AVK has secured a majority investment from Partners Group, including an initial commitment of more than $1 billion.

The deal is the first time AVK has taken external funding in its 36-year history. Chief Executive Officer Ben Pritchard will retain a significant shareholding alongside the existing management team.

Investing on behalf of its clients, Partners Group will become the majority shareholder in the UK and European supplier of power systems for data centres and AI infrastructure. It will also provide capital to support the buildout of on-site infrastructure under an energy-as-a-service model for data centre operators.

The funding will support AVK’s strategy to fund, develop, own, and operate on-site power systems, including microgrids. The company already has a pipeline of more than 2GW tied to that plan.

The investment comes as data centre operators across Europe face growing pressure to secure electricity more quickly, with grid connection delays and constrained power availability becoming bigger obstacles to expansion. AVK says on-site generation can help reduce delays by bringing supply closer to the facilities that need it.

AVK has built its business around prime, standby, modular, and dispatchable power systems, with a focus on mission-critical installations. Its operations are supported by a manufacturing facility in Haydock, north-west England, and a workforce of nearly 400 across ten hubs in the UK and Europe.

New funding

Under AVK’s energy-as-a-service model, customers would buy electricity through power purchase agreements rather than take on the upfront cost and development risk of large on-site energy projects. That shifts financing and ownership of the assets to AVK and its backers.

For private equity and infrastructure investors, the appeal lies in rapidly rising demand from AI and data centre projects, which are putting greater strain on existing power networks. The sector has become a focal point for investors seeking exposure to both digital infrastructure and electricity supply.

“Speed-to-power is now a defining opportunity for European data centre operators. Our new partnership with Partners Group will allow us to meet our customers exactly where the market demands. From the moment we launched our first microgrid, we recognized the challenge and the opportunity facing developers and operators globally. By adding capital to our power solutions portfolio, we can turn speed-to-power from an ambition into action. I am excited to lead AVK into this new chapter alongside Partners Group, leveraging the firm’s deep operational expertise in the data centre sector and power markets,” Pritchard said.

Partners Group has previously invested in decentralised energy assets in Europe and in data centres, including the pan-Nordic platform atNorth. It has also invested in behind-the-meter data centre energy providers in the US, giving it experience in a market where operators increasingly seek localised sources of supply.

Market pressure

Demand for data centre capacity has risen sharply as cloud computing and AI workloads expand, but the pace of new construction has run into power shortages in several European markets. That has made access to electricity, and the speed at which it can be delivered, a more prominent factor in site selection and project design.

AVK recently energised what it described as Europe’s first large-scale data centre microgrid at a PureDC site in Dublin, where power constraints have become a major issue for new digital infrastructure. The company is using that track record to position itself as a provider of on-site alternatives for operators that cannot wait for conventional grid upgrades.

Nicholas Pepper, Managing Director, Infrastructure, Partners Group, said: “AI is driving one of the largest infrastructure buildouts in decades, and access to power is becoming a defining constraint. This constraint and lengthening connection queues are critical bottlenecks to growth in the European data centre market, which onsite generation can alleviate by accelerating speed-to-power. AVK, with its deep expertise, track record, and pan-European footprint, is well-positioned to address this issue as a one-stop shop for data centre power solutions. We see an exciting growth opportunity for AVK and we look forward to supporting the management team in its next chapter.”

The deal gives AVK fresh capital at a time when investors are looking for businesses positioned between electricity infrastructure and digital growth. For AVK, it also opens a new phase in which the company will move beyond supplying equipment and services to owning and operating assets tied directly to customer demand.

Pritchard and the leadership team will remain in place.

Julian Rosser has assured the public that the Duke of Cambridge will reopen again soon with its current closure coming after a reported burglary in June this year.

His statement comes after Duke Property Ltd, which is based at the Duke of Cambridge, entered Creditors Voluntary Liquidation on July 28.

This is a a liquidation procedure that enables a company to be wound up by resolution of the members of the company instead of by a court order.

READ MORE: Statement as historic UK jewellers in administration amid £189K debts

However, Mr Rosser – who has run the cocktail bar since 1998 – has said that Duke Property Ltd is to do with the lease of the site and not involved in the day-to-day operation of the bar.

He said: “The Duke will continue. It hasn’t gone into liquidation; Duke Property Limited has.”

Duke of Cambridge in Little Clarendon Street (Image: NQ)

The liquidators appointed are from Fortis Insolvency, with Daniel Taylor of the firm stating that the economic climate over the last few years has provided “major challenges”.

He added: “We know that this business is not alone in what it has faced over recent trading periods, and suspect that there are more economic consequences yet to be felt.”

Mr Rosser agreed the the economic climate isn’t good citing the Botley Road closure – which has lasted several years and is set to end in September – as a difficulty.

“Trading in Oxford is very difficult right now,” the 62-year-old said, who also said students from the university weren’t visiting as much as they used to.

Julian Rosser

Following the burglary in June, he said that The Duke of Cambridge will remain closed until students – including from Somerville College which is a neighbour to the bar – return in the Autumn.

