Nyobolt And The Power Bottleneck Inside Modern AI Infrastructure

[00:00:04] - [Speaker 0]
What happens when AI ambition runs head first into the limits of power? Well, today, we're gonna talk a lot about models, agents, and automation, and, yes, real time intelligence. But underneath all of this hype sits a quieter constraint that's starting to dictate what is and what isn't possible and possibly something we all take for granted. I'm talking about energy and power, not as a background utility, but as a deciding factor on whether AI systems stay online, scale efficiently, and deliver on their promises. So today, we're gonna zoom in on that reality because my guest is gonna unpack the growing power squeeze inside AI infrastructure, robotics, and industrial automation.

[00:00:55] - [Speaker 0]
He is a senior leader at NioBolt, which is a company focused on high power, fast charging energy systems that are designed for environments where downtime is simply not an option. So you'll hear why AI training and interference workloads can cause a power demand to spike in milliseconds and how those simple swings can put stress on grids and operations and why oversizing infrastructure has become an expensive default response. But he will explain what's happening inside warehouses, factories, and data centers right now, places where robots and automation fleets lose productivity to charging cycles, thermal limits, and insufficient battery design. So if you are someone that's responsible for AI infrastructure, strategy, or any operation that depends on always on performance, today's episode will hopefully challenge how you think about power, uptime, and scale. So let's get into it.

[00:02:02] - [Speaker 0]
So let me officially introduce you to my guest now. So a massive warm welcome to the show. Can you tell everyone listening a little about who you are and what you do?

[00:02:13] - [Speaker 1]
Thanks, Neil, and my pleasure to be here on on your podcast today. So my name is Ramesh Narasimhan. I'm with a company called Niobold. I'm their executive vice president and and their chief commercial officer, and hello to everyone.

[00:02:30] - [Speaker 0]
Well, thank you for sitting down with me today. One of the reasons that I invited you to join me on the podcast is to talk about the power squeeze inside AI infrastructure. So tell me a little bit more about how AI training and inference workloads are now so dynamic that power demand can swing dramatically in milliseconds, and also why this matters.

[00:02:53] - [Speaker 1]
Yeah. I mean, you know, AI itself, you know, the the number of AI data centers that that are being built is creating a huge stress in terms of the overall power requirement. And and then when you look at specifically the AI data centers, they have this pattern where it creates massive swing in terms of power change, and it happens in milliseconds. And what that does is it puts a huge pressure on the grid and has the ability for the grid to actually fail and also create powers power fluctuations around the communities where it's based in. So it's got a number of repercussions because of those traumatic swings as as you just said.

[00:03:46] - [Speaker 1]
And what we have done is we have come up with a solution on how to address these millisecond surges at a a data center rack level. And what it does is you could have the right sized grid capacity. I mean, we as as I mentioned earlier, you have huge capacity requirement already. And if you have to then build in for these surges, it's just expensive and also time consuming, and it's probably not physically possible to anticipate the surges to that extent. So when we eliminate it, you can right side right size the investment, for data centers and makes it more operationally more efficient.

[00:04:37] - [Speaker 0]
And I'd love to dig a little bit deeper on the problem here and bring it to life people that could be listening around the world. So if we talk about something like the hidden cost of automation growth, tell me more about how, let's say, robotics or AMR fleets in warehouses, because they're scaling quickly right now, but those operations still lose significant productivity to charging cycles, thermal cutoffs, and oversizing. Right?

[00:05:03] - [Speaker 1]
Yeah. Absolutely. I mean, you know, there is an increasing move towards, you know, sustainable solutions within robotics and AMR fleets, I e, electrification. But what that does also, is the the electrification has a need for charging the the robots at a certain point in time. And the speed of charge and the speed and the battery's ability to accept that speed of charge becomes a very critical part of the solution.

