Scotty Petty, CTO of Vodafone Group, explains the far reaching implications of efficiency and how he and his team are contributing to that goal, including the pivotal impact of AI. By Contributing Editor Annie Turner.

Scott Petty, CTO of Vodafone Group, has been with the operator for 18 years, in several senior positions. He has been in his current role for three years and has dual responsibilities: first he leads Vodafone Technology which has more than 30,000 employees spanning Europe, Africa and Asia. Secondly, he is a member of the Group Executive Committee, accountable for Vodafone Digital and IT, with worldwide oversight of Cybersecurity, Vodafone Business Platforms & Solutions, and Technology Assurance & Strategy.

A cornerstone of Petty’s approach could be summed up as it’s better to build exactly what the business needs in-house “where it really makes a difference or adds value or differentiates,” he says. “I’m never going to write my own ERP [enterprise resource planning] software. That doesn’t make any sense – I can buy SAP and implement that – but in Generative AI [GenAI], in autonomous networks, things that really make a difference to our business, that’s where we want to drive differentiation.”

In these critical areas, he explains that if what is on offer from vendors isn’t a proper fit it results in “compromise and complexity”. Note that under the leadership of Margherita Della Valle, who became Group CEO in January 2023, the company’s watchwords are Customers, Simplicity, Growth.

Powered by shared ops

How much has the DIY approach been adopted for newer kinds of AI? Petty says one of the toughest things about adopting GenAI then agentic AI has been sticking to the common IT and data architectures that are used across the group. He acknowledges it was hard to resist hype-driven pressure to incorporate various vendors’ AI solutions in the interests of speed and short-term expediency.

Maintaining the integrity of that common, shared architecture is foundational to the entire group’s strategy. This was illustrated on another of Della Valle’s slides in the earnings report on H1 of the current financial year in November 2025, ‘Powered by shared ops’ to become the ‘Most efficient telco, leveraging local & cross-border scale’ (see slide below).

Source: https://reports.investors.vodafone.com/H1FY26presentation/15/

 

‘Shared ops’ is a vision long held at Vodafone, so when GenAI emerged in late 2022, the operator set up a cross-company AI taskforce to decide where the most benefit could be derived. An initial 10 use cases fell within Petty’s areas (outlined in the slide below). He notes that November 2025 was a watershed: for the first time during an earnings call, Della Valle attached numbers to the benefits from those use cases, which form the AI@vodafone initiative. Its goal is to embed Generative AI in core operations, at scale, and is one of the group’s five strategic priorities for growth.

Source: https://reports.investors.vodafone.com/H1FY26presentation/22/

Petty notes that this slide is the most viewed in the presentation and the topic that attracted the most questions on the earnings call. He elaborates, “Particularly in 2023, people were saying, ‘Just do millions of PoCs, throw GenAI at everything, test it out everywhere and see how it works out’. We felt that that wasn’t going to go anywhere. You’d test lots of technology, but nothing would scale. We forced the business to focus on our 10 scale use cases and then learned how to apply GenAI to them. That’s helped us scale much faster than others.”

Cloud and data strategy

The ‘powered by shared ops’ also sheds much light on the centralised but company-wide, standards-based approach to cloud. Petty says, “It is a really important part of the strategy. First, it enables you to modernise infrastructure so you can be more innovative and drive new capabilities, but it also enabled our data infrastructure which is critical to AI.”

The foundation stone was building a data ocean (as it was described then although platform is the preferred term now) with Google Cloud and Microsoft which began in November 2019. Vodafone chose Google Cloud to host its strategic cloud platform, Neuron, for data analytics, business intelligence and machine learning.

According to Petty, Neuron brought multiple data sources together in a standardised format. It also improved operations by making Vodafone’s extant software cloud-compatible, so that local markets could use the new platform’s capabilities without major disruption.

In May 2021, Vodafone and Google Cloud announced a new, six-year strategic partnership to leverage reliable and secure data analytics, insights and learnings to support the introduction of new digital products and services for Vodafone’s customers simultaneously worldwide.

They announced they would build a new, integrated data platform that can process and move huge volumes of data globally, from multiple systems, into the cloud. The platform, Nucleus, houses a new system called Dynamo which is designed to drive data throughout the operator group, enabling it to offer customers new, personalised products and services across multiple markets more quickly.

Co-innovation to benefit customers

In January 2024, Vodafone and Microsoft agreed a 10-year strategic partnership: Vodafone is to invest $1.5 billion over the next decade in cloud and customer-focused AI services developed with Microsoft, which in turn will use Vodafone’s fixed and mobile connectivity services.

Then in October 2024, Vodafone and Google deepened their ties, announcing a new, 10-year strategic expansion of their partnership “to bring new services, devices and TV experiences to millions of Vodafone’s customers across Europe and Africa, supported by Google Cloud and Google’s Gemini [large language] model”. The agreement is to bring storage, security and ‘AI assistance’ to Vodafone’s customers in 15 countries, as well as its partners in another 45 markets around the world.

In Petty’s view, there is little point in telcos building their own large language models (LLMs). The trick is to pick the most suitable one for any particular application because they all have different characteristics including the size of the model, its performance and licensing terms. Also, things change too fast to make home-built models viable. He stresses, “Differentiation comes from applying your own data for inferencing, which you and only you control, then building applications on top.”

“We’re really impressed with Gemini, it’s capabilities continue to grow and we run a whole set of capabilities across Gemini, OpenAI [creator of ChatGPT], increasingly Claude [Anthropic] a little bit on Bedrock [AWS]. So we’ve got good experience across all of them,” he adds.

Right AI in the right place

Vodafone is a big supporter of TM Forum’s autonomous network programme, but says it is wrong to think about it as the journey from automation to autonomy: the point is to deploy the “right” kind of AI in the “right” place. For example, machine learning is largely what will be deployed within the network itself, with closed loop mechanisms used to identify or predict and fix all kinds of faults to minimise or avoid their impact. Petty comments, “It might be possible to deploy agentic AI to extract information in this environment, but it would be an unnecessarily expensive way of achieving the same goals.”

In short, the higher up the stack you go and the closer you get to customers, via customer-facing processes and functions, the more likely it is that GenAI and then agentic AI are better options.

“The real power of generative AI is it enables you to have systems that service customers in a much more authentic, interactive way than the how you build traditional systems where everything has to be built into a flow and therefore feels robotic when you’re interacting with it,” he adds. “The more natural the interaction to solve customers’ problems, the closer it brings you to the customer and that depends on how well you apply GenAI.”

How will AI change networks?

Has AI in all its guises altered how Petty sees Vodafone’s future networks? For example, some claim that AI will create a big rise in uplink traffic. Another argument is that data flows are more distributed and that expectations around latency, throughput, resilience, security and energy efficiency are changing.

Petty states, “Today, typically 90% of traffic is downlink, which has been the case for many years. In future, we expect uplink traffic will increase by a few percentage points, driven by new AI driven use cases such as associated with smart glasses or connected cars, but it will remain predominately downlink.

“The amount of new, specific AI traffic we see on our networks currently is small, but AI is a key factor in driving existing sources of traffic: for example, AI-generated videos and recommendations which are expected to increase levels of video streaming.”

On the other hand, Petty concludes, “AI will massively improve our customer experience, particularly to support sales and journeys etc. That will improve our NPS [Net Promoter Score] and therefore customers’ willingness to attribute a premium to us. More importantly, it can help us dramatically reduce costs and put those resources into other things.”