Decode: What does it mean to scale AI/GenAI and become 'AI-native'?

The AI promise is massive and being widely embraced: According to the latest McKinsey Global Survey on AI 65% of respondents report their organizations are regularly using generative AI (GenAI), nearly double the percentage from just 10 months ago.

The report states: “If 2023 was the year the world discovered generative AI (GenAI), 2024 is the year organizations began truly harnessing and deriving business value from this transformative technology.” This rapid increase highlights a significant shift from experimentation to integration, indicating that GenAI is no longer a novelty but a critical component of business strategy.

But, despite this enthusiasm, to date, few organisations have successfully adopted AI/generative AI at scale, because knowing where and how to apply AI, and critically, what ROI it will deliver, remains uncertain. Thus, many CxOs are entering a critical phase in their relationships with GenAI. As most organizations are learning, it is relatively easy to build GenAI pilots but turning them into at-scale capabilities is more complex, and no more so than for communications service providers (CSPs), that have to address rapidly-scaling data traffic and an increasingly diverse range of services, use cases, users, and devices on their networks, all whilst under pressure to deliver improved profitability. So, can AI/GenAI enable CSPs to create scalable solutions and tackle the increasing complexity of their business needs?

This was the proposition of a double workshop ‘decode’ session presented by McKinsey & Company at the recent Innovate Americas event in Dallas, hosted telecoms industry association TM Forum.

The first decode session, led by McKinsey & Company’s Jorge Amar (Senior Partner), Guilherme Cruz (Partner), and Borja Belda (Associate Partner) focused on what CSPs need to do to rewire how they work and explore the seven truths about scaling GenAI. It began by considering two primary concerns for executives faced with building an AI strategy: how to deploy and scale AI to enable better decision-making and improve productivity, and how to position and capture value​ the ever-expanding AI value chain?​

For CSPs, AI is typically applied across three vectors: However, despite initial investment, many telcos struggle to move from PoCs to value at scale​, due to a range of challenges:

Workshop leaders Amar and Cruz outlined how overcoming these challenges required domain-focused action across areas including technology, data, partnerships and ecosystem, operating model, skills and change management as pillars of scale, each of which requires multiple solution building blocks to achieve scale. For instance, to drive scale, the data ‘pillar’ requires a unified data lake, data governance and data products solutions to be in place, whilst the change management pillar requires workflow change, incentivization and impact tracking and progress benchmarking.

Fellow leader, Belda, went on to describe how a GenAI platform of reusable services, reusing code and APIs, for example, could streamline and standardize use case delivery, minimizing rework​ for CSPs, and operationalizes internal AI, maintaining and ensuring performance of microservices​ and service components, so the latest technology is readily and securely available for consumption​. The assertion was that CSPs can then convert the GenAI platform into a GenAI factory​ which could enable the continuous delivery and automatic deployment of gen AI apps, bolstered by ongoing monitoring & feedback​. This could include features such as pre-provisioned services, pre-loaded components (and recipes, wizards, docs), and dynamic resource containers, and support hosted open-source/custom LLMs or external LLM services​ as well as the full GenAI lifecycle.

This session was followed by an illuminating fireside chat with special guest Kalyani Sekar, SVP & Chief Data Officer, Verizon, who was in conversation with Joshan Abraham, Associate Partner, McKinsey & Company. In their discussion, Ms. Sekar outlined Verizon’s experiences to date with AI and genAI, and positive response to the GenAI factory concept. In particular, she described some of the CSP’s successes so far in utilising AI, to improve fault management in the network, and in leveraging genAI to enhance customer experience, but also noted some of the challenges, such as putting in place an AI-centric data strategy, to maintain the accurate and consistent flow of data through the business, which is critical for building LLMs and supporting other AI-driven use cases.

This discussion was followed by another decode session, 'How Telcos Can Capture Other Opportunities Across the (Gen)AI Value Chain', led by Sebastian Cubela, Partner, and Miguel Frade, Associate Partner, McKinsey & Company. They began by outlining how the AI opportunity is expected to create >$100B of value to be unlocked and how CSPs have been experimenting with multiple strategic moves, such as building their own LLMs, and making data centre infrastructure investments. But despite this, many questions remain. For instance, how much value is realistically available for CSPs and how should they position themselves to win in the AI ecosystem?

This decode session considered what roles CSPs could play in the value chain; and how they should win, particularly given the dominance of hyperscalers, actively building out capacity in new and atypical locations, outside of core datacenter markets. ​But, according to Cubela and Frade, there are still opportunities for CSPs to explore across the (Gen)AI tech stack and ecosystem​, in:

Each of these were explored in the workshop, generating much discussion and interactivity with participating CSPs, rounding out the deep-dive into scaling AI/GenAI, and AI positioning for CSPs, as part of the journey to becoming an AI-centric telco.

The event also featured keynote speaker Alex Singla, Senior Partner and Global Leader of QuantumBlack, AI by McKinsey, who had this to say about his participation:

“Partnering with TM Forum for Innovate Americas was an incredibly enriching experience. The collaboration was seamless and highlighted the depth of expertise and innovation that both of our organizations bring to the table. The first TM Forum event of its kind this decade for North America, showed a commitment to fostering a dynamic and engaging environment, making it a truly memorable and impactful experience across multiple sectors and functions. As a senior partner at McKinsey and leader of QuantumBlack, AI by McKinsey, I have seen firsthand the transformative power of AI and GenAI across various industries. During my session, I emphasized the reality of AI and GenAI, the critical role of speed as a strategy in adopting new technologies, and the challenges of adoption and scaling in organizations. These points underscore the importance of proactive governance and effective integration of AI tools into business processes to stay ahead of the competition. I am grateful for the opportunity to share these insights and engage with industry leaders in such a robust setting.”