Summary of Key Points
At the 2026 Taipei Computer Show, NVIDIA unveiled two seemingly unrelated products: the humanoid robot reference design H2+ (Isaac GR00T system) in collaboration with YuShu, and the AI PC chip RTX Spark developed jointly with MediaTek. These moves are not isolated; they represent NVIDIA's strategic shift from a "data center graphics card vendor" to an "AI platform provider." By offering complete reference designs that integrate hardware and software, NVIDIA aims to extend its CUDA ecosystem (the "universal language" for AI development) to new areas such as PCs and robots. Its goal is to become an indispensable partner for developers in every emerging AI application, betting on growth across various use cases over the next decade.
NVIDIA's Transformation: From Hardware Sales to AI Infrastructure
In the past, NVIDIA made its money by selling GPUs to cloud providers and large model companies, with its CUDA ecosystem as its core competitive advantage—ensuring that millions of developers became accustomed to writing code on its hardware, creating a barrier where "you can't do AI efficiently without using our cards." However, the growth in data centers has reached a ceiling. Therefore, Jensen Huang plans to apply this approach to more areas:
- For the robotics industry: By providing ready-made development frameworks (reference designs), researchers can start working with NVIDIA's hardware and software without having to build from scratch.
- For the PC industry: By introducing AI chips, NVIDIA aims to transform PCs into "personal AI computing centers," enabling ordinary users to run AI applications locally.
In simple terms, NVIDIA is now selling a "full set of tools" that enable rapid AI development, essentially setting up a stage for anyone interested in entering the AI field, with the message: "Just use our devices and follow our rules."
RTX Spark: A Game-Changer for AI PCs or a Repeat of Past Failures?
AI PCs have been around for two years without gaining significant traction due to two main issues: software compatibility (Arm-based PCs can't run x86 software) and consumer confusion about the practical uses of AI PCs. NVIDIA and MediaTek's RTX Spark aim to address these challenges:
- Hardware Advantages: The chip incorporates NVIDIA's cutting-edge gaming GPU technologies (DLSS, G-Sync), which could motivate game developers to adapt their games to the Arm platform—a feat Qualcomm hasn't achieved in seven years.
- Smart Positioning: The chip is marketed as a "personal AI computing chip," suggesting it's versatile for both gaming and work tasks such as AI-powered graphics and voice assistance.
- Alliance Building: Qualcomm welcomes NVIDIA's participation, as both companies are part of the Arm ecosystem, working together to counter Intel/AMD's x86 dominance.
However, there are risks: Jensen Huang claims the chip will be compatible with all Windows software, but this is a bold bet. If successful, it could revolutionize the AI PC market; otherwise, NVIDIA might face the same compatibility issues that Qualcomm has encountered.
The Robot Reference Design: Aiming to Be the "Android of Humanoid Robots"
The YuShu H2+ robot uses hardware manufactured by YuShu, but its "brain" (Jetson chip) and operating system (Isaac GR00T) are provided by NVIDIA. This reference design's impact lies in:
- Lowering the Bar: Researchers in universities can start their work immediately with a purchase of H2+, saving time and money.
- Ecosystem Integration: It includes the necessary components for AI development, such as computing power, basic models, simulation training platforms, and data generation tools (like Cosmos, which addresses the lack of first-person perspective data for robots).
- Extending the CUDA Ecosystem: Similar to how CUDA made GPUs essential for developers, Isaac GR00T aims to make NVIDIA's framework indispensable for robot development. Morgan Stanley predicts that humanoid robot sales in China will double this year to 28,000 units; NVIDIA's goal is for every future robot to utilize its technology.
Challenges Ahead
NVIDIA's new strategy faces two major hurdles:
- PC Market: Consumers are still focused on traditional factors like brand (Intel/AMD), processor type (i7/Ryzen), and gaming capabilities, with the value of AI PCs not being clearly communicated. Intel and AMD are also developing integrated NPU/GPU chips, potentially eroding NVIDIA's advantages.
- Robotics Market: The robotics industry is still in its early stages, with high costs, maintenance challenges, and unclear regulations. If growth doesn't materialize until five years from now, NVIDIA's investment may take a long time to pay off.
The Long-Term Gamble: Establishing a Presence in All AI Use Cases
The vision behind NVIDIA (HYPERION cars, SPACE-1 satellites, HOLOSCAN medical devices) represents its "ten-year roadmap"—each project corresponds to a potential AI application. NVIDIA aims to establish its tools in these areas before they become mainstream. The strategy is simple: it doesn't need to win in every market; achieving a 20% market share in key areas (e.g., with AI PCs or robots) could generate significant revenue. Its goal is to be present in every transformative AI application, ensuring its longevity.
RTX Spark and H2+ are NVIDIA's offers to enter these markets. Whether they succeed depends on consumer and market acceptance over the next five years.
In Summary: NVIDIA is transitioning from being the "king of AI GPUs" to a provider of AI infrastructure, using its ecosystem and tools to dominate the new AI landscape. However, its success will depend on whether consumers and the market embrace these innovations.