Is the H-1B visa controversy just a distraction from US' real immigration problems?

The fierce row over Indian immigrants and H-1B visas that's roiling US President-elect Donald Trump's team even before he takes office misses the point and isn't borne out by data, said Aravind Srinivas, founder of Perplexity AI.

Illegal immigration is the real issue, said the IIT graduate, who's worked at DeepMind and OpenAI, and is himself in the US on an H-1B visa.

"Clearly, it surprised everybody online and it's not exactly data driven -the claims that the H-1B or Indian immigrants are taking away the jobs of laid-off American tech workers," he told ET in an interview. "Indians are also being laid off and honestly, the real problem... is illegal immigration. That's what the government needs to cut down on."


The issue has polarised Trump's top supporters, with the opposing camps being dubbed MAGA (make America great again) and DOGE (department of government efficiency).

Elon Musk and Vivek Ramaswamy, who will head DOGE, the proposed cost-cutting agency, have championed H-1B visas, while Steve Bannon and Laura Loomer are MAGA supporters, who have accused their opponents of being billionaire tech bros, with little interest in the welfare of the Trump base.

Srinivas also said it was unfair to drag Sriram Krishnan - Trump's pick for artificial intelligence (AI) policy adviser- into the dispute by citing his past tweets supporting H-1B visas and immigration. "We should all feel good about someone as high calibre as Sriram influencing AI policy," he said. "People need to understand he has nothing to do with immigration, and stop taking his past tweets out of context!"

The Perplexity founder, however, expects tempers to cool. "If people actually took a look at the data and saw it for themselves, they wouldn't be this angry," Srinivas said. "I feel it's going to all settle down."

In an extensive conversation with ET, Srinivas also spoke about AI's evolution from "question answering" into becoming a true "executive assistant," an "agent" capable of performing multiple tasks, all in the background.

'Worth Paying for AI'
"Question-answering is very hard to personalise because most of the time, you get roughly the same answer. (That is,) until it starts doing things for you and, over time, figures (things) out, and begins to feel like a true executive assistant. When that happens, you're gonna have a pretty different relationship with every AI you use," said the Perplexity founder. "That'll become a big theme next year. When AI begins to take actions on your behalf and gets more and more personalised, you will begin to feel you do have an assistant - that's going to be very powerful."

This is set to enhance AI firms' monetisation capabilities. "I think this is going to be the beginning of what everyone's been waiting for - when is AI really going to make a ton of money?" Srinivas said. "Because the data centre build-out costs are just insane. The revenue being made in AI right now is so much lower than the amount spent to train these models. As you can see, OpenAI loses more than $5 billion a year. The true promise is going to come from people realising it's actually worth paying for the AI."