As AI continues to evolve from buzzword to business driver, infrastructure leaders face mounting pressure to turn proof-of-concepts into enterprise-scale results.

In a recent Leadership Lounge webinar hosted by Digital Realty, Chris Sharp, CTO of Digital Realty and Matt Hull, VP of global AI solutions at Nvidia, shared a powerful conversation on what it takes to scale AI successfully.

Key takeaways from the Leadership Lounge

I had the opportunity to listen in – and below are five insights that stood out to me as especially relevant for today’s enterprise leaders navigating the AI journey.

1) AI factories are the backbone of scalable AI

To achieve success with AI at scale, enterprises need high-density, liquid-cooled, and performance-validated purpose-built infrastructure – what Nvidia calls “AI factories.” These aren't just centralized data centers but interconnected ecosystems that span from the cloud to the Edge.

“We pride ourselves on what we call ‘AI factories.’ It’s about scaling your AI factory from the multi-tenant cloud to the self-driving car or the robot performing surgery at the Edge,” Hull noted.

These factories serve as the foundation for real-time, high-impact AI applications across various industries – from model training in centralized colocation to low-latency inference at the Edge.

2) Shadow AI is creating siloed chaos – CIOs must reclaim control

As enterprise teams independently adopted AI, many created isolated GPU-enabled environments outside of central IT oversight, leading to what Hull termed "shadow AI."

“They turned to their internal IT departments but often got the Heisman. So these enterprises have these little pockets of AI, and I think what we’re seeing is a realization that they need to centralize these efforts.”

Consolidating these siloed initiatives isn’t just a technological challenge, it also requires fostering cultural change, data sharing, and governance.

3) Where your data lives matters more than ever

Enterprises are recognising that having data is only the first step – it must also reside near GPU clusters and interconnection fabrics to enable fast, cost-effective training and inference.

“You’re going to bring your compute to your data – whether that’s in your core data center or out on the light post at the edge,” Hull stated.

This is where Data Gravity and infrastructure proximity become crucial for optimizing AI performance.

4) The market demands simplicity and results

AI solutions should be straightforward and deliver immediate value, not a complex mix of hardware, software, and services.

Digital Realty and Nvidia are addressing this need by creating repeatable, performance-verified reference architectures, such as ServiceFabric and Nvidia DGX deployments, to de-risk and accelerate AI adoption.

5) CIOs are the linchpin of enterprise AI success

Hull concluded the discussion by urging IT leaders to take a strategic role in shaping the AI landscape.

“They are the heroes. There are three legs to the AI stool: data, compute, and people. The data and compute aspects are firmly within the CIO’s domain.”

CIOs who embrace this moment – by aligning infrastructure with business outcomes, forming the right partnerships, and breaking down silos – can unlock significant business value with AI.

Embracing the AI revolution

AI isn't just a technological advancement; it's a fundamental shift in how we operate. Success in this new landscape hinges on moving quickly, building trust, fostering interconnectedness, and having strong leadership. The future is here, and those who adapt will thrive.

Enterprises that proactively align their data, compute, and interconnection strategy – while planning for TCO, lifecycle refresh, and application-specific latency requirements – are best positioned to thrive.

“AI is going to be uncomfortable,” Sharp reminded us. “So get comfortable with being uncomfortable – and use that to your benefit.”

We're proud to collaborate with organisations like Nvidia to help make AI real, repeatable, and valuable. As you chart your next move, consider how your infrastructure, data strategy, and organizational model can rise to meet the moment.

Are you optimizing for innovation? Watch the Leadership Lounge webinar AI: Enabler or Obstacle to hear the full discussion between Chris Sharp, CTO, Digital Realty and Matt Hull, VP, Global AