IBM Think rocked up to London last week, giving SDxCentral a chance to explore where IBM currently stands in the AI infrastructure boom, as well as hear from British clients on how they are keeping pace with said buildout.

In a survey released to coincide with the event, IBM claimed 80% of surveyed European executives experienced an AI-related incident that reached their board or C-suite in the past year, with 40% changing their AI governance policies or frameworks as a result. Meanwhile, 56% of execs continue within agreed risk limits while governance adapts, as 33% delay deployment until AI governance is mature.

Without further ado, here are five thoughts from IBM Think on AI governance and deployment, open source, and even post-quantum security.

All businesses will be neoclouds by 2030

“We believe within the next five years, every company in the world is going to be a neocloud.” So claimed Rob Thomas, SVP of software and chief commercial officer for IBM, who told the Think audience that businesses will increasingly need to design their infrastructure around AI, rather than rely entirely on traditional in-house data centers or on-demand public cloud. Instead of providing AI services through GPU rental, IBM’s version of a neocloud is a hybrid setup comprising company-owned GPUs and CPUs, public cloud, sovereign deployments, and edge-based AI running when needed.

“It will be that every company, no matter your size, is going to look like a neocloud in the next few years. AI demands that,” Thomas stressed.

Britain is boldly agentic

After the keynote, Leon Butler, chief executive, IBM UK & Ireland, reminded a media roundtable that the U.K. is the world’s third-largest tech market.

“There is a huge amount of innovation going on. (In our global survey) from an agent perspective, 35% of people are adopting and using agents. Actually, in the U.K. it is 57%, so we are certainly well ahead, which is fantastic. But we're still seeing, and certainly myself and my team are struggling to see people move out of the pilot phase,” Butler said.

Companies that have moved out of this phase include British insurance provider Markerstudy, which coincidentally was recently presented by Nutanix at another roundtable attended by this title.

While Markerstudy was not in attendance at the IBM panel, Butler said the company has gone all in on vibe coding by using multiple development agents, including IBM’s Bob, to validate and rework 6,000 lines of code for a core application in four weeks.

Craig Aston, CEO at Celerity, said at the panel that his IT consulting firm was “absolutely completely bought” into Bob.

“It widens out the whole adoption. It's not just about applying a model. It's not just about how you use a model to make something better. It takes you through the whole process lifecycle … and drives real outcomes, and it actually gives you measurability,” Aston said.

Governance is king

In a tip of the hat to IBM family member Red Hat, Butler would go on to say open source was fundamental to British technology growth.

There were no mentions in the Think keynote of Project Lightwell, IBM and Red Hat’s $5 billion joint commitment to open source security. But governance defined Think 2026, with IBM’s survey results finding just 37% think their governance is designed to scale with agentic AI deployment.

As such, IBM is presenting itself as the governance partner to beat in the infra buildout. Ana Paula Assis, SVP and chair for IBM Europe, Middle East and Africa and Asia-Pacific, gave a nod to Lightwell when she stressed that open source is “an area that requires a lot of focus and attention.”

“With Red Hat, we are investing in 22,000 engineers to build a clearinghouse where we're going to identify vulnerabilities in the open source space and create the corrections and the patches for those vulnerabilities, regardless of the version of the product that you are using. So you don't need to upgrade your product to have access to those patches. Those patches are going to be certified by Red Hat, so that is going to be another key component because we believe that that's an area that is going to be exploited by agents,” Assis said.

Celerity also uses Bob for governance, with Aston revealing the tool helped it find an example of problematic open source software getting in the way of its ISO/IEC 42001 accreditation, the international standard specifying requirements for AI management systems.

Watson can track tokenomics

On the governance topic, Celerity believes AI needs governance around cost, as well as clear guardrails because expenses can rise quickly. According to Aston, the firm uses the Watson suite to track the use of AI tokens.

"We use WatsonGov (Watsonx Governance) to actually look at what tokens are being used, what they're being used for, how many are being used. So it's built into a system that's linked to other open source systems to be able to measure that," Aston said.

For the CEO, having such guardrails doesn't hinder companies from becoming IBM-envisioned neoclouds, instead allowing for innovation to be "done in a controlled fashion."

You’ve probably already been harvested

Quantum security came up in the Think panel, with David Debnam, operational risk and innovation director at insurers Aviva revealing its biggest problem is its data has “probably already been harvested.”

This was in reference to harvest-now, decrypt-later attacks, in which threat actors steal encrypted data with the intention of waiting for post-quantum decryption technologies capable of breaking current cryptography methods.

Celerity is currently defending itself against such threats by working with IBM on quantum-readiness assessments using IBM’s Guardiam, a data protection suite designed for multiple environments.

“It's going to become a really big subject in the next 12, 18 months when you look particularly in the public sector in the U.K., where there's a requirement that organizations will have to report on their quantum readiness in 2028, and that is not very far away. So yeah, we're absolutely looking at it,” Aston said.