From technical debt to capability debt: CTI Digital CEO on the evolution of enterprise AI

Speaking at MD Future 26 last week (17 September), Chris Burgess discussed the need for a considered approach to AI solutions, rooted in business process.

“As a CTO or a developer, you're often challenged with technical debt. It's very annoying. It can hold you back, and you have to know how to challenge it. But businesses are now dealing with capability debt,” says CTI Digital CEO Chris Burgess.

“I don't blame boards and businesses struggling to understand where to go because of how quickly it's changing,” he added.

Burgess was discussing the rapid evolution of AI infrastructure and large language models, and the inertia that many businesses are feeling when trying to rollout new solutions.

Tech is “evolving quicker than businesses can adapt”, added Burgess. “Yes, we need to host the data and manage the information flow… but there is also training. We're talking about training new models and training new intelligence infrastructure. Business needs to think about where their data is going to be stored, how it gets to that location and how they evolve that data in that location.”

“The speed, agility and optionality that you need with infrastructure needs to be built in, and that's something we need to figure out,” Burgess said, speaking at Manchester Digital’s MD Future 26 conference last week (17 September) at Campfield in Manchester.

“The challenge you're trying to solve now dictates the infrastructure you need.”

Burgess argued for a considered approach to AI within the enterprise, and against a ‘one size fits all’ solution.

“It isn’t just like ordering a faster broadband connection,” he said.

“What is the right thing for your business process, and what you're trying to achieve? It could be efficiency, growth, or a niche area. The challenge you're trying to solve now dictates the infrastructure you need.”

Burgess was contributing to a panel discussion titled, ‘Who will build the infrastructure of the intelligence economy?’, alongside speakers from Hewlett Packard, Kyndryl and Pulsant. Earlier this year, CTI Digital launched its own sovereign AI platform, Croft, and offers clients their own private LLM platforms (including managed hosting and on-prem infrastructure) and AI training.

“Never before have we been so gaga about a technology,” cautioned Burgess. “‘Let’s get AI, let’s do something with AI.’ But why? What are you actually trying to achieve? Okay, you've made a case for it. What's the right infrastructure for that? It might be a data center or it might be a box on your desk…”

“We need optionality… we need to be able to experiment”

On AI infrastructure, Burgess’ keyword was ‘optionality’.

“I think we need to be able to experiment and build in experimentation, because AI is a great leveler – for a one-mad band, all the way up to an enterprise-level business. So when you've leveled those odds, you need to be able to have an infrastructure that adapts to that agility as well.”

Here, the CTI CEO pointed to new innovations in AI that could potentially disrupt the market:

“We're talking about large language models, which make billions of calculations across billions of parameters really quickly, and it's compute intensive, and it uses GPUs… It's great, but actually, [researchers are] already developing spike neural networks, and neuromorphic chipsets, which can calculate and work like the human brain. So it doesn't activate 200 billion parameters, it might activate a couple of thousand and give you the same result. That's not 10 years into the future.

“We're not in an age now, where things take five to 10 years, we’re in an age where it takes six to 12 months, so we have to think about that.”

CTI Digital CEO Chris Burgess was speaking at ​MD Future 26, which brings together the leaders shaping the future of AI, compute and frontier technology in the North.