NVidia’s Latest AI Chip Can Be The Door To The Next Era Of Humanoid Robots

Nvidia

Every science fiction enthusiast’s dream comes true: Nvidia has increased its lead in artificial intelligence with the release of a new “superchip,” a quantum computing service, and a suite of tools to assist in the creation of all-purpose humanoid robots. We’ll examine the company’s actions and their potential implications.

Nvidia: What Does It Do?

At the company’s annual development conference on Monday, the main announcement was about the “Blackwell” family of AI processors. These chips drive the highly-priced data centers that train the most recent iterations of GPT, Claude, and Gemini, as well as other cutting-edge AI models.

One is a rather straightforward upgrade over the company’s prior H100 AI chip: the Blackwell B200. An enormous AI model the size of GPT-4 would require roughly 8,000 H100 chips and 15 megawatts of power to train, according to Nvidia. That would power around 30,000 typical British households. It would only take 2,000 B200s and 4MW of power to complete the same operation using the company’s new chips. This may lead to the AI industry using less electricity, or it may soon result in the AI industry using the same amount of electricity to run far larger AI models.

Why Is A Chip Such a Great Thing?

The B200 and GB200 “superchip” are the two parts of the Blackwell line that the business introduced simultaneously. Nvidia claims this method has “30x the performance” of server farms that use ChatGPT Claude or other similar chatbots without first educating them. It accomplishes this by placing the company’s Grace CPU and two B200 processors on a single board.

According to the business, this approach might reduce energy use by up to 25 times. Placing everything on the same board increases efficiency by reducing the amount of time the chips spend communicating with one another. This implies that the chips will have more time to work with the algorithms that enable chatbots to sing, or at least communicate. Worth almost $2 trillion (£1.6 trillion), Nvidia would be pleased to contribute.

Use the company’s GB200 NVL72 as an illustration. There are around two miles of wires extending from the one server rack that houses the 72 B200 chips. Is that insufficient? Instead, have a look at the DGX Superpod. It builds an AI data center the size of a shipping container by assembling eight of those racks. The event did not specify the price, but it is safe to assume that you cannot afford it if you have to inquire. Even the newest chips have a price tag of almost $100,000 per.

Nvidia Superchip

How Does That Affect Robots?

Project GR00T from Nvidia is a new foundation paradigm for operating humanoid robots. Though unrelated to the arboriform extraterrestrial from Marvel, it has that moniker. On top of a foundation model, you can construct specific use cases, such as StableDiffusion for image production or GPT-4 for text. Even though they are the most expensive to produce, their ability to be “fine-tuned” to meet certain needs in the future is what propels all future advancements.

Nvidia’s foundation model for robots will observe human movement and language to enable them to “quickly learn coordination, dexterity, and other skills in order to navigate, adapt, and interact with the real world.”

Jetson Thor, a system-on-a-chip designed to function as a robot’s brain, is used by GR00T. It connects to another Marvel device by pairing with another piece of Nvidia technology. Ultimately, the aim is to have an autonomous computer that can be instructed to perform generic activities using natural human speech, even if it hasn’t been trained for those particular jobs.

Where Does Quantum Computing Come In?

One of the few hot sectors in which Nvidia is not a leader is quantum cloud computing. Though the technology is still in its infancy, Nvidia is now joining the fun after Microsoft and Amazon have already included it in their products.

But there won’t be a link between a quantum computer and Nvidia’s cloud. Rather, the business offers a service that imitates a quantum computer using its AI chips. Experts can test their theories in this way without having to pay for access to the expensive and uncommon actual one. But over time, Nvidia promised that the platform would enable customers to establish connections with other businesses’ quantum computers.

People will undoubtedly mention OpenAI as the firm driving the current AI frenzy on a daily basis. Its CEO travels the world, it expands rapidly, and it offers incredible public tools. However, if you ask a tech expert, they’ll probably pick Nvidia.

Nvidia has been producing processors that have been the foundation of every significant technological advancement since its founding in 1993. The business is currently the largest manufacturer of the chips driving the AI boom after placing a risky early bet on machine learning. Its market capitalization is more than $2 trillion (yes, trillion with a “t”), and its stock price is surging. “The Woodstock of AI” is an event that the corporation is hosting this week to showcase much-anticipated new technology.

Jensen Huang Is Now One Of The Greatest Leaders In The Industry

The same individual has led Nvidia since its founding: Jensen Huang, a co-founder. Compared to many other IT companies, this is unique. Huang is a Taiwanese immigrant who had a difficult beginning in life. His company is at the forefront of new technology, and as a result, he has guided it to become worth approximately $80 billion.

How has he continued to be so successful for so long in an industry that is so competitive and changes so fast? He recently gave two speeches at Stanford, where he attended college, discussing Nvidia’s future AI goals and his leadership approach. He made some excellent remarks on the Business School’s View From the Top series and at the Stanford Institute for Economic Policy Research. Both are highly recommended for a comprehensive viewing. Though Huang’s two-hour speeches covered a lot of ground, his five words provided a succinct summary of his thoughts and a useful leadership lesson for practically everyone.

View From the Top’s Shantam Jain invites Huang to discuss how he approaches business design and what other people should consider while doing the same. In his five-word response, Huang states, “No task is beneath me.” This is extremely modest coming from a CEO with a billion-dollar net worth. It is, indeed. Huang then ridicules his early low-level occupations and how they influenced his mentality. Cleaning the toilets was my responsibility. I did a lot of loose cleaning. He makes people chuckle by saying, “I’ve cleaned more toilets than all of you put together, and some of them you just can’t forget.”

Many people, including Professor Adam Grant of Wharton, Jeff Bezos, and psychologists, have stated that future leaders should possess the quality of humility. According to these various thinkers, humility enables us to connect with people, be receptive to new information, and make wiser decisions.

Huang is displaying humility only out of self-interest. He claims that his maxim “no task is beneath me” goes beyond just overcoming arrogance and sloth. It’s also about empowering others that collaborate with him.

Regarding Possessing Power And Being Humble

In his address to the Stanford Institute for Economic Policy Research, Huang expands on this. Surprisingly for a CEO, he oversees over 50 direct reports that he does not personally interact with. The supervisors of these reports don’t regularly meet one-on-one or receive written evaluations. Unless they require assistance with something, they are mainly left alone.”Then I’ll drop everything for them,” Huang declares. Why? Huang elaborates: “In that way, our company was designed for agility, for information to flow as quickly as possible, for people to be empowered.”

Huang isn’t simply a lovely guy; he will stop at nothing to support his team in any situation. Additionally, he acts in this way because he believes that his primary responsibility is to train his employees so they can function independently and swiftly.

“Our job at the company should depend on how well we can solve hard problems, motivate others to do great things, and help, inspire, and empower others.” According to Huang in View From the Top, there is a management team to assist all other employees of the organization for these reasons.