What is Deep Learning?:
Deep learning is a breakthrough in machine learning that involves self-writing software and self-learning computers. It is considered almost magical because it can find important features from a large number of examples without explicit instructions.

Feature Detection through Convolutional Neural Networks:
Deep learning utilizes convolutional neural networks to detect features, starting from low-level features and gradually building higher-level abstract features. This allows the network to identify intricate details and even differentiate between species of cats.

Generalization and Robustness:
Deep learning models are robust to changes and can generalize learned knowledge. Even if the color of an object is altered, the model can still accurately identify it, demonstrating its ability to recognize subtle details while being difficult to deceive.

Applications of Deep Learning:
Deep learning is not limited to computer vision; it has been applied to various other domains. It is particularly effective in analyzing messy data when traditional software development approaches may be impractical. By training on numerous examples, deep learning can automatically determine features and make predictions, classifications, and recommendations based on the data.

Rise of Deep Learning and Challenges in Computing
Jensen Huang highlights the growing exploration of artificial intelligence capabilities using deep learning. The stagnant performance of CPUs due to limitations in transistor performance and architectural advancements has created a demand for more computational capabilities.

GPU-Accelerated Computing: A Time Machine
NVIDIA pioneered GPU-accelerated computing to intercept and offload key kernels of programs, enabling significant acceleration. The speedups achieved by GPUs are comparable to decades of progress under Moore’s Law. This acceleration has made it possible to solve problems that would have otherwise taken much longer, effectively acting as a time machine.

Unsolvable Problems Become Solvable
The tipping point in GPU adoption has led to widespread application in various fields. Previously unsolvable problems in sectors such as transportation, healthcare, finance, enterprise, farming, retail, and consumer applications are now being addressed through GPU acceleration.

Opportunities for Startups in Israel
Israel, known as “Startup Nation,” is well-positioned to leverage these dramatic changes in computing. Startups thrive on discontinuities and embrace new technologies that challenge the status quo. NVIDIA is building a computing platform and tools for researchers, developers, and entrepreneurs seeking innovative solutions.

Collaboration with NVIDIA
NVIDIA supports Israeli startups by providing a platform and opportunities for global exposure. Collaboration with NVIDIA enables startups to address industry-critical issues and access valuable resources.

Holodeck: Virtual Reality for Complex Creation
NVIDIA’s Holodeck platform allows users to immerse themselves in virtual reality for complex creation and collaboration. Prototyping and collaboration can occur virtually, regardless of physical location or scale of the project.

AI Platform and TensorRT for Deep Learning
NVIDIA’s AI platform empowers deep learning researchers worldwide. TensorRT optimizes and tunes deep learning networks, enabling faster performance and reduced costs.

Cost Reduction and Scalability
The speedup achieved by TensorRT translates to significant cost reductions for startups using cloud services. Instead of spending pennies per transaction, startups can potentially reduce costs to a hundredth of a penny per transaction when using GPU acceleration.

Introduction:
Jensen Huang presented at an AI event, discussing the topics of cloud-based services, autonomous machines, virtual reality, and the future of consumer graphics cards.

Cloud-Based Services:
Huang introduced NVIDIA’s platform for cloud-based services, enabling businesses to reduce costs and operate their services from any cloud provider worldwide.

Autonomous Machines:
Huang described autonomous machines, starting with self-driving cars. He emphasized the importance of teaching these machines to avoid collisions and interact effectively with the environment.

Virtual Reality:
Huang introduced Holodeck and Isaac, NVIDIA’s platforms for creating virtual reality environments. Holodeck allows users to design and interact with virtual objects in a holographic display. Isaac is a virtual robot that can learn from user interactions and eventually be transferred to a real-world robot.

Consumer Graphics Cards:
When asked about the future of consumer graphics cards, Huang hinted at the possibility of moving beyond the current numbering trend. He remained secretive about the specific plans for the next generation of cards.

