China is six months behind the U.S. on AI, the US has to move faster: Div Turakhia

Divyank Turakhia is a tech entrepreneur and investor based in Dubai. He has launched and sold a number of innovative companies, of which the most well-known is Media.net , which he founded in 2010 and sold to a Chinese consortium for $900 million six years later. Div is originally from India, where he grew up and started his entrepreneurial journey when he was just 14 years old.

Div spoke to Rest of World ’s editor-in-chief, Anup Kaphle, at an event organized by Rest of World , in partnership with Luminate, that featured a series of conversations about the role of AI in emerging markets.

The following interview was prepared from the transcript of that conversation, and has been edited for clarity and length.

Div, one of my favorite stories about you is that you learned to code when you were 8 or 9 years old. Today, you’re a serial entrepreneur and now you’re working on your fourth internet business — Ai.tech. Why did you want to start an AI company?

I thought I was very fortunate that I found what I loved doing very early. And so the journey started early. That company you mentioned is the fourth one. And after the third one, which I did back-to-back, I took a long break, which was awesome.

How long is a long break?

I’d said it was a two-year minimum; it became a three-year break. And then started the fourth one. I built that as a holding company from which we can build and incubate multiple businesses. AI became a word that everybody knows now, starting around November 2022 when ChatGPT launched. And everybody thought it was awesome. I’d started playing in the field a long time ago. Just before I launched Media.net, we wanted to do something called contextualization of large amounts of content. And at that point in time, there was nothing like AI. We were using CPUs to do better matches across a lot of content, as we browsed the internet. So I had that exposure for a long time. When building any business, if you want to be successful, you want to identify what your core strengths are, you want to double down on them. And I realized over a period of time that my core strengths were deep tech and operational efficiency, which is picking up any process and figuring out how to make it cheaper, faster, and more scalable. I felt like AI is going to be in everything. We’re in the very early stage of understanding the uses. But over the next 20 years, it will be a part of everything and will be almost invisible.

What’s the one AI use case that you’re most excited about?

I’d say efficiency. If you think about what it is that AI does for you, it adds an efficiency layer, which boosts productivity. All the folks that I’ve spoken to, in surveys that I’ve read, with today’s tools like ChatGPT, whoever is using them either in coding or in writing something or analyzing something, you’re seeing somewhere around a 10% to 15% boost. I think that is the very beginning of what we’re going to see because two separate tracks happen: One is, the large language models become much smarter by using a lot more data. And the training costs of these large models are insane. They started it for $10 million, now every large language model costs over $100 million and will soon be billions. The second track is small models, which are far more specific and we’ll see a lot more small models used across a variety of applications because the number of large models that can exist is a handful. They’re not going to solve for every use case.

One of the first things people talk about when they hear about AI is what it means for job security. What happens when AI comes to outsourcing countries like Kenya or the Philippines?

Or India. We’ve seen this time and again: Each time there’s an efficiency tool that’s used in the world, everybody thinks [about] what happens to their jobs. Humans are innovative enough. They’ll figure out how to use and incorporate that additional tool. Whether you’re talking about the Industrial Revolution or the move from horses to cars or from cars to planes, or from no computers to computers. Different economies will have to figure out what their core strength is. I’ll give you India as an example: India historically has not been the best at infrastructure or infrastructure investment, so I don’t really expect India to build large data centers. They have built data centers, they’re very proud of 10,000 GPUs, and 10,000 GPUs doesn’t really move the needle if you’re basically looking at some sort of large language model that is going to be important enough.

At the same time, it has 1.4 billion people — it’s great human capital. Because the large models are not going to solve everything, they’re just going to be tools to use to incorporate into one of the things. If you look at the current Indian outsourcing companies, like Wipro or HCL, all of them are advising companies around the world on how to incorporate AI into their business processes. You will see a lot more of that come from an outsourcing place where they have learned the new tools. Because they have a lot of tech-based knowledge already, they have the advantage that they can learn new stuff faster, and they can help businesses just like they’ve been helping them in the past with other things.

You are now based in Dubai, and you’re also a member of the UAE government’s AI Council. Companies have been on a huge spending spree when it comes to investments. Can you tell us what it is that countries like the UAE are hoping to achieve?

