India is our fastest growing market, doubling every year, says Pure Storage CEO Charles Giancarlo

Data solutions and platforms company Pure Storage is on a high. The NYSE-listed firm’s stock price has been skyrocketing as it kept beating street estimates on revenue growth. This 15-year-old corporate entered India in 2020, at a time when data centre expansion by global tech giants had just started taking off amid changing online behaviour during the pandemic.

Today, India is its fastest growing market worldwide.

In an interview with Moneycontrol, the company’s Chairman and CEO Charles H. Giancarlo spoke about the company’s outperformance, its India expansion strategy, the US government’s IT spends, AI’s impact on the data centre and storage industry, and more.

Edited excerpts:

Your stock prices have shot up nearly 80-90 percent this year and recent revenue performance has beaten expectations. What’s driving this?

A number of things. We're the one company that considered data storage to be high-tech rather than a commodity. A lot of the investment, a lot of the innovation that we've  put in place over the last several years has revolutionised the industry. That has helped. We're the only data storage company which has been growing over the last several years and acquired significant market share.

Recently, for the first time in the industry, we acquired a major hyperscaler as our customer. To put that in context, the four or five major hyperscalers have all built their own storage.

This is the first time a hyperscaler is going to use a system supplier. We provide systems for their storage. And we're the only company that  doesn't use SSDs (solid state drives) for flash storage. We have something called direct flash, meaning that we manage the flash chips. It's the same storage you have in your phone.

So until now, they bought raw hard disks, that's 90 percent of their storage, or raw SSDs, that's maybe roughly 10 percent.

In comparison, we come in at a lower price because we use flash more efficiently compared to a hard disk. We need a 10th of the space, power, and cooling. That helps the hyperscaler use their existing data centres for expansion without having to build new ones.

Corporates are thought to have been cautious with their tech spends and even AI experimentation in the past couple of years. Has the mood started to shift with Trump’s victory? What’s next?

I can't say, it's a little early. We're all hoping for an improvement in the overall economy. The IT industry has been stressed over the last year or two. There was a lot of spending during Covid, but that's pulled back a bit. Also, there was a lot of price escalation in the SaaS and software space, and that impacted all the other spending. Much of that was unexpected.

As we go into the new year, I think we'll see prices stabilise, which will allow additional spending. But hope is not a strategy. We have to wait to see if the scenario improves.

There was news that the Department of Government Efficiency (DOGE), led by Elon Musk, was likely to bring down the US government’s IT spends, which are expected to cross $100 billion in 2024, and the government has already allocated $75 billion for 2025. Do you see this having an impact on IT services companies like Accenture? Will this increase demand volatility for the technology services sector?

It's a complex topic. If you ask any American, they would say there's a lot of wasteful spending by the US government. But a lot of that is on labour, which is not very productive.

A study has shown that few government employees are actually coming to  office. So I view IT spending as a solution for more efficient governance. And when you get someone like Musk and other tech leaders, I think they would agree that we need to make the government more efficient.

Greater  efficiency does not come by spending less on IT. It's not necessarily spending more, but spending better on IT.

Pure Storage entered India right before the pandemic, which is when we started seeing the rapid expansion of data centres in the country. How has the market matured since? What’s been the demand like for Pure Storage?

We've been scaling very rapidly and have more than doubled (in India) in the past year. The plan is to double again next year. You can't say that Covid was the right time to do anything, but I think we came in right at the right time. It's not so much the spending environment, it's not even so much the building of data centres. What's really happened is that India is looking at doing everything in a new way because they're not challenged with a lot of old infrastructure.

You have an technologically educated population, and a very vibrant, high-tech community. You can take all the new techniques and start building them into your services, your infrastructure as you go forward. The community that I speak to is so excited about taking advantage of all that and creating something that's efficient and customer-focussed.

We're coming at the right time because storage is a very old and conservative industry. So, being able to speak to people that are looking to change the way things are done is very refreshing for us.

