At the GTC conference, Jensen Huang has already told you the code.

Yesterday, the global tech and capital circles once again focused their attention on NVIDIA's Taipei GTC conference. There was a pre-conference warm-up event before the main conference began, where the new generation of chips tailored for AI PCs, RTX Spark, was announced, marking NVIDIA's official entry into the PC market.

The strength of the new chip isn't about adding another AI plugin to the PC; it's about redefining the PC as a personal AI computer that can run local agents.

Traditional computers of the past were like "kit cars." The CPU, which handles basic tasks, was the chassis, and the GPU, which processes graphics, was the engine. These two parts were separated by traffic lights and congested roads, resulting in extremely slow data transfer. RTX Spark, however, directly merges the chip engine (Blackwell GPU) and the luxury custom chassis (MediaTek Arm CPU) onto the same frame using a superhighway (NVLink C2C), sharing a massive 128GB of unified memory. This architectural paradigm shift leaves traditional x86 architecture-based systems in the dust.

What's truly important behind this is that the way we interact with PCs might be reshaped by agents. PCs used to be the gateway to apps; in the future, they might become the gateway to personal Agents.

The biggest expectation gap in the current market is that the vast majority of people still see NVIDIA merely as a server chip vendor selling shovels in data centers and believe the consumer PC market is already a stagnant pond. But this keynote speech revealed the industry's biggest underlying card—NVIDIA is using RTX Spark to begin dismantling the x86 hegemony that Intel and AMD have ruled for decades, turning every consumer-grade notebook into a decentralized micro AI factory through edge-side AI.

Furthermore, the NVIDIA DSX AI Factory Ecosystem reveals the scope of NVIDIA's, and even the entire AI industry's, landscape: encompassing everything from the cloud and software to infrastructure and energy cooling. For a detailed interpretation, you can check out the two charts I made below. The consumption of AI hardware will ultimately be transmitted to the real economy.

(Not for investment purposes)

$STAR Chip Index(000685.SH) $NVIDIA(NVDA.US) $Taiwan Semiconductor(TSM.US) $Arm(ARM.US)