An article by Rayan Salha, Director Product Marketing at Infovista.

Artificial Intelligence is driving a renewed wave of transformation in telecom, but unlike previous tech shifts, no single vendor can deliver everything a CSP needs to become AI-native. At FutureNet World 2025, this reality was made clear through candid insights from leaders including Infovista CEO Rick Hamilton and VMO2’s Director of Planning, Transformation and Performance, Anita Tadayon. The message: AI success depends not on finding a vendor, but on building an ecosystem.

This  blog explores why the “full stack” promise of AI is often misleading, what CSPs really need from their partners, and how to approach transformation in a way that balances vision with real-world complexity.

1. The myth of the full-stack AI vendor

In a market racing to capitalize on AI hype, vendors often position themselves as “end-to-end” providers. They offer platforms that claim to manage data ingestion, model development, orchestration, deployment and insight delivery across the entire value chain.

But according to Hamilton, this kind of promise is not just unrealistic, it’s counterproductive. “Any vendor out there telling you they have the magic bullet or the full stack, disregard them,” he said during the panel. “It’s simply not true.”

In reality, telco environments are too fragmented and context-specific for one-size-fits-all solutions. Use cases vary dramatically, legacy data is scattered, and governance models differ by region and business unit. No single partner can meaningfully cover all these bases.

2. Start with outcomes, not offers

One of the most consistent themes across FutureNet World 2025 was the need for CSPs to start with clarity on business outcomes, not technology features. That clarity is essential when choosing AI partners.

“Vendors all have their areas of expertise,” Hamilton noted. “The question is, how do we fit into your AI strategy?”

For VMO2, this meant starting with messy but valuable operational data and defining real-world use cases, such as understanding the cost drivers of service faults, before selecting platforms or partners. Only then could they engage the right expertise to extract value.

This is why for AI the traditional IT procurement model must be reversed. Instead of sourcing a tool and finding a use case, CSPs need to define the problem and build the solution outward, selecting and working with the right blend of partners.

3. Work with your data – even if it’s messy

The most valuable use cases for AI often lie in the messiest data. As Tadayon explained, when it comes to data, it’s not about how “messy” the data is, it’s about how “accessible and diverse” it is.

CSPs often assume they need to “fix” all their data before starting, and in so doing risk delaying transformation indefinitely. But as VMO2 has demonstrated, real-world results come from doing AI with messy data, not waiting for perfection.

The key is identifying partners like Infovista who understand the operational complexity of telecom environments, from navigating fragmented, legacy data across siloed OSS/BSS systems to addressing the broader business impact of service degradation. CSPs need to work with domain experts who operate in messy environments and who understand the specific contours of telco data complexity.

4. Fit-for-purpose over “one partner to rule them all”

AI in telecom spans a broad range of domains, from predictive maintenance in networks to churn analysis in marketing, and fraud detection in finance. Each of these areas requires different data sources, models and expertise.

Rather than trying to consolidate everything under one provider, CSPs must adopt a modular partnership model, selecting best-in-class partners for specific needs and integrating them into a coordinated operating framework.

That means:

• Cloud providers for compute and scalability
• Model specialists for specific business problems
• Systems integrators for orchestration
• Domain experts with telco DNA
• Ethical AI consultants to oversee governance and transparency

5. ROI still matters

Amid all the hype and seemingly unlimited upside to AI, it’s easy to lose sight of a metric that still really matters: ROI.  This is because with AI, as Hamilton explained, “there’s infinite possibilities to transform your organization”, and with it a natural tendency to “cut people loose” to experiment.

“The problem is, when teams start to work on this unchecked experimentation, for a CEO or for a business leader, that’s money you’re spending. It’s interesting, and it’s great that people are innovating, but you’re burning resources.”

As CSPs navigate their transformation journey to get to their desired outcomes, AI or no AI the one thing doesn’t go away is that ROI matters. This is as true for in-house AI projects as it is for the AI partners collaborating on the projects.

Or, as Hamilton summed it up: “You’re going to spend money and you better get the return, or life’s going to be difficult.”

Conclusion

As CSPs accelerate their AI journeys, the partner landscape will be a defining factor in success. But the smartest telcos will resist the allure of all-in-one solutions that promise the world. Instead, CSPs will take control of their own ecosystem, built to their own requirements and mapped to their own business objectives.