'Anthropic Ban = Tariff Moment For Indian Tech': Why Diversification Is India's Best Bet, Again

[00:00:00]This is unprecedented, at least in the technology space.

[00:00:02]I don't think I can't remember that, you know, we've had a moment where suddenly this entire technology was access has been revoked.

[00:00:10]>> Well, it seems to be deja vu from when tariffs were earlier implemented. The same thing happened with Indian exporters. They were suddenly hit. Or when H1B fee was suddenly increased over overnight, which again sent Indian techies into a tizzy. So, would you say this was like a tariff moment for India's AI dependence and diversification is now basically the only option?

[00:00:30]>> Absolutely. I think in the short term the idea would be we use a combination of three, and but in the [music] long term keep working on how to improve our own our own models and own become much more self-reliant.

[00:00:41]>> HCLTech and Sarvam AI also signed a kind of first-of-its-kind deal, right? So, do you think that will kind of change the playbook for how Indian IT firms partner on AI technology?

[00:00:52]>> It's very exciting news that we are all I'm I'm sure like me a lot of a lot of us are thinking about how do we start leveraging that and start using them and making our own technology more successful. We are leveraging AI, but it is still we still I still monitor by humans, making sure the, you know, the output is right and, you know, because it it affects our customers, right? So, it's definitely assisting us. It and but it is it is not going to displace humans, right? So, it's it will create more opportunities. Cost to utilizing it it can go haywire and and if not properly, so you know, one starts questioning whether I should use this technology. Is it really benefiting me?

[00:01:43]>> Hello and welcome. You're watching Mint and I'm Sana Marwah. A few days ago the United States banned the use of Anthropic's top AI models Mythos 5 and Fable 5 for non-US users.

[00:01:55]Technology, which was the one thing that has always been globally pervasive, is now also being restricted.

[00:02:02]This is the first time that the government of a country that controls a majority of the world's tech has actually gotten involved in the access to private companies' technology.

[00:02:13]This highlights the Trump administration's push and their increased focus on tech sovereignty, the control, and the role of geopolitical risks.

[00:02:23]Now, the thing is, most companies in India are also highly dependent on foreign technology, especially US tech products.

[00:02:30]So, what does this whole situation mean for Indian companies and startups? And why did this move leave them rattled?

[00:02:37]And what's the road forward from here?

[00:02:40]To discuss this in detail, we're joined by Ashish Tandon, the founder and CEO of Indus Space. Thanks so much, Ashish, for joining us here today.

[00:02:49]>> Thank you. Thanks for having me over.

[00:02:52]>> All right. Well, first of all, could you help us understand what was the initial reaction of Indian companies after this ban was announced for the first time?

[00:03:01]You mentioned that the access was revoked within about 90 minutes. So, what exactly happened in and after those 90 minutes?

[00:03:08]>> No, it's Just think about it that, you know, you have uh you're using an application or a software, and in this case, uh you're using the the Anthropic models, and you have integrated all of them within your applications. You're using them for your uh enterprise functions and and all the things you created around it. And uh and and your business is starting to rely on it on some of the critical functions.

[00:03:33]And uh imagine within uh within within a short span of time, you are those accesses revoked. So, it it it it leads to business disruption, right? So, uh you're you're you're a lot of those integrations stop working, and uh and and and you have no idea and no reason, and they're not good enough time to plan an alternative and the stuff like that.

[00:03:55]So, that really that really led to a lot of dissatisfaction. You know, it has never happened like this, right? That you were suddenly an entire entire uh uh technology is uh is is not available to you.

[00:04:10]Uh so, I think that left a little bit of a uh you know, it was surprising and also it has uh it is it has created more questions and in in in the in enterprises and startups' minds than uh really answers. AI is supposed to give you answers, right? But this really created a lot of questions now in our minds as to what should be our strategy, what should be done, how would can you be how can you be safeguarded against those things in future, etc. >> Right. So, in this context then, how much exactly do Indian companies depend on this kind of US technology for their work? Because basically we understand that there are layers to how firms essentially build on top of US technology or foreign technology. And most Indian firms then add to those layers to kind of make it their own. So, can you explain to us how exactly that works and hence to what extent is the impact of such a move?

[00:05:01]>> No, right. So, you know, you would you know you know, you are one this is it most of the most of the enterprises and businesses are using is to assist uh the technology is used the AI technology used to assist them in certain functions of getting outcomes on on what are the what are the what are the what are the whatever they're offering as part of their stack or current product. Uh for efficiency, for uh for quality, for speed, etc., right? So, uh but you would then you are dependent on them because it becomes a part of your offering, right? Uh and uh and this technology itself is new. It's you know, AI has just been uh you know, the has been growing very very as an extensive pace. And especially uh this aspect of the model which is towards security and cyber security and we are using cyber security. It's supposed to be a superior model which can help uncover a lot of vulnerabilities and a lot of new security flaws better than other models. So you know that hype also led to a lot of us using this and almost zeroing down on them.

[00:06:06]And now with that not available, you know, one has to start looking at not that they were they were not many alternatives or they were there weren't alternatives but you know now the thought process is clear that you will have to have multiple alternatives for anything we do and do not and not have dependency only on on a particular model or you know and and maybe use a combination of open source, made in India or and you know, even if it has to be a international model, one has to calibrate and make sure that it's not it's not they're not always depending on one model. So I think that's the strategy.

[00:06:44]I'm sure going forward everybody will everybody will follow.

[00:06:47]>> Yeah, I was just going to ask you what would you say is the one lesson for Indian enterprises from a situation like this because it seems to be deja vu from when tariffs were earlier implemented.

[00:06:57]The same thing happened with Indian exporters. They were suddenly hit or when H1BC was suddenly increased overnight which again sent Indian techies into a tizzy. So would you say this was like a tariff moment for India's AI dependence and diversification is now basically the only option?

[00:07:13]>> Absolutely. I mean you know, all these examples you have quoted are you know, are have been kind of reminder kind of like a reminder to us that you know, we need we need alternatives. You need to become self-reliant.

[00:07:27]We need to look at options which can and and plan strategies which can help us overcome these these controls, right? I mean you know, you are now almost at you almost become at the mercy of a next country or a next technology and that is it's never good, right? In any in any any situation. So, you know, in having the confidence, the trust that you know, whatever is available, you know, will be available to you. And you this is unprecedented. This is the technology space.

[00:07:58]I don't think I can't remember that you know, we've had a moment where suddenly this entire technology was access has been revoked.

[00:08:06]So, so this is this is this is unprecedented.

[00:08:12]Yes, deja vu that other aspects also have been you know, there has been a continuous or a consistent approach. Uh And one needs to now think about this on how you can safeguard your own self as a business, as a country. Uh you know, and and how do you how do you how do you come out and emerge stronger through this? And I'm sure everybody's planning what is their plan B or plan C to make sure that you don't encounter a similar situation in future.

[00:08:39]>> Right. So, then kicking off from there, where would you say India currently stands when it comes to AI or building AI? Because we are of course late on developing the models themselves. But many experts seem to be optimistic that Indians have always kind of been better at optimizing the use of technology rather than, you know, being the first to kind of develop it themselves. At least that's been past precedent even with the internet, etc. So, how how do you feel about that? Like, where do you think India stands on that front?

[00:09:07]>> No, you're right. I mean, But but there are we are already seeing developments that the homegrown models are coming up. Uh which is a great sign.

[00:09:17]Uh there are open-source models, you know, which which uh specific verticals or specific areas, like in our case cybersecurity. We start using some of the open-source models, train them, use our use our you know, our database and knowledge and and and make it better.

[00:09:35]And of course, you know, continue to leverage some of the some of the frontier models. So, but I think I think in the short term the idea would be we use a combination of three and but in the long term keep working on how to improve our own own models and own become much more self-reliant and have the same kind of capabilities which any of the frontier models currently today have, which are international.

[00:09:59]So, it has to be a combination of three.

[00:10:01]Uh but I I and I think there is this is this is not a wake-up call. This is basically it it's just it's just accelerating what has been the thought process going on. Now, I think this is cable stakes. One has to get this thing done. And I'm sure you know, in my circles or wherever I am going this is definitely a thought process and and you know, and it it is required. So, I'm sure you will see a lot more acceleration. And as you rightly said, we have the talent, the capabilities. Uh but it just needs to kind of get uh we have to kind of move very fast.

[00:10:34]>> Right. So, now a couple of days ago HCL Tech and Sarvam AI also signed a kind of first-of-its-kind deal, right? So, do you think that will kind of change the playbook for how Indian IT firms uh partner on AI technology? Like rather than, you know, uh just relying on the foreign tech, they'll increasingly pursue a sovereign AI platforms. And do you think that that is one thing that, you know, will turn into the norm, especially with these uh big Indian IT firms?

[00:11:01]>> Absolutely. I mean, this is a great start, great example, you know, Sarvam and uh HCL getting together.

[00:11:08]Essentially, these models and these capabilities require investment, right?

