Summary of Key Points
NVIDIA, in collaboration with Microsoft, is “reinventing the PC” with a superchip called RTX Spark – integrating data center-level computing power into thin and light laptops, enabling computers to run large models with over 120 billion parameters locally. They have also introduced the Vera CPU, designed specifically for AI agents, to revolutionize the computational foundation of data centers. Additionally, they have created an “agent operating system” using a four-layer software stack, transforming users from mere “tool users” into “commanders” that direct AI operations. This transformation not only challenges the dominance of Intel and AMD’s x86 architecture, which has lasted for four decades, but also aims to turn PCs into 24/7 AI partners. Brands like ASUS and Dell are set to launch related products by the end of the year, though the prices have not yet been announced, sparking intense discussions among netizens.
Detailed Analysis
#### 1. Integrating Data Center Power into Thin and Light Laptops: How Powerful is RTX Spark?
In traditional PCs, the CPU and GPU used separate memories, resulting in inefficient data transfer when running large models, which was akin to “borrowing data” back and forth, causing significant delays. RTX Spark eliminates this bottleneck by sharing a 128GB memory pool between the CPU and GPU, offering bandwidth dozens of times higher than traditional solutions. This allows for the local, offline execution of models with over 120 billion parameters (equivalent to reading the entire “The Three-Body Problem” series in one go), without the need to upload sensitive data to the cloud.
What’s more remarkable is RTX Spark’s versatility: based on the Arm architecture, it is 100% compatible with Windows and CUDA (commonly used by programmers). It can handle games at 1440p resolutions with 100 frames per second, doubles the speed of AI functions in Adobe software, and enables smooth editing of 90GB 3D scenes and 12K videos on battery-powered laptops. This fall, brands like ASUS and Lenovo will release thin and light laptops with a thickness of just 14 millimeters – marking the first major upgrade to the “brain” of Windows PCs in over four decades.
#### 2. The CPU Designed for AI: How Vera Boosts Data Center Efficiency
Traditional CPUs are designed for human users (with slower performance and billing based on core count), but AI agents require nanosecond-level processing speeds, making them a bottleneck for GPUs. NVIDIA’s Vera CPU is the first to be tailored specifically for AI:
- Optimized Single-Core Performance: It can process 10 instructions per clock cycle (setting a world record), with 88 cores communicating 50% faster than traditional CPUs.
- Energy Efficiency: It performs SQL queries three times faster and real-time stream processing six times faster while using less power.
- Advanced Platform: The Vera Rubin rack system uses liquid cooling, is cable-free, and assembly time has been reduced from two hours to five minutes, with production capacity doubling compared to the previous generation.
This means that the “commander” of the data center (the CPU) can finally keep up with the pace of the “orchestra” (the GPU), significantly boosting the efficiency of AI systems.
#### 3. AI as Your Personal Computer Assistant: NVIDIA’s Ecosystem Locks in the Future
Hardware is just the foundation; NVIDIA’s true ambition is to build a comprehensive “agent operating system.” Traditional PCs require users to manually operate software, but with this new approach, AI agents can handle tasks such as drawing with Rhino, rendering with Blender, and coordinating across applications.
To ensure developers rely on their platform, NVIDIA offers a four-layer toolkit:
- Model Layer: The open-source Nemotron 3 Ultra model, with 550 billion parameters, reduces inference time by 50% while cutting costs by 30%.
- Framework Layer: It directly supports popular AI frameworks like Claude Code.
- Tool Layer: Thousands of CUDA-accelerated libraries come with pre-configured “skill sets” that can be used by AI agents.
- Security Layer: The OpenShell sandbox isolates risks and protects user privacy.
For example, chip-designing agents developed in collaboration with Cadence can automatically run simulations and fix bugs, reducing the verification cycle from weeks to hours. In the future, you might be able to use local AI agents from the Windows taskbar to write emails or control smart homes – once this ecosystem is established, developers will find it difficult to switch to other solutions.
#### 4. Who Is This Transformation Affecting?
- Intel and AMD Are Concerned: The forty-year-old x86 architecture is being challenged by RTX Spark (based on Arm) and the Vera CPU, leading to a shift in both PC and data center “brains.”
- Users Become More Relaxed: They no longer need to learn complex software; they simply state their requirements, and the AI agents handle the tasks.
- The Industry Direction Is Changing: PCs are no longer just tools for humans but become 24/7 AI partners. Whoever controls the AI ecosystem will dominate the future.
Some may argue that this isn’t as groundbreaking as the iPhone, but the iPhone changed the relationship between people and devices; NVIDIA, on the other hand, is changing who truly “owns” these devices. In the future, the AI in your PC will understand software better than you do, and all you’ll need to do is give commands.
#### 5. What About the Prices? Netizens Are Speculating, but NVIDIA Keeps Them Secret
Netizens are debating whether prices should be around 5,000 or even 10,000 yuan, but NVIDIA tends to keep prices under wraps until the launch. Based on similar configurations:
- Thin and Light Laptops: Incorporating RTX Spark will likely result in prices above 10,000 yuan due to the advanced capabilities.
- Workstations: Versions with up to 768GB of memory could cost several times more than regular workstations.
- Desktop AI Servers: 24/7 online versions will be more expensive than thin and light laptops but cheaper than dedicated workstations.
Regardless, this AI-powered PC, which combines data center capabilities in a portable form, is undoubtedly one of the most anticipated technology products of the year.
Final Conclusion
NVIDIA is overhauling the 40-year-old logic of PCs with its chips, CPUs, and ecosystem. In the future, your computer will be an AI partner that helps you complete tasks – whether you’ll be willing to pay for it depends on the price. What do you think the price should be? Feel free to share your predictions in the comments!