Delhi-born Vivek Mohindra is the Senior Vice President, Corporate Strategy, Dell Technologies and works in close partnership with the offices of the CEO and COO of the Texas-based tech major to develop and drive strategies that ensure Dell’s continued market leadership.
businessline spoke to Mohindra during his recent visit to Mumbai about how Dell is looking at AI developments and the challenges he is seeing on ground when it comes to using AI.
How is Dell thinking about developing and deploying AI?
When we think about our AI strategy, we think about it in the context of our business strategy and that’s the approach we took to deconstruct it.
There are 4 elements of AI – “In, on, for, with.” The ‘In’ part is about how we put AI into our product for customers tp use it much better and much more seamlessly. The ‘On part is, what are we doing with our products for customers to extract those benefits? And the ‘With’ part is, how do we deliver it to our customers with partners and finally, how do we use AI for ourselves and think of ourselves as customers zero to make sure that it accentuates our business strategy.
So, we have very thoughtfully taken the four parts and prioritized the used cases along these different dimensions.
As customer zero, what percentage of your business is now completely AI driven?
The first step we took as customer zero was yp thinking about – where should we apply these tools? That was part of our business strategy lens that I just shared. The second thing that we did was to look at where the start-ups and VCs are investing.
The third part was benchmarking, based on the early evidence, where are our people are seeing the benefits? In terms of percentages, in terms of software development and all the other areas that we talked about, generally speaking, you can get anywhere between 10 to 40 per cent better productivity in these processes.
As customers zero, we had to figure out which use cases., what particular data you need to use to be able to mix and take advantage of these use cases. You also have to start by simplifying, standardizing these processes for a global company like ours.
So, we started this journey last year and we are at different stages of this journey.
Has this helped Dell in helping customers because a lot of companies are yet to figure out use case for AI
Some of the customers who were very thoughtful about which particular use cases they should focus on, and how to objectively measure the outcome of the initial proof of concepts (PoCs) are seeing phenomenal benefits.
But vast majority of the companies are not thoughtful about where they should apply the use cases because in many cases either the CEO or the Board or both have said AI is everywhere, you have got to do something and they said, “Well we got to do something. Let’s jump in.”
Now, they have run these ill-defined PoCs. They have not seen the benefit of AI. What we did is to help these customers, understand and run the right POCs because if they don’t do that, they will be left behind, without even realizing it.
My point to a lot of customers is if you are having those discussions, I totally get it. It’s very understandable. But if you haven’t started, then you are already late. And guess what? All the other competitors who have started in the right way are already beginning to see the benefits and deploy it.
Most companies still consider tech spends as discretionary. If AI is so critical to cutting costs and enhancing productivity, why aren’t corporates thinking about tech spending as core to their business?
Companies and their boards doing that are being short sighted in my view. What we are beginning to see is that companies who are doing it right along the lines of what I described earlier, have increased conviction to think about IT spending not as a discretionary spending but as a legitimate investment they are making, both in terms of productivity and growth.
When I look at the statistics of companies who are either doing POCs or moving from POC to production, those percentages are systematically moving up. In some of the more recent surveys, I have seen that number is as high as 59-60 per cent.
Now do I believe that 59-60 per cent of global companies are don’t see IT as discretionary? I am not sure about that, but I think, my personal view is that it is increasing at such a rapid clip and more and more companies will start coming out in quantifying it in their earnings call.
Some are beginning to do that which will actually build a momentum for their competition to also move into not thinking about IT as discretionary. But it’s inevitable. It’s going to happen.
Deploying large language models requires huge amounts of energy. Do you see small language models as a better alternative for most use cases ?
Everybody talks about LLM but it’s not all about LLM. It’s also small language model and purposeful language models.
Right now, the compute distribution for AI is 90 per cent training and 10 per cent inferencing. When you look at various estimates out there and how the companies will really strike the benefit from AI, it’s all going to be around inferencing.
By the end of this decade, the estimates are 70-90 per cent of the compute will be for inferencing, not for training. When you think about inferencing side of the equation, you have far greater number of tools to play with. You don’t have to think about having a major GPU cluster, you can think about racks within our data centre, and you can think about PCs and AI PCs which will be able able to orchestrate some of these inferencing workloads as well.
How do you see opportunities for India?
We continue to view India as a hugely important footprint for us to be able to fulfill our AI strategy. From a market perspective, India has phenomenal growth exhibited already and we are very bullish on India’s potential going forward.
I genuinely believe that a lot of these AI trends are moving in India’s favour and this AI revolution could truly be the future of Indian economy.