NVIDIA's NEW AGE Blackwell B200 Artificial Intelligence Development...

"While rumors about the Hopper H100 Tensor Core chips, launched in 2022 and used by companies like OpenAI to develop and deploy their own artificial intelligence algorithms, continued, Jensen Huang announced a stunning chip announcement."

The name of the new chip; Blackwell B200


I would like to explain the power of this new chip to you with these words. The Hopper 100 chips, which were released just two years ago, banned from export to US rivals, and used in training artificial intelligence models, have now become "old" with this new product.
In fact, what made Nvidia a trillion-dollar company and elevated it to the position of the world's third most valuable company was the Hopper H100 GPU cards.
Blackwell B200 cards deliver performance that is 5 times more powerful than the Hopper H100 cards.

Here's the translation of the part you provided from CNBC, given my expertise in English:
"Compared to Hopper, Nvidia Blackwell GPU offers 128 billion more transistors, 5 times more artificial intelligence performance increased to 20 petaFlops per chip, and 4 times more embedded memory."

NVIDIA's RTX 4090 cards, which are considered one of the most powerful graphics cards on the market today, offer 90 teraflops of processing power per second. We can liken this value to the TPS (transactions per second) in blockchain networks. Similar to processing power per second...
On the other hand, Blackwell cards can soar to 10,000 teraflops of processing power per second.

Meanwhile, NVIDIA is set to reduce the cost and energy consumption compared to the H100 by up to 25 times with this new product.
NVIDIA has officially begun to leap from generation to generation in just a few years.
The year 2023 was hailed as a revolution for technology.
And we are now seeing that 2024 could be a revolution for generations.
Imagine thousands of Blackwell chips being placed inside a computer. Think about the tremendous computing power it could achieve.
Amazon has already ordered more than 20,000 Blackwell chips.
Of course, the subject wasn't just artificial intelligence.
Now, let's move beyond artificial intelligence a bit.



Now, after discussing these chips, we have also seen that a real virtual universe can be created. So, the words "we don't have enough power to build Metaverse technology in previous years" have become a thing of the past.
Jensen Huang emphasized the need for different training for robots to enter our lives faster.
There is only one way for robots to evolve.

-Make a mistake

You heard it right. Unfortunately, according to Jensen Huang's statements, there isn't enough room for mistakes in our world.
For this, there are two remaining paths.
1- Internet of Things (IoT) Technology
You heard it right.... The Internet of Things technology allows all our data to navigate on the IoT technology. Thus, artificial intelligence algorithms can learn new things even without making mistakes by collecting data from here.
Although it is expected that 100 billion devices will be connected to the internet by 2025, this may not be sufficient.
Jensen Huang is resorting to a new method here.

- OMNİVERSE

Omvinerse, bir kavram olarak Metaverse'in ötesinde yer alır ve çoklu evrenlerin, bir arada var olduğu varsayılan paralel evrenlerin bir kümesini ifade eder.
My first thought is that most likely the Metaverse sector will become hype again in this regard.
Anyway, it's up to you to think about it, I'll continue sharing my notes.
Nvidia aims to establish a virtual system. This virtual system will have a replica of our real world. By connecting artificial intelligence models to this system, they will be enabled to make mistakes faster.
Of course, Nvidia will not do this alone. Other tech giants will also be involved in this system.
As a result of all these plans, the goal is to bring robots closer to individuality in the near future.
We can also say that metaverse universes that companies like Apple and Meta are working on are no longer difficult. This technology could fully enter our lives in just a few years.
This system, which we call omniverse, will evolve to the point where even we will build on omniverse first when we want to produce a product.