AI Infrastructure Is Entering a New Era 

officially moved past the AI hype phase and now it's all about massive full-scale enterprise rollouts.

Now hyperscalers are on a mind-boggling $850 billion cap expense to build up the tech infrastructure, but data centers are also rapidly running out of power and physical space.

Meanwhile, tech giant IBM, which rang the opening bell this morning here at the New York Stock Exchange, is expanding its C17 platform and Linux 15 family with.

The new Rockhopper 5 Analytics 15 express.

Now these architectures are designed to drop straight into standard 19 inch data center racks eliminating the need for expensive new infra tears or separate GPU clusters.

Well joining us live here at the New York Stock Exchange to break down this infrastructure shift is Stephen Dickens, CEO of Hyper Frame Research.

Steven, good morning.

Great to have you here.

Hey, thanks for having me on the show.

Well first and foremost, we have.

Blue here that rang the opening bell.

So tell us about the announcements and the implications for enterprise.

So IBM, I looked it up this morning, launched on the New York Stock Exchange November 11th, 1915.

So it's quite apt for me to come onto the FinTech show and talk about kind of the OG of FinTech.

IBM's core franchise is its mainframe franchise.

90% of the Fortune 50 still run on that platform, still processing 80% of the financial transactions, so very apt that they're here today, underpinning what we talk about on Wall Street and so many of those financial institutions.

They're bringing to market new infrastructure solutions today as part of the launch, that's why I'm here with that party today.

And I think the way to think about this is new form factors, new sort of capabilities really to underpin those transactional workloads.

Whether that's Visa, Mastercard, JPMC, Wells Fargo, all of those big banks, telcos and retailers and governments as well.

Particularly Wall Street runs on that platform.

And I do want to expand on this from the technology side because when we're talking about the AI trade, there are a lot of players in the ecosystem here.

So when we're talking about the Z17 as well as the rockhopper, what are the implications for highly regulated industries.

So many of those organizations are trying to bring AI workloads closer to the transaction, so this isn't creating videos or doing sort of chatbot type work or even coding.

This is really how do I do a fraud scoring in real time in the.

Transaction flow and IBM has got capabilities with the Spire accelerator where they're bringing that XPU type technology directly into the transactional flow.

So that really enables some of those financial institutions that I mentioned to make better, smarter decisions closer to the transaction.

Let's talk about the Costs here because we know that the expense for this type of technology is not cheap.

So when it comes to enterprises, what do they need to keep in mind and what's the reality, especially at a time when we're talking about raising capital for some of the hyperscalers as well as the big AI names?

So huge build out the hyperscales that you mentioned, what is it?

700 to $800 billion that they've raised.

Big names like Google and Oracle making debt issuances to invest in Capex, so we're at the sort of inflection point.

We're going to see a lot of that moved to inference.

And as that build up, so we've seen the Frontier Labs and those hyperscalers invest.

I think where IBM is positioned is to catch that inference wave, as I mentioned, transactions and being able to do that AI scoring closer to the transaction.

That's an inference workload.

That's not a training workload.

So I think the viewers here have got to look at it in a couple of contexts.

You're going to see a lot of that infrastructure build out for the frontier labs and those hyper scanners, but we've also got to think about inference.

That's happening in the enterprise, that's happening with sovereignty, that's happening with these organizations wanting to take control of their data.

Yeah, and we're talking about inference here, so for the layperson who's trying to understand where we are in the time horizon, where would you say we are right now and how much longer do we have to go?

So that's the that's the multi-trillion dollar question there Remy, but no I think we're still we've seen that obviously early build out for um over the last sort of three years for training.

I think we're still first game of a 7 series World Series, but I'd almost say we're almost in the 2nd or 3rd innings of that first game.

We're still so early.

This is a 20 year mega trend for me in the same way that the cloud was and it still is for those hyper scales.

Those trends build on top of each other.

I think we're 3 years into what's going to be a 20 year trend.

Trade.

So I still think there's a lot of infrastructure space to run.

Inference is going to be that next wave.

Finally, before I let you go, for investors out there who are looking at this space, where would you say the opportunities are right now?

I think IBM is a classic example.

I'm looking here.

The stock is up 1 or 2% in early morning trading.

I think obviously those big mag 78 names.

Get all of the hype.

I think it's looking at some of those infrastructure names the Ciscos, the IBMs, the HP, Dell's caught a lot of hype, Lenovo, those types of names that are maybe not those biggest hype names, but those names that are really providing that core foundational infrastructure that's going to underpin that inference.

And finally, before I let you go, less than 60 seconds here, but We're about to head into earnings season, so when it comes to some of the names in this space, what are you paying attention to?

So I'm going to be looking for the picks and shovels of AI.

So it's looking beyond some of those frothy metrics and really looking at where core companies that I mentioned, some of those names are actually seeing deployment, so looking for some of the questions that get answered on those.

On those earnings calls rather than the press releases.

Well, Stephen, always great having you on the show.

Thank you so much for joining us today and thank you so much for sharing your perspective and your insights.

Fantastic.

Thank you very much.