizing the industrial revolution, Jenson mentioned that how it all started was “water comes in” and “electricity comes out”. In the new industrial revolution AI factories are the

NITIN UCHIL – Founder, CEO & Technical Evangelist
nitin.uchil@numorpho.com

REFERENCES:

• Staying in Sync: NVIDIA Combines Digital Twins With Real-Time AI for Industrial Automation | NVIDIA Blog
• Ingesting massive, global-scale, real-world geospatial data into virtual worlds is fundamental to building physically accurate industrial metaverse applications and digital twins. These digital twins are critical to simulating real-world environments in architecture, engineering, & construction, smart cities, defense, autonomous driving, robotics, climate science, and more. Geospatial data and virtual worlds provide context, spatial coherence and relationships, and real-time monitoring capabilities. We’ll share how we built Cesium for Omniverse, an open-source extension for NVIDIA Omniverse that enables an accurate WGS84 globe, an efficient 3D Tiles streaming engine for massive 3D geospatial data and level of detail selection, and integration with Cesium ion for easy and optimized access to 3D tiling pipelines and global data. Learn how Omniverse’s OpenUSD and Fabric, combined with 3D Tiles and glTF, enable high-fidelity streaming and rendering for global-scale digital twins using open standards.

• Shehzan Mohammed

  Director of 3D Engineering & Ecosystems, Cesium

  Shehzan is the director of 3D engineering and ecosystems at Cesium, where he oversees product management and 3D engineering of Cesium’s products and runtime engines, including Cesium for Omniverse. He’s a contributor to 3D Tiles and glTF. Shehzan previously worked at ArrayFire on GPU computing and high-performance GPU libraries. Shehzan has been teaching GPU programming and architecture at the University of Pennsylvania for the past seven years.

• A Smart Robot Born in the Industrial Metaverse, Enabled by OpenUSD – In robotic applications, the arduous and costly process of optimizing the vision system for top-notch performance is a significant challenge. Siemens Italy Digital Industries’ pre-sales consulting team, known for its strong collaboration with local original equipment manufacturers, delves into cutting-edge robotic feeding systems, exemplified by ‘Supata,’ a groundbreaking solution tailored for the efficient handling of small-batch assembly and insertion tasks. The pivotal goal here is to augment the robot’s adaptability by transitioning from the traditional vision system to an artificial intelligence-driven one. This transformation empowers the robot to adeptly identify and adjust to various objects, facilitated by the powerful Omniverse framework. The journey commences with 3D models of objects, where the robot is visually trained to distinguish these diverse objects, including those synthetically generated via Isaac Sim Replicator.

• Cristian Sartori

  Country Business Segment Manager, Siemens

  After 10 years as product manager for networking devices, Cristian changed sectors and went to Siemens. He started as product manager and sales specialist of digital solutions for the manufacturing sector. Over the course of his career, he acquired skills in product market management and sales, and as a technology enthusiast he’s fascinated by the impact of edge computing and AI in industry. Now Cristian is the director of business segment automation systems for the Italian market.

• Vincenzo De paola

  AI Technical Sales Support, Siemens

  Vincenzo graduated from Politecnico di Milano with a master’s degree in automation and control engineering, then transitioned into industrial automation, working as a digital application engineer for over two years. Now enrolled as Ph.D. executive at Politecnico di Milano associate with the AIR LAB, he’s studying applicability and business values for the AI Italy project as a technical sales support professional.

• The Transforming AI Panel features the authors of “Attention Is All You Need,” the groundbreaking paper that introduced the transformer neural network architecture. Transformers have since dominated all areas of AI and revolutionized the industry. Join Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N. Gomez, Lukasz Kaiser, and Illia Polosukhin, hosted by NVIDIA founder and CEO Jensen Huang.