I listened to the entire NVIDIA press conference at CES, and there was a lot of valuable information. Jensen Huang's vision for the future will soon be felt by ordinary people. Here are the key points.
The automotive sector is actually part of NVIDIA's physical intelligence. At the conference, Huang announced that the era of autonomous driving has arrived and outlined a grand blueprint: the release of the fourth-generation Thor autonomous driving computing platform, with computing power increasing 20 times compared to the previous generation;
In the fiscal year 2026, NVIDIA expects to expand its automotive business scale to $5 billion, launching the world foundational model Cosmos, which provides low-cost data generation solutions for autonomous driving and robotics development, comprehensively accelerating the intelligent transformation of automobiles, which will have a profound impact on industry patterns, technological trends, and the mobility ecosystem.
<h2>Part 1 NVIDIA's New Layout in the Automotive Field</h2>
● Technical Breakthroughs of the Thor Autonomous Driving Platform
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/549255f8-aaf9-470e-90a3-7f1eef008295.png?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/549255f8-aaf9-470e-90a3-7f1eef008295.png"/>
We have previously written &#34;Technical Analysis | How Advanced is NVIDIA's Thor Chip?&#34; The fourth-generation Thor autonomous driving computing platform released by NVIDIA this time represents the core changes in the iteration of intelligent driving technology. The full-powered Thor has 20 times the computing power of the previous Orin platform, which is not just a simple performance iteration but a comprehensive upgrade of functionality.
With Thor, NVIDIA provides powerful real-time computing capabilities for in-vehicle computers, supporting multi-sensor fusion (including cameras, radar, lidar, etc.), achieving autonomous driving capabilities at L4 and even L5 levels.
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/9e5d5511-d74e-4028-bbc5-1fdd04f8eac8.png?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/9e5d5511-d74e-4028-bbc5-1fdd04f8eac8.png"/>
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/1200be9c-8f50-481c-a263-d9c5365d1871.png?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/1200be9c-8f50-481c-a263-d9c5365d1871.png"/>
The breakthrough performance of Thor enables it to support multiple functional integrations, including autonomous driving, cockpit experience, and vehicle networking, providing automakers with a one-stop solution. The scalability of Thor means it is not limited to the automotive field but can also be widely applied in robotics and other scenarios, further extending its market potential.
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/40ed54e0-5dfb-444d-bf39-ca4ddb9479f2.png?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/40ed54e0-5dfb-444d-bf39-ca4ddb9479f2.png"/>
NVIDIA's approach to autonomous driving iteration is pioneering, focusing on the integration of virtual and real, leveraging real-world data for feedback.
As vehicles equipped with NVIDIA chips hit the road, they collect various driving data, from daily commuting in urban traffic to long-distance highway driving. This first-hand data becomes a &#34;treasure trove&#34; for optimizing autonomous driving algorithms, utilizing virtual platforms like Omniverse
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/bbebd197-80e4-40df-ae71-a6ba446fcc9c.png?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/bbebd197-80e4-40df-ae71-a6ba446fcc9c.png"/>
Based on Blackwell chip computing power and software services, the automotive manufacturing production line is reconstructed in the virtual world, and new virtual scenes can be generated based on real data for training. This is akin to creating a &#34;digital sandbox&#34; for autonomous driving systems, allowing for infinite simulations of extreme weather and rare road conditions, enabling algorithms to adapt in advance and reduce real-world trial-and-error costs.
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/b2f56f39-e747-43ef-9b6f-e6602ed83251.png?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/b2f56f39-e747-43ef-9b6f-e6602ed83251.png"/>
The iteration of autonomous driving technology requires a large amount of training data and simulated scene support. NVIDIA provides three key computing systems—AI training system, virtual world &#34;Omniverse,&#34; and data synthesis system &#34;Cosmos,&#34; forming a closed-loop system from data collection to model training to simulation.
Omniverse: Provides high-fidelity virtual scene simulations for testing and optimizing autonomous driving algorithms.
Cosmos: Generates new virtual data using real data obtained from autonomous vehicles, further enriching training samples.
Through these tools, automotive companies can significantly reduce R&amp;D costs, accelerate technological iteration, and improve the accuracy and reliability of autonomous driving algorithms. Even strong players like Tesla are using NVIDIA's core systems to enhance themselves.
● Ecosystem Cooperation: Building an &#34;Autonomous Driving Circle of Friends&#34;
Toyota is one of NVIDIA's important partners, and the two parties will engage in deep cooperation in the field of next-generation autonomous vehicles. Toyota's scale advantage complements NVIDIA's technological strength, providing strong support for the future implementation of autonomous driving technology.
In addition to Toyota, NVIDIA has also attracted several automotive companies, including Tesla, Mercedes-Benz, BYD, Li Auto, Xiaomi, and ZEEKR, to join its ecosystem.
These partners cover a range from traditional automotive companies to new force brands, such as Aurora and Continental Group, showcasing NVIDIA's strong appeal in building the autonomous driving ecosystem.
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/beab6029-154b-4d4a-95af-f43ff2113a73.png?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/beab6029-154b-4d4a-95af-f43ff2113a73.png"/>
Here, domestic companies XPeng and Nio have exited NVIDIA's latest circle, with each enterprise making its own choice.