NVIDIA’s Strengths:
Huang highlighted NVIDIA’s status as a computer graphics leader, dedicating significant resources to R&D and producing the most complex processors in the industry.

Virtual Reality Goals:
Huang expressed NVIDIA’s ambition to create immersive shared virtual reality worlds, requiring advancements in physics simulation, rendering, lighting, and material reflections.

Expansion in Israel:
NVIDIA’s presence in Israel is currently limited, but Huang acknowledged the country’s exceptional talent and innovation culture. He expressed excitement about exploring future growth opportunities in Israel.

NVIDIA’s Growth in Israel:
NVIDIA plans to expand its presence in Israel, a hub for technology and engineering talent, by increasing its workforce from 20 to a larger team. The high concentration of GPU developers in Israel, particularly NVIDIA GPU developers, makes it an ideal location for NVIDIA’s growth. Israel’s ecosystem fosters a passion for technology and entrepreneurial spirit, aligning with NVIDIA’s goals for innovation.

Autonomous Vehicles and NVIDIA’s Role:
NVIDIA has been involved in the automotive industry for 10 years, initially focusing on infotainment and now shifting towards artificial intelligence and autonomous driving. The company aims to reshape infotainment by transforming it into an AI-powered agent to enhance the user experience. The automotive industry’s cautious approach to innovation is due to the significant impact of new technologies on society and the need for meticulous engineering. NVIDIA emphasizes the importance of balancing traditional engineering practices with the integration of groundbreaking AI technology to revolutionize vehicles. Autonomous vehicles are expected to hit the roads sooner than initially anticipated, with level 5 taxis projected to emerge in 2020 and level 4 autonomous vehicles by 2021-2022.

NVIDIA’s Market Share in Autonomous Vehicles:
NVIDIA anticipates a substantial market share in the autonomous vehicle industry due to its focus on level 4 and level 5 cars and its dedicated efforts in this field for almost a decade. The company believes that the market is vast, encompassing various types of vehicles and applications, and it’s too early to focus on specific market share percentages. NVIDIA prioritizes developing and refining the technology, ensuring performance, safety, and assisting partners in bringing autonomous vehicles to the market.

Competition with Mobileye and Intel:
NVIDIA acknowledges the success of Mobileye and Intel in the autonomous vehicle space. It recognizes Mobileye’s expertise in ADAS technology and anticipates strong competition from the company in the autonomous vehicle domain. Intel’s capabilities and determination in pursuing new ventures are acknowledged, and NVIDIA views this as a competitive challenge. NVIDIA emphasizes its unique strengths in creating an open platform with a fully driveable stack, positioning it well in the autonomous vehicle landscape.

Coexistence of Drive AX and Drive IV Technologies:
Drive AX enhances and augments the driver’s capabilities, while Drive IV enables fully autonomous driving. Autonomous driving capabilities may be limited in certain situations, such as adverse weather conditions, where the car opts for manual driving mode. Even in manual driving mode, AI remains functional, monitoring the surroundings to ensure the safety of the driver and passengers.

NVIDIA’s Vision for AI-Powered Cars:
Jensen Huang emphasizes the potential of AI to enhance cars’ capabilities, including driver assistance, passenger comfort, and entertainment. Huang envisions a future where cars become fully autonomous, leveraging AI for navigation, safety, and interaction with passengers.

Challenges in Automotive AI Adoption:
Industry experts discuss the challenges in implementing AI in the automotive industry, particularly the lack of standardization and the uniqueness of each car platform. Huang highlights the slower pace of development due to the need to adapt AI solutions to different car models and architectures.

NVIDIA’s Approach to Domain-Specific AI:
Huang explains NVIDIA’s strategy of focusing on domain-specific AI rather than general-purpose AI. He emphasizes the benefits of domain-specific AI, including the ability to optimize for specific tasks and the cost-effectiveness achieved through economies of scale.

AI and the Potential for Gaming:
Huang confirms the possibility of using AI for gaming, envisioning a future where players can interact in virtual reality environments with friends worldwide.

