CNN
–
“The US plan to “industrialize” technology manufacturing is “the right thing to do,” says Jensen Fan, CEO of one of the world's leading AI chip manufacturers.
Huang, who heads NVIDIA based in Santa Clara, California, said in an interview with CNN's Fareed Zakaria that the US should invest in manufacturing and is now “missing that whole band in our industry.”
“That passion, the skills, the skills to make things. The ability to make things is valuable for economic growth. It is valuable for people and a stable society who can create great lives and great careers without having to earn a PhD in Physics,” Huang said.
The Trump administration has enacted a number of policies, including wiping tariffs to revive America's declining manufacturing industry. This was to boost investments in the automotive and energy sectors and technology.
“President Trump has made it clear that the US cannot rely on China to manufacture important technologies such as semiconductors, chips, smartphones and laptops,” White House news director Caroline Leavitt said in an April statement after a temporary suspension of tariffs was established on smartphones and other electronic devices.
Huang said that if supervised manufacturing and manufacturing removes pressure from Taiwan, Taiwan Semiconductor Manufacturing Company (TSMC), the world's largest semiconductor manufacturer, will be based. Trump announced in March that the chip-making giant would invest at least $100 billion in US manufacturing.
“With a rich ecosystem of industry and manufacturing, we can make the US better on the other hand, but reducing dependency on other countries — only dependency, is a wise move,” Huang said.
The rise in AI investment, which has driven a massive technology boom in recent years, has raised concerns about whether technology will threaten future employment. A survey released by the World Economic Forum in January shows that 41% of employers are planning to reduce their workforce by 2030 due to AI automation.
With the market value quickly reaching $4 trillion, Nvidia has created data centers in its power data centers that companies like Microsoft, Amazon, Google and other use to operate AI models and cloud services.
“Whose jobs are affected. Some jobs are lost. A lot of jobs are created and what I want is that the productivity gains seen in all industries will grow society,” Huang said.
He explains that all NVIDIA software engineers and chip designers use AI and encourages that they “mandate it.”
Artificial intelligence tools, particularly generative response platforms such as Elon Musk's Grok and Openai's ChatGpt, have faced a significant portion of the controversy these days.
Last week, Grok began responding with posts after Musk's Xai tweaked the chatbot to allow users to provide more “politically wrong” answers. Among other graphic descriptions, I have begun to create anti-Semitic hate posts.
Xai posted a statement on Saturday that the “deprecation code” update made Grok more susceptible to existing user posts from X, including extremist views. According to the X statement, that code was subsequently removed.
Huang commented to Grok, saying that the chatbot was probably “young” but Musk “has made a lot of progress with Grok in 18 months.”
“Of course there's fine tuning, there's guardrails, and that's just taking Poland's time,” he said.
There were also concerns about the proneness of AI models to “hapticism.” In other words, AI models will fall out of scripts and spit out inaccurate information. Similarly, some experts have expressed concern about losing control of powerful AI models because they are susceptible to operations.
But Huang believes that “boundaries scare people” that don't know how AI systems are interconnected to keep technology safe. He explained that most AI models use other AI tools to provide resources and fact checking. He added that global standards and safety practices should be in place.
“It's overwhelmingly positive. There's some degree of harm. The world needs to jump on top of it when it happens, but it's overwhelmingly incredibly powerful,” he said.
Use AI in healthcare and real-world use
Huang said that diseases can be cured by using AI models to teach tools about the meanings of chemicals and proteins and chemicals that contain their interactions.
It's similar to the process of drug discovery, but more complicated than teaching AI about the language for the data needed, Huang pointed out.
“We not only accelerate drug discovery, but also improve our understanding of diseases. But over time, we will have virtual assistant researchers and scientists to essentially treat all diseases,” he said. “I think that day will come.”
There are also actual physical use cases of AI. Like Google's VEO 3, today's generative models can generate videos of physical behavior. The next step is to create a robot that can complete similar tasks, such as picking up glass. The process becomes a different Vision Language Action (VLA) model than a Big Language Model (LLM).
“There's technology here today. It works today,” Huang said, adding that there's a lot of technology in “three to five years.”