[00:05:33] - [Speaker 1]
Otherwise, it ends up where the equipment has a a much larger downtime leading to inefficiencies again and having a much larger infrastructure than what is required. And that's that is the the biggest challenge that that you have. For example, I won't name names, but one of our customers is in the is in this space. And their current solution effectively means that they can use the asset only for 50%. So you can imagine the inefficiency that you gotta carry to have an asset utilization of 50%.

[00:06:18] - [Speaker 1]
So that's, one of the things that we address with our, power solution.

[00:06:25] - [Speaker 0]
And across sectors, power is no longer just a a background utility. So tell the listeners a little bit more about why energy systems are becoming somewhat of a strategic priority now.

[00:06:38] - [Speaker 1]
Yeah. I mean, you know, power is a critical part of the growth that we are seeing in AI in the in the automation that we just spoke about in terms of robotics and AMR. And the the the key is going to be, I mean, this is gonna drive the performance for the the data centers and and the automation and everything else. And and looking forward, this is going to be a critical infrastructure. And when when we talk to our customers, a lot of the companies are worried about this.

[00:07:17] - [Speaker 1]
They are looking at the resilience of their supply chain, especially around power, and, and also cycle life of the of the battery then, therefore, becomes very, very critical. And the speed at which the industry is moving is phenomenal, and you you would have read probably the number of data centers that are being built. And if you specifically look at AMR and automation, if you're interested in that industry, you'll see the same trend. So power has become kind of the the critical infrastructure that backs all this up.

[00:07:57] - [Speaker 0]
It really has. And I I think it's something that we're hearing more and more about now. And companies in everything from logistics to automotive manufacturing. And, of course, AI infrastructure wants systems that can stay online continuously with high availability. So I'm curious from what you're seeing here, the conversations you're having with some of these organizations, are you seeing a shift towards high power, high uptime operations?

[00:08:24] - [Speaker 0]
Are these demands becoming the norm now?

[00:08:27] - [Speaker 1]
Yeah. Absolutely. I mean, with the automation, what it leads to is an operational efficiency need. An an operational efficiency need, the equipment to be working twenty four hours, seven days, three sixty five days a year. And that's driving a lot of the demand for, you know, much, much quicker charging than than what it is.

[00:08:54] - [Speaker 1]
Today, you know, a very efficient, power footprint, is absolutely critical because the cost of power then becomes a larger portion of your operating cost. So you want it you want that to be efficient as well. At the same time, you want to have you wanna display a sustainable, you know, concern towards how how, the behavior of the of the of the customers, right, of the of the clients. So all of this is we're certainly seeing a shift towards towards this type of, approach.

[00:09:30] - [Speaker 0]
Now something we also should mention here that many operators are being forced to oversize infrastructure to handle some of the peak lows that we're talking about here and which increases capital spending and energy waste, which is something that many don't wanna hear about. So looking at the industry's next challenge, we record this right at the beginning of 2026. What what is the secret to scaling without overbuilding? I'm sure that is almost a podcast episode on its own, but anything you can share around that?

[00:10:02] - [Speaker 1]
No. I mean, you know, in very simple terms, it's high uptime is the key to avoiding, the oversizing the infrastructure. But along with that, I mean, it's very simple to say that, but the in reality, the solutions are quite challenging. Right? But as I mentioned earlier in my example, where one of our customers has got a 50% uptime, and with our solution, you actually reach about 90% uptime.

[00:10:32] - [Speaker 1]
That's the kind of solution that we need to enable, you know, higher productivity, rightsizing the infrastructure, putting less pressure on the critical, needs in terms of power, etcetera, and the size of operation. I mean, in fact, because of the higher uptime in in one of our other customers, the footprint the the entire manufacturing footprint is half of what they they needed. So those are the type of responsible solutions that is required to drive, a reduction in terms of the, oversizing the infrastructure or drive drive towards an operating efficiency.

[00:11:22] - [Speaker 0]
And I think this is a great moment to introduce everyone listening to Niovolt. And for people hearing about you guys for the first time and and what you're doing, can you just tell me a little bit more about how you're helping businesses overcome many of the challenges that we're talking about here today? Tell me more about what you do and how you're doing that.