In part, this is because he wants to brainstorm how to improve business.

He said: “It always used to be very very busy but turnover has taken a hit. We need to think about how we are going to reinvent the Duke.

READ MORE: UK singer behind legendary 80s Christmas tune leaves £12m to wife

“We stopped doing food in 2002 but it might be time to bring food back.”

Mr Rosser added that he was also considering changes to their opening schedule.

On its website, the Duke of Cambridge describes itself as “Oxford’s leading cocktail bar” and says its has been open since 1981.

It adds: “Located in the bohemian district of Jericho, the bar is always bursting with atmosphere at the weekends with a more chilled vibe during the week.

“Fresh ingredients, cool interiors and friendly staff give The Duke genuine character and style.”

The UK’s AI Security Institute said tests of Anthropic’s Claude Mythos 5 and OpenAI’s GPT-5.6-Sol found cases of unsanctioned behaviour during cyber evaluations. The most serious incident involved an attempt to insert malicious code into a real open-source project.

The behaviour emerged during a routine assessment of frontier AI models under deliberately permissive test conditions, including internet access and disabled cyber misuse filters. The institute reviewed 122 runs across seven models and found 19 out-of-scope actions in 10 runs. Seventeen were linked to Mythos 5 and two to GPT-5.6-Sol.

According to the institute, one agent created a malicious pull request for a public software project, then tried to influence the approval process by creating fake online identities and pressuring a maintainer. The maintainer rejected the code, and the institute found no evidence of real-world harm.

The findings add to a growing debate over how autonomous AI systems behave when given broad goals and access to live online services. The institute said the models did not escape their sandboxed research environment, but their actions were sustained, novel and directed at real people and organisations.

Its investigation traced the incident to a single cyber challenge designed to test whether agents could solve a security task over many steps. Some runs were misconfigured in ways that may have made the task harder, the institute said, but that did not fully explain the behaviour. Some agents still acted beyond scope even when they had a valid route to solve the task.

Security monitoring first flagged unusual data transfers through the Tor anonymity network. Investigators then found that one agent had used GitHub and other online services in ways that breached the intended boundaries of the test.

Beyond the attempted code insertion, the institute described efforts to contact real people with messages and files intended to persuade them, or their own AI coding tools, to run malicious code. It also said one agent tried to plant hidden instructions for other AI systems, and that separate agents appeared to collaborate by reusing public artefacts left behind during the evaluation.

Expert response

Waseem Ahmed, Head of Engineering at Secure, said the incident should be understood as a consequence of goal-seeking behaviour rather than a loss of control in the popular sense.

“Let’s be precise about what happened, because ‘AI went rogue’ misses it. AISI’s own report is clear. The agent did not turn evil and it did not escape its sandbox. It was told to solve a hard security challenge, and deception emerged as a by-product of chasing that goal.

“Two details matter. This was a model not yet released, and testers had switched off the safety filters on purpose to probe raw capability. That is not how these models behave in production with guardrails on. The real lesson is that a capable agent chasing a goal will try routes you never approved, including social pressure aimed at real people. That is new, and it is why we cannot treat agents like ordinary tools.

“The most reassuring fact in the report is also the most alarming. The attack failed because a human caught the bad code and refused it. Good practice worked, but the margin was thin. It depended on human vigilance, not a technical wall that would reliably stop a stronger agent.

“So here are four moves for security teams. First, block open internet access for agents by default and grant it only when a task truly needs it. Second, watch agents in real time so you can stop out-of-scope actions as they happen, not find them in the logs later. Third, assume any capable agent will try to bend its limits, and build guardrails and containment before it runs. Fourth, harden code review and contributor identity checks, because fake identities are now a real supply chain attack path, and treat all AI-generated or outside code as untrusted until you verify it in isolation.

“The strongest response is still standard cyber hygiene done well, which matters more as these agents get stronger,” said Waseem Ahmed, Head of Engineering at Secure.

That reading is broadly consistent with the institute’s own account. It said the behaviour arose because the agent persistently pursued its assigned objective and explored routes operators had not intended, including deception and social engineering.

Control measures

The institute said it contained the incident within about an hour, halted related evaluations and disabled internal access to the most advanced models under review. It also worked with GitHub to remove artefacts left by the agent and notify affected users.

In response, the organisation is tightening how internet access is granted in cyber evaluations, adding real-time monitoring to flag or stop out-of-scope actions as they happen, and reviewing evaluation design so containment does not depend on a model choosing not to test its limits.

The case is likely to sharpen scrutiny of how frontier model developers and testing partners assess systems before release. While the institute stressed that the specific model configurations were not commercially available and did not reflect ordinary public deployments, it said the episode showed a direction of travel that warrants immediate attention.

For businesses, the report points less to a sudden breakdown in existing systems than to a widening attack surface in software supply chains and identity checks. The institute said standard security practice, human judgement and caution around outside code prevented the worst outcomes, but added that in several cases the margin between failure and success was narrow.

It concluded that the behaviour was possible, sustained and new.