April 20, 2024

Earlier this week, Nvidia launched their developers conference in San Jose, CA, and rolled out new products that will power the future of AI.

As tech conferences go, this one was different. I have been to every GPU Technology Conference, (GTC) since Nvidia has held them, and this was the largest and most impactful event they have ever had.

The keynote venue alone was impressive. To handle the over 10,000 attendees, Nvidia had to rent the Shark Tank, the 11,000-seat SAP Center, where the NHL San Jose Sharks play their games. This conference must be the most prominent tech event for which a company needed a vast arena to handle the keynote crowd. The conference occurred at the San Jose Convention Center, a mile from the Shark Tank.

Nvidia has become the darling of Wall Street as their GPUs power the AI revolution. It has become a trillion-dollar company and made its CEO, Jensen Huang, a billionaire. I met Jensen before he started Nvidia and have followed the company since its inception. I have watched the company close for decades and saw its fortunes rise, fall, and rise to a level of success well beyond any of us could have predicted.

Mr. Huang is among the most intelligent people I have met in the tech industry. Nvidia’s success is based on Jensen’s keen ability to see the future and invent the technology to make it happen. Over my years of covering the company, I have had many personal discussions with him and marveled at how Nvidia has become one of the most essential companies driving our AR future.

The most significant announcement from Jensen’s keynote was the introduction of the Nvidia Blackwell B200 GPU and GB200 “super chip.” Nvidia says the new B200 GPU offers up to 20 petaflops of FP4 horsepower from its 208 billion transistors. Also, a GB200 that combines two of those GPUs with a single Grace CPU can offer 30 times the performance for LLM inference workloads while potentially being substantially more efficient. This “reduces cost and energy consumption by up to 25x” over an H100, says Nvidia.

In his keynote, Mr. Huang illustrated the power efficiency of their new Blackwell B200 by saying that this chip could handle a 1.8 trillion parameter model that would have previously taken 8,000 Hopper GPUs and 15 megawatts of power. Now, 2,000 Blackwell GPUs can do it while consuming just four megawatts. On a GPT-3 LLM benchmark with 175 billion parameters, Nvidia says the GB200 has a somewhat more modest seven times the performance of an H100, and Nvidia says it offers four times the training speed.

This game-changing processor could allow more extensive parameters and faster inferencing, requiring much less power than past chips.

My colleagues at Forbes have done a deeper dive into the Blackwell B200, and it is worth a read for more details on this remarkable and powerful new processor.

Although there were many other significant announcements made during Mr. Huang’s keynote, he showed an image of a cat that was created by an AI prompt. He then postulated that using AI to create code has major ramifications for the future of programming.

Earlier this year, Jensen Huang began making statements about the concept of no-code programming. He predicts that AI will make traditional coding relatively unnecessary in the future, and will democratize programming.

If you have used any of the current AI programs that allow you to enter a prompt and then draw a picture of what you asked, that is an example of no-code programming. Open AI’s Sora is an even better example of AI coding. Give a prompt to Sora, and if done correctly, it could deliver a short video of your request. While that seems magical, Sora generates the code that, in the past, had to be done by human programmers over hours of work and delivers the results in seconds.

After the keynote discussion with analysts, Mr. Huang reinforced this concept. He said that, “the end user is connected to a supercomputer that does the coding for them to act on whatever they want to create.” A user’s request creates the code to answer the question and delivers results. The prompt is searching an AI inference database and generating code or content to produce a person’s desired program.

When asked whether programming will remain a helpful skill in the age of generative AI prompts, Huang said, “I think that people ought to learn all kinds of skills,” and compared code to juggling, playing piano, or learning calculus. However, “programming is not going to be essential for you to be a successful person.”

Generative AI, Huang said, is “Closing the technology divide. You don’t have to be a C++ programmer to be successful,” he said. “You just have to be a prompt engineer. And who can’t be a prompt engineer? When my wife talks to me, she’s prompt engineering me. We all need to learn how to prompt AIs, but that’s no different than learning how to prompt teammates.”

This statement is a big deal. If Jensen is correct, traditional programming is on the brink of transformation, shifting the programmer’s role from coder to highly skilled prompt engineer who hones specialty skills. For instance, programmers creating Hollywood special effects will no longer need traditional coding to produce those effects. Instead, they will use prompts to manifest their desired effects into being.

All traditional coders and programmers need to follow these developments closely. Using standard language to create the commands to write a program or code to perform some action will soon be done by just giving it a command, and AI will write the code for the programmer.

Programmers will use AI to do the heavy lifting of the written code and customize it for any particular application or use. However, we are on the verge of a severe earthquake in traditional programming that will impact all levels of software engineering over the next few years.

When I started my career in tech in the mid 1970s, I learned to do basic coding using a DEC PDP 11 computer. But I was terrible at it, as I was trained as a marketing specialist, and I was only asked to do this because the company needed more programming talent. However, I admired those who could program and always wanted to be able to do serious coding sometime in my career.

AI-based programming now gives me hope that I can achieve that goal, albeit with the help of AI.

Disclosure: Nvidia subscribes to Creative Strategies research reports along with many high tech companies around the world.