NVIDIA assists in the construction of the autonomous driving ecosystem closed loop from the optimization of technical solutions to hardware adaptation and integration, covering the entire link from algorithms to vehicle implementation. You can also exit part of this circle, but completely leaving NVIDIA's system to create your own unique system, in China, only Huawei can do that.
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/b85107f3-c342-43db-ba9d-6dd40344ea3b.png?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/b85107f3-c342-43db-ba9d-6dd40344ea3b.png"/>
<h2>Part 2 Physical AI: NVIDIA's Strategic Perspective</h2>
In NVIDIA's view, physical artificial intelligence is the key to unlocking advanced capabilities in autonomous driving and robotics. It is not merely a theoretical model but an intelligent form closely related to the interactive logic of the real physical world.
Traditional AI models often struggle when faced with the complex physical dynamics of robotic manipulation and autonomous driving, as the real world is full of uncertainties, intertwined with factors such as object motion, mechanical relationships, and unpredictable human behavior.
Physical AI must accurately simulate, understand, and respond to these situations, such as adjusting the center of gravity when a robot is lifting heavy objects or estimating physical collisions when an autonomous vehicle is avoiding obstacles. This is a core threshold for advancing towards mature intelligent applications.
Physical AI is a new concept proposed by NVIDIA, empowering the intelligence of the physical world through AI technology.
The Cosmos world foundation model is a key tool for achieving this goal. The model can convert information such as text, images, and videos into tasks that robots and autonomous vehicles can execute, thereby providing developers with rich training data.
Cosmos generates realistic physical scenarios through 20 million hours of video training, used for optimizing algorithms in autonomous driving systems and robots.
Developers can utilize Cosmos to generate simulation scenarios under specific weather conditions, road conditions, or traffic situations, allowing for more effective evaluation and training of autonomous driving systems and the development of physical AI models, which previously required massive real-world data collection and testing, incurring high costs and time.
Taking XPeng as an example, the cost of collecting autonomous driving training data under different lighting and terrain conditions was previously high. With Cosmos, similar effect data can be synthesized on a virtual level, greatly saving resources.
Cosmos not only provides general foundational data but also allows developers to fine-tune and build customized models. This attracts companies of various sizes and application scenarios.
Ride-hailing giant Uber aims to optimize its autonomous delivery algorithms, while robotics company 1X seeks to enhance the coordination of humanoid robot movements, both of which can adjust based on Cosmos as needed, lowering development barriers and stimulating industry innovation.
NVIDIA positions Cosmos as the &#34;ChatGPT&#34; of the physical AI field, aiming to allow more developers to access high-quality data resources and AI tools through an open foundational model.
● This strategy has two implications:
Lowering technical barriers: By opening the Cosmos platform, developers can easily deploy models on NVIDIA's accelerated computing systems, avoiding high data collection costs.
Expanding ecological influence: NVIDIA further expands its ecological influence in the physical AI field through collaborations with automotive companies, XPeng, Uber, and other partners.
No one can stand alone; full effort is needed to collaborate with the global developer ecosystem. Uniting various forces to involve university research teams, startup tech companies, and mature industrial giants in the construction of physical AI
It's like a feast of open-source, where all parties contribute wisdom and share results. NVIDIA firmly sits at the technical foundation and ecological hub, driving the wave-like advancement of physical AI technology, continuously expanding from basic theory to application scenarios, radiating across diverse fields such as automotive and robotics.
Autonomous vehicles are not only an application scenario for AI technology but also a major source of generated data. Through the combination of Omniverse and Cosmos, NVIDIA provides automakers with comprehensive support from production line design to product optimization.
Production line optimization is based on virtual simulation technology, reconstructing the automotive manufacturing process to improve efficiency and flexible production capabilities.
Data-driven R&amp;D utilizes the massive data collected from autonomous vehicles to generate more real-world training samples, further enhancing the capability boundaries of autonomous driving technology.
<h2>Summary</h2>
Jensen Huang's speech at CES 2025 clearly conveyed NVIDIA's strategic ambitions in the automotive field: to lead the technological transformation of autonomous driving and physical AI with leading computing power and ecological layout.
From the Thor platform to the Cosmos model, NVIDIA provides automakers and developers with a complete set of solutions from hardware to software, accelerating the process of automotive intelligence.
As Jensen Huang said, &#34;The era of autonomous driving has arrived.&#34; In this transformation, NVIDIA is not only a driver of technology but also a definer of future mobility.
Author: Tao Yanyan, Source: ZhiNeng Automobile, Original Title: &#34;CES | NVIDIA's Multi-Dimensional Assault on Automotive Business: The Era of Autonomous Driving Has Arrived&#34;
Risk Warning and Disclaimer
The market has risks, and investment requires caution. This article does not constitute personal investment advice and does not take into account the specific investment goals, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific circumstances. Investing based on this is at your own risk