What is the India opportunity looking like for Pure Storage in the coming years? How much revenue contribution do you see coming from India?

We have three major development facilities around the world. When I started at Pure Storage a little over seven years ago, we were 100 percent US West Coast. Since then, we've opened global development centres in both Bangalore and Prague in Czechoslovakia.

In fact, our VP, Engineering, in Bangalore heads one of our two major products globally. He's responsible for the teams worldwide that focus on the product, and accountable for its performance. We're more than doubling our investment in our sales and marketing in India.

India is  our fastest growing country and we're very excited about that. I expect India to grow and double each year for the next several years. Though right now it's still pretty small on a global basis.

But if we continue to double, I can see India vying with some of our major European geographies, which would be great because right now it's just Japan, France, Germany, and the UK that compete for the number one spot every year. I'd love to see more competition.

How big is the Indian team? Do you have any expansion plans?

We've been growing our India headcount by more than 50 percent a year for the last four-five years.

On the enterprise business side, how have the compute costs and energy requirements trended in the last two years with AI experimentation taking centrestage?

We have been focussed for many years on our performance from an efficiency standpoint. I've already mentioned that we require a 10th of the space, power and cooling of a hard disk. We're also at least twice as efficient as our flash competitors.

Our customers don't necessarily have to expand their power requirements in order to be able to put in AI or more compute environments. That's really nice.

Storage is the number two user of power in a data centre. If we can reduce it by a factor of five or 10, it means that all of that power can be used for their expansion rather than having to go and build nuclear or other power plants.

Who are your customers in India?

We have two of the top telcos. We've got a number of the top banks, and also manufacturing and transportation firms, many of the large companies here. But we're still on day one. We're  yet to  get to the point where we have all of the customer's business because we're the new vendor. They still have a lot of the legacy vendors in place, but we expect to continue to grow.

There were media reports that OpenAI and Google have been struggling to further develop  advanced AI models as returns and revenues have diminished. How will this  impact computing / data loads on your infrastructure around the world?

The rate of improvement has slowed down dramatically. I think  we're going to see some  deflation of the AI bubble, because the current models have effectively  reached competency saturation.

Before LLMs (large language models), AI was based not on statistical models but on algorithms. They'd take facts, and relate them to  each other. So you could always reverse engineer and see why the model  came up with an answer. For example, protein folding models are based on physics, they train the AI on the physics. So the physics will determine what the answer is. If the answer is wrong, they can go back and see which part of the physics was incorrect and fix that. And then it'll come back with a better answer.

LLMs are entirely different. It's all based on statistics. Like Mark Twain (the author)  said, there are lies, damned lies, and statistics. The problem with LLMs is that they're first and foremost based on  algorithms and models that mathematicians have come up with, and then the data.

If the algorithms don't improve, if the models don't improve, then eventually the data won't make it any better. The big change was that about two years ago,  several new data models  were created which enabled a dramatic improvement, but since then no new models have come out. We've kind of reached the end of the improvement cycle for LLMs right now.

What’s going to be your key business focus areas in the next two-three years in India and globally?

The Indian market is much more open to doing things differently than some other markets in the world that have a lot more legacy in place. So one of the things we want to drive here is establishing your own data environment, which looks like a cloud rather than  individual arrays. That's a big opportunity. We think AI will be a major driver in that area.

The second, which we've innovated on, is instead of having customers actually buy storage arrays and manage it all themselves, we offer storage-as-a-service. Now, you might think of storage-as-a-service as exclusively cloud based. We  provide something very similar, but the storage can be on the customer's premises. It doesn't actually have to be in the cloud. And we then manage all of that remotely in a completely automated fashion.

The customer interacts with it through APIs and a GUI (graphical user interface), just like they would with a hyperscaler (such as Amazon Web Services and Microsoft Azure). We guarantee the uptime and performance. The customer only monitors via SLAs (service level agreements). We think this a much more modern way to operate and we hope to promote that here.