[00:11:11]And I think uh it is good now that we see that uh the industry is getting together and and providing the providing the capital which is required, right? Because these are capital-intensive uh projects and uh and and uh and and see more and more investment coming in.

[00:11:27]Uh you know, and this should this This this is just going to help the help everyone in the everyone uh in the ecosystem.

[00:11:36]But we need more and we need some more of more of that to happen because we need a few more and and absolutely, you know, one would when you build a strategy now that you definitely make sure that you know, you will also make sure that you kind of use some of your home grown models and which are easier accessible. You don't have any of these you know, any of these uh uh apprehension software, you know, whether the model will be available or technology available or not. So absolutely each each one about it's very exciting news that we are all I'm I'm sure like me a lot of a lot of us are thinking about how do we start leveraging that and start using them and making our own technology more successful and so that more and more investments happen and more and more such models come and we kind of we kind of are able to overcome this challenge quickly.

[00:12:27]>> Right. Okay, moving away from the specific example of this particular crackdown of AI exports to a more generic sort of scenario on the AI situation. Many have been flagging that AI valuations are extremely high right now. The firms which AI firms valuations. So some experts seem to think that there's going to be an AI bubble burst at some point. Is that likely according to you and if yes, then what would that mean for the Indian technology sector like good or bad?

[00:12:57]>> Whether it's a bubble or not, time will tell.

[00:13:00]The technology definitely is relevant is heavily is definitely very very impressive. It really helps. There is no discussion, no doubt about that.

[00:13:10]I think uh my my thinking is that the current because it's heavily funded and the the idea behind uh the models is to kind of start making a lot of people hooked on to it, start using it and I feel it's currently subsidized. Uh and uh and and chances are that the uh the cost of using it might keep going up. So, uh I think I feel that would be a more uh I I think that would be a more uh possible scenario that the then the costs of using the technology might go up because cur- uh you know, the current chatter is that it's highly discounted, right? So, uh whether it's a bubble or not, it will keep I would say I think it's more the market should decide, but the technology definitely is very useful and uh and here to stay.

[00:14:01]>> All right. Well, one last question.

[00:14:03]Since you spoke about subsidies, are do you think and you spoke about cost, are these AI models at this point uh ending up costing more than humans? Like, where do where does Because, you know, India has a relatively younger population right now. We have a workforce. So, is there any view you have on this whole AI versus jobs debate? Like, is it going to end up costing more? Like, do you think ultimately, you know, things will just kind of reach this equilibrium where we just learn to be comfortable with AI rather than maybe, you know, pushing it into every sphere like a lot of company heads seem to be doing?

[00:14:38]>> Again, I don't have a magic wand, but uh from my experience, I definitely AI is definitely a very, very efficient and effective technology for improving a lot of your processes, a lot of your capabilities. You're able to do things at a much faster pace. So, there are all these benefits. Uh at least in our space, we are not seeing that we are able to lose or uh or we are able to uh displace humans. I mean, in our in our space, that is in cyber security or application security, uh we are uh leveraging AI, but it is still uh we still I still monitor by humans, make me sure the you know, the output is right, and you know, we because it's it affects our customers, right? So, uh it's definitely assisting us. Uh it and but it is it is not going to displace humans, right? So, it's it will create more opportunities and hence you you know I don't see us I don't see this that this is going to eliminate jobs. I feel that the that the skills and the job capability have to have to evolve with this. And they'll definitely be a requirement and they will continue to be there.

[00:15:46]I somehow don't see this at least in my space I don't see this completely automated or or AI taking over taking over doing the entire end-to-end function by itself.

[00:15:58]In terms of costs uh you know, again you need to know how to use it. It definitely can if you don't if you if you if you if it's not used efficiently, that is one discussion and chatter continuously happening that you know, the token cost or the cost of utilizing it can go haywire and uh and if not used properly, so you know, one starts questioning whether I should use this technology. Is it really benefiting me? So, a lot of those discussions are already on.

[00:16:31]Already in the in the you know, we see we hear about them very often not only here even in even countries like US where there are large organizations that are using it, deploying them. They are questioning whether the outcome of of deploying this technology is actually cost-saving or is actually reason for the cost. So, so I think that's that's we will see it will as time goes by I think people will start using it efficiently and and figuring out the right equilibrium or the balance which is using AI and humans. I don't see either one going away.

[00:17:06]>> All right. Well, that's that then. We have some key messages out of this.

[00:17:10]Thank you so much Ashish for joining us here.

[00:17:12]>> Thank you so much. Thanks for having me.

[00:17:15]Pleasure talking to you.

[00:17:23]>> [bell]