NVDA

NVIDIA announced at the CES press conference that the era of autonomous driving has arrived, launching the fourth-generation Thor intelligent driving computing platform, with computing power increased by 20 times. It is expected that by fiscal year 2026, the scale of the automotive business will reach $5 billion. The Thor platform supports multi-sensor fusion, achieving L4 to L5 level autonomous driving, providing a one-stop solution widely used in the automotive and robotics fields. NVIDIA optimizes autonomous driving algorithms through real data, promoting the transformation of automotive intelligence and influencing the industrial landscape and mobility ecosystem

- NVIDIA's CEO announced the arrival of the autonomous driving era at CES 2025, unveiling the fourth-generation Thor computing platform, which boasts 20 times the computing power of its predecessor.  
- The company aims to expand its automotive business to $5 billion by fiscal year 2026, leveraging the Cosmos model for low-cost data generation to accelerate automotive intelligence transformation.  
- NVIDIA's strategy emphasizes the integration of physical AI and collaboration with major automotive partners, positioning itself as a leader in the evolving landscape of autonomous driving technology.

Wallstreetcn

NVIDIA

Toyota

Huaru

Direxion Semicon Bull 3X

GraniteShares 2x Long NVDA Daily ETF

iShares Self-Driving EV&Tech ETF

IFLYTEK

VanJee

KSW-TECH

PIA Automation

UBTECH ROBOTICS

Hesai

Mercedes-Benz Group AG

Global X Autonomous & Electric V

VASTDATA

YieldMax NVDA Option Income Strategy ETF

Tesla

Naruida

GLARUNTECH

Toyota Motor

SENSETIME-W

XPENG-W

NIO-SW

BYD COMPANY

Huawei

VanEck Semiconductor ETF

Focuslight

Thor Industries

TM ADR US

Suochen

Intel

OpenAI

Aerospace Nanhu

NVIDIA's multi-dimensional advance into the automotive business: The era of autonomous driving has arrived