I can talk more about the UAE specifically because I’ve been there for a long time. The UAE started its AI journey earlier than most. It appointed an AI minister in 2017. The next country, to my understanding, that appointed an AI minister was South Korea in 2020, three years later. If you think of the UAE specifically, it’s only got a population of about a million people that are local. It is a resource drought because one of the Emirates, which is Abu Dhabi, has oil money. The rest of the Emirates have to figure out their own things. In the case of Dubai, it only has access to 3% of the oil resources. So it pivoted very early to tourism and did so very successfully. And Dubai is very good with infrastructure. The second thing about the UAE is that it is a smaller version of what America is in some sense, because 90% of the population is expats. So it’s got 10 million people in the country, 90% of the people are people that have moved countries to find opportunity. As soon as you’re talking about a population that moves countries to find opportunities, these are people that generally have a bit more of a hustle, generally want to get a little more done, generally want to push the boundary a little more. And so they adopt newer technologies or figure out how to capture newer opportunities faster. And so, from the UAE’s perspective, every time that there is something new that is game-changing, they want to incorporate that. Hence, they are already in AI.

The other bit around that is sovereign funds. The UAE has one of the largest sovereign wealth funds. It’s not just one — they have more than eight of them. Together between Abu Dhabi and Dubai, they manage over $2 trillion. These sovereign wealth funds have been investing around the world with a lot of the usual private equity funds like Blackstone or Apollo or whoever else has been raising money from sovereign wealth funds and hedge funds. They see these trends early enough where they feel like this is a place where they want to invest heavily because it’s going to have an outsized economic return on the investment. That will be the other reason.

In addition to having a smaller population, which means AI is needed to achieve many things because of the automation and scale, there is enough money to basically make that investment. The third bit is energy. Any large AI investment requires a lot of energy. And the advantage that the UAE specifically has is, because it was good with the infrastructure, it had always overplayed that infrastructure. So both Abu Dhabi and Dubai have spare power, which is not the case in many countries, and that spare power can be used to build large data centers and kind of add scale process, larger models that can be built.

What is it like watching from the Gulf, as the U.S. and China compete to own the backbone of AI? And who would you put in charge?

From whatever I’ve seen and read, my understanding is that China is about six months behind the U.S. — it’s not very far behind, which is concerning. So the U.S. needs to do more to figure out how to move faster, how to ensure that it’s not overregulating in any form so that it can move faster because China doesn’t have that problem. It has a different kind of problem. The advantage of being in the UAE is you get to see both sides. Because the UAE is so small, the advantage it has is that it can be friendly with everybody. And it is. It’s one of those unique places where 90% of the population is expats. Within that population, you will see Russians, Chinese, Americans, Europeans all hanging out together and nobody cares about the greater politics. People care about economic opportunity more than they care about the immediate politics that are happening.

That’s how I think. I think China’s really good with hardware, the U.S. is really good with software. You see those differences. China produces at scale, it’s really hard to do that outside of China, and obviously the U.S. has realized that that’s a mistake and is making investments in as many additional places as possible, so that it does not have that single point of failure issue with scaled hardware. The UAE has made a choice in this already because it was asked to make a choice and it chose the U.S., as far as AI is concerned. So all of the UAE’s platforms are U.S.-only and they don’t use any of the China stuff.

Last question. What’s one thing about the state of AI that keeps you up at night?

Nothing keeps me up at night; I sleep very well. I’d say that I think of it positively. Yes, there are challenges with any new tech, and yes, it can be used for bad stuff. That’s the case with everything that gets launched. On the positive side, there’s so much efficiency boost for humanity already. If you read the news every day, you think we’re in the worst time ever. If you don’t read the news every day, you think technically we’re actually in the best time ever.

I grew up in India. Just 70 years ago, I would not have the same luxuries that I have today. That applies to everyone. Most people have better quality of life, and with AI, the quality of life will improve very significantly as we keep thinking about how to make it more cost-effective because it’s not cost-effective today. It’s not something that everyone can afford today. In fact, I was looking at a stat: ChatGPT only has 11 million paying users. That basically means that everybody that is not a paying user doesn’t have access to the best models and doesn’t have access to that level of efficiency that can be generated. I think, with time, that will obviously change. That’s exciting because I think it improves the quality of life for everybody and it improves pretty much everything, including relationships between countries because as things move fast and things get more automated, things get more efficient. And that’s a good thing. ▰

Anup Kaphle is the editor-in-chief of Rest of World. He is based in New York City.