The most important thing today is NVIDIA’s GTC conference—an AI version of a brief history of humanity.

The most important thing today is the NVIDIA GTC conference—it’s basically an AI version of a human history of mankind.

Huang Renxun hasn’t even taken the stage yet, but the leaked information beforehand is already enough to fill a whole book.

WannaWanna has sorted out three big takeaways. Come on, fat friends—follow me.

1）AI compute costs cut straight to one tenth

The previous-gen Blackwell was already pretty powerful, right? The next-gen chip, Vera Rubin, is about to go into mass production.

What’s so strong about Vera Rubin? In plain terms: it’s cheap.

Running the same AI model, the number of chips drops to one quarter, and the inference compute cost falls by 90%. Falls by 90%, friends. AWS, Microsoft, and Google—the three major cloud providers—are directly among the first to adopt.

2）Groq, bought for $20 billion last year, delivers its homework today

Before, at an earnings call, Huang Renxun said Groq would be connected as an expansion architecture into the NVIDIA ecosystem—like how, back then, buying Mellanox completed network capabilities.

Groq’s LPU and NVIDIA GPUs sit in the same data center: GPUs understand the problem, while the LPU rapidly spits out the answer.

With the two types of chips working in tandem, Agent scenario latency gets knocked down directly.

AI Agents do the work for humans. One task might involve dozens of model-tuning rounds back and forth, and each round burns inference compute—while the user is waiting. If it’s even a bit slower, the experience breaks.

Inference happens in two steps: first understand your question, then output the answer word by word. The GPU is good at the first step, but on the second step—the speed and stability of “typing out” tokens—Groq’s LPU is stronger.

Is $20 billion expensive?

Think about it: in the future, every company runs hundreds of Agents, and each Agent tunes the model thousands of times every day.

3）NVIDIA’s OpenClaw version goes live, called NemoClaw

It’s an open-source platform: once enterprises install it, they can deploy AI employees to handle workflows normally done by real people—running processes, processing data, and managing projects. It’s said to already be in talks with Salesforce and Adobe.

What’s interesting is that NemoClaw doesn’t require you to use NVIDIA chips. Think about that logic. Selling chips only makes money from the hardware layer; setting the rules is what lets you make money across the entire chain. Huang Renxun has clearly done the math.

4）Huang Renxun says he’ll demonstrate “chips the world has never seen before”

Most likely, it’s the first public appearance of the next-next-gen architecture, Feynman, with mass production in 2028 and TSMC’s most advanced 1.6nm process.

Also, there’s one lesser-known piece of news I think is pretty interesting.

NVIDIA is making laptop computer processors—two of them—focused on gaming. The GPU sellers are coming to fight for the CPU’s dinner.

WannaWanna, I feel like Huang Renxun is going to become a great figure of an era in the future.