[00:11:41] - [Speaker 1]
Yeah. Sure. I mean, a very quick introduction about Nibold itself. We're we're six years old. We're headquartered in UK and Cambridge, but we've got operations, around the globe.

[00:11:52] - [Speaker 1]
And our main focus is on, you know, high power, fast charging battery, that is aimed very much at, increasing the uptime, in terms of where it's mission critical, operations. Right? And that is you know, whether it's been AIDC, robotics, industrial automation, all of those, areas, and and in automotive as well.

[00:12:24] - [Speaker 0]
One thing I'd love to ask you is you said that you you were formed in 2019. So 2019, you've created this fast charging energy solutions that deliver mission critical uptime for power demanding industries. And that was a time before all the AI went mainstream. I know AI has been around a long time, but that mass adoption and that mass energy consumption was still a few years away. Did you have any idea just just how quickly that demand would go up?

[00:12:53] - [Speaker 1]
Great question, in fact. I I wish I could say that, you know, we knew about AI, which I don't think anybody knew back in 2019. Think the the aim at that time was we certainly felt that there was a gap in the market in terms of the power at which the batteries were charging. And, you know, if you take automotive, for example and we actually showcase this through our concept vehicle that we built. Today, you know, customers are charging their vehicles at forty minutes thirty to forty minutes, you know, full charge.

[00:13:35] - [Speaker 1]
We proved that we we showcased that we could do that in under five minutes. We went from 10% to 80% in four point three seven minutes, in fact. And and the so we we believe that this is not restricted to AIDC or industrial automation. It can actually change the entire customer experience, at a product level, you know, for automotive and and beyond that. So that was the thought process at that time that we wanted to break this view that you needed to make the batteries bigger and bigger and bigger, but have a much larger downtime to something that is much more cost efficient because you can have a much smaller pack, quick to charge.

[00:14:28] - [Speaker 1]
Of course, you need charging infrastructure. I'm I'm not taking that away from the one of the fundamental requirement. But you can imagine if you could charge a vehicle in in five minutes, you don't need to do all of this planning that seems to be, you know, point of discontent today amongst drivers, you know, when you're driving for three hours and then you need to have a an hour break, it's like a seamless transition from, petrol diesel to to electrification. So that was really the thought process that drove, where you know, how Niobold started thinking about this, and moving away from batteries being commodity to batteries enabling a a delight in terms of the customer experience.

[00:15:23] - [Speaker 0]
And I'm curious if you look at the last eighteen months of your work at Niwa, have you have you noticed a huge upsurge in interest here?

[00:15:32] - [Speaker 1]
Yeah. I mean, certainly, there there's you know, we're having quite a bit of discussions amongst different industries, including mining, automotive, industrial automation, and the data center and and beyond as well. But some of the industries take a little bit longer in terms of speed of adoption as opposed to, like, the data centers. Right? They move much, much more rapidly because they have an immediate need for a solution, which does not exist today.

[00:16:09] - [Speaker 1]
So that's driving some of the the focus internally, from our perspective.

[00:16:18] - [Speaker 0]
And looking forward there, the the next wave of innovation across AI, robotics, and indeed industrial automation, that's also gonna rely on high energy systems that can keep pace with some of the real time demands that we're talking about today. For business leaders listening, I always try and give people listening a valuable takeaway. Any advice that you would offer around how and and why organizations can re or should be rethinking their approach to power? Anything you'd like people to walk away with and think about when thinking about power?

[00:16:51] - [Speaker 1]
Yeah. I I think a lot of the thought process when we talk to our customers, what we find is their solutions that they're thinking about is within the boundary of what they see in the market. And what Niabolt offers is something way beyond what most people have actually seen. So I think, the key is to understand what is the most efficient operation that you would like and look for that solution that you could actually enable and be in the forefront of innovation in terms of implementing those solutions. So I I think the the boundary is is actually, you know, set by ourselves, and we just need to break that boundary to go beyond that.

