AI changed when software finally got the hardware it needed: NVIDIA’s Vishal Dhupar

For many years, the popular understanding of artificial intelligence (AI) has been shaped by the idea that it is simply software that automates or speeds up tasks. But Vishal Dhupar, managing director of NVIDIA’s South Asia division, suggests that this view misses the real point in the development of artificial intelligence.

The breakthrough came not just from lines of code, but from the hardware that made modern neural networks work at scale, Dhupar shared during a fireside chat at TechSparks 2025with Shraddha Sharma, founder and CEO of the company Your story.

He explained that the relationship between software and hardware is interdependent, using a simple analogy to understand this.

“Think about it. Everyone says fuel is important, but if you didn’t have an engine, how would you consume it? Now everyone says you have an engine, but if you didn’t have fuel, how would you start it? It just depends on how you look at it,” Dhupar noted.

For decades, researchers knew what they wanted from neural networks, but lacked the computing power to make those models learn well enough to be useful. Algorithms were not new; there was an opportunity to train them.

“For 60 years, we kept finding solutions to the problem of perception. We built neural networks, we had enough data, we kept trying,” he said. “And suddenly a bunch of smart people in Toronto figured out…a computer would write software that no human could write. And that was solved with GPUs.”

The turning point was not accidental. NVIDIA was already building CUDA (Compute Unified Device Architecture), its parallel computing platform, long before the commercial benefits were clear. At the time, it seemed like an expensive, unnecessary bet to many.

“CUDA was introduced at a time when everyone was saying that you as a company are not doing well, why are you increasing costs. For six years we had a burden trying to get CUDA up and running,” he recalled. It wasn’t until deep learning made its way to hardware in 2012 that the industry realized what had changed: “All humanity has rationalized.”

Industrialization of intelligence

Dhupar described the current moment in AI not as a technological upgrade, but as a shift in how intelligence itself is produced and distributed. Instead of simply storing and retrieving information from databases or search engines, AI systems now generate new answers in real time. This means a shift from accessing information to producing intelligence.

“Today we actually ask a question, the system thinks and generates a new answer that sometimes surprises, increases intelligence,” he explained. “Until today, we mostly extracted information from web pages or databases. Today, the system generates.”

This new capability has changed the role of data centers. They have evolved into what Dhupar calls artificial intelligence factories: not just storage centers, but facilities that actively create intelligence.

“Data centers have become factories of artificial intelligence. And these factories of artificial intelligence mainly produce tokens. And tokens are the new currency. And most importantly, it is the industrialization of intelligence,” he said.

India’s AI capabilities

Dhupar argued that India has a decisive advantage in the global AI race: a unique combination of data, talent and domestic demand. But these assets will only translate into leadership if the country builds the necessary infrastructure to turn data into intelligence.

“The end game is how we serve 1.4 billion people who officially speak 22 languages,” he said. “We have our common sense, our sensibility … It can’t be done from the Western world. It has to be done here.”

The basis for this, he emphasized, are artificial intelligence factories: national-scale computing power that allows the country to train and deploy models adapted to the Indian context.

“You require AI factories to produce tokens and have a network to bring intelligence to each of us. Just like electricity needed a network to reach every household.”

He clearly outlined a roadmap: rapidly building computing infrastructure, cleaning and streamlining datasets, and encouraging society to embrace AI as a productivity multiplier rather than a threat.

“The first thing I’m going to do is speed up the pace of infrastructure… The second thing is to speed up the data sets and clean them up… The third thing is to fully embrace it,” he said.

Export of intelligence

Dhupar also emphasized that India’s longstanding reputation as a software exporter will continue to grow. Instead of exporting manpower and code execution capabilities, India can export intelligence itself: AI models, systems and solutions shaped around its scale and constraints.

By building models that take into account linguistic diversity, accessibility and infrastructural constraints, India is creating technologies that naturally serve much of the Global South. Countries in Africa, Asia and Latin America have the same restrictions and will adopt India-developed artificial intelligence, just as many countries already accept UPI.

“Your IP can be exchanged with countries like ours. And so you will be a benchmark for the global south,” he said.

This marks a shift from a global back office to a global source of intelligence production.

“We are becoming a nation that not only exports software, we export intelligence. People come here. And so we continue to become the intelligence capital of the world.”