Ethical Considerations in AI:
Huang acknowledges the potential for AI to be used for both good and evil, citing examples such as fire, chemistry, and the internet. He emphasizes the need for open dialogue and awareness to prevent dystopian outcomes associated with AI. Huang stresses that AI should remain under human control and suggests that it is more likely to be misused by evil individuals than becoming inherently evil itself. NVIDIA’s strategy to mitigate risks includes keeping the computing platform open, supporting multiple frameworks and cloud providers, and promoting collaboration and competition in AI development.

AI Revolution:
Jensen Huang describes AI as the most significant computer science revolution since its inception. He envisions computers capable of writing their own software and learning from experiments, surpassing human capabilities.

Unsolvable Problems Solved:
AI enables the solution of historically unsolvable problems, such as self-driving cars, early cancer detection, and optimizing crop productivity.

Electric and Autonomous Vehicles (EVs and AVs):
Huang predicts a surge in car sales in the next 10 to 15 years due to the transition to EVs and AVs. He believes that autonomous electric vehicles will be highly enjoyable, leading to widespread adoption.

Long-term Transportation Vision:
Huang hopes for a future with fewer cars on the road, reducing parking congestion and capital underutilization. He envisions shared transportation models, including ride-hailing, ride-sharing, and improved public transportation, as beneficial for society.

Personal Excitement:
Huang expresses eagerness to purchase the Porsche Mission E, an upcoming electric sports car, highlighting his personal interest in innovative vehicles.

Collaboration with Startups:
NVIDIA seeks to actively invest in promising startups that align with its mission of creating groundbreaking ideas. The company’s goal is to accelerate the growth of these startups and bring their products to market faster. NVIDIA offers unique assistance through its technology platform, technical capabilities, and marketing exposure. Startups are encouraged to reach out to NVIDIA for potential collaboration opportunities.

Mellanox Collaboration:
NVIDIA has a special partnership with Mellanox to address the challenges of data transfer for high-speed computation. The collaboration resulted in the development of GPU direct, a software and architecture solution that allows data from Mellanox NICs to directly enter NVIDIA GPUs. GPU direct optimizes data transfer and enhances the performance of NVIDIA’s GPU computing platform.

Robot Taxis and Autonomous Cars:
NVIDIA expects robot taxis to operate relatively soon in geo-fenced areas with complete confidence in their maps and surroundings. These areas could be large cities like San Francisco, New York City, or Phoenix, which have been extensively mapped digitally. Robot taxis are equipped with numerous sensors that record everything, providing valuable data for accident analysis and fault determination. Determining the regions where these technologies will emerge first is challenging, but the benefits of robot taxis are most significant in urban environments with a high demand for taxi services.

—

Autonomous Cars and Lane Markings:
Lane markings or digital virtual lanes in the car’s HD map make a significant difference in autonomous car navigation.

AI in Education:
AI offers benefits in education by expanding quality education to a broader population. It addresses the shortage of teachers and limited time, allowing for personalized learning and expanding access to education. While AI can enhance education, it cannot replace gifted professors or the passion of dedicated teachers.

Jensen Huang’s Car Collection:
He does not own any collectible cars. His preference is for simple living and avoiding unnecessary possessions. His daily car is a Tesla.

NVIDIA’s Relationship with Tesla:
NVIDIA’s technology is used by Tesla for training, deep learning, and in their autonomous cars. Tesla’s current models, including P100s, Model X, and Model 3, all utilize NVIDIA’s technology. Rumors suggest that Tesla may switch to their own ARM-based chips, but NVIDIA remains supportive and committed to helping them build cars.

Competition between NVIDIA’s Tesla Processor and AMD’s Vega Processor:
Tesla and Vega serve different markets. Tesla is designed for high-performance computing, with specialized testing, reliability technology, software stack, and developer support. Tesla’s popularity and extensive developer base make it a valuable platform for high-performance computing.