[00:17:43] - [Speaker 0]
And before you joined me on the podcast today, was doing a little research on you guys, and I quickly learned that Symbiotic last year, the Boston based leader in AI enabled robotics technology, they turned to Niabolt for a major product rollout. So huge success there, and I think towards the end of last year. But this year, anything that excites you this year? Anything you're working on or any big goals or anything you can share about the road ahead and what what 2026 means for you guys?

[00:18:12] - [Speaker 1]
Yeah. I mean, 2026 is gonna be a huge growth for us. Yes. Symbotic was a was a you know, was very, very it proved our technology. Symbotic was very cooperative with us as well in terms of being a a fantastic customer for our product and having that thought leadership, if you will, in in driving the change in their operation and enabling that technology our technology in their in their warehouse.

[00:18:51] - [Speaker 1]
But, yeah, this year is gonna be very, very exciting. We we're entering the the data center world, and we're gonna be entering couple of other new applications, which I cannot talk about too much yet. But I'll just say wait for the exciting announcements to come.

[00:19:09] - [Speaker 0]
Oh, you've left us on a teaser there. It sounds like we need to get you back on later in the year. But for people listening that already have their curiosity sparks somewhat and wanna keep up to speed with what some of those announcements might be and how you might be able to, work with them, for example. Where would you like to point everyone listening? Where's the best place to keep up to speed with everything?

[00:19:33] - [Speaker 1]
Nyball.com is a great place to to to learn a lot about what what we do and with the latest news on what we are what we are up to. But yeah. I mean, if you're interested in understanding what Nibel can do for your operations, please feel free to reach out to us, either through our, our website or LinkedIn. We would, would love to have conversations with you.

[00:20:01] - [Speaker 0]
I'll I'll add links to everything. So anybody listening should be able to just click on the link in the show notes. But such an important topic right now, unlocking next gen industrial performance across everything from AI data centers, warehouse robotic automation, and electrified heavy duty applications. These are things that many of ourselves and as individuals and indeed industries and workplaces take for granted. We just plug something in and miraculously that power appears.

[00:20:30] - [Speaker 0]
But it I think as that demand increases, solutions like Art and I are about are more and more important. So just thank you for shining a much needed light on this. Really appreciate your time today, and we will get you back on later in the year to find out more about some of those announcements that you teased us with. Thanks for joining me.

[00:20:47] - [Speaker 1]
Thank you, Neil, and look forward to the next chat.

[00:20:50] - [Speaker 0]
So if this conversation shifted how you think about AI infrastructure, that's because power is no longer a side issue. The next phase of AI robotics and automation, they're gonna be shaped as much by energy systems as by software Because high uptime, fast charging, and right size infrastructure, these are few things are becoming strategic decisions, not just engineering footnotes. And you heard why oversizing to handle peaks actually just creates waste and how very often higher uptime can shrink physical footprints and why rethinking battery performance can unlock productivity across data centers, warehouses, and industrial environments. And there was also a little glimpse into where all this is heading next, especially as demand accelerates and new applications emerge faster than many of you listening will expect. So if you're building or scaling systems that need to stay online, need to receive power around the clock, this is your moment to reassess how power fits into your broader strategy.

[00:21:56] - [Speaker 0]
And excuse the pun, but if this episode sparked any ideas or raised questions about your own operations, please let me know your thoughts and share it with someone that you think may find this conversation valuable. So here's the question I'm gonna leave you with today. As AI pushes performance expectations higher and higher every single year, is your power strategy ready to keep up? Food for thought indeed. But that's it for today.

[00:22:24] - [Speaker 0]
So thank you for listening as always. Remember, you can contact me over at techtalksnetwork.com, but it's time for me to prepare for tomorrow's guest. Speak with you all again very soon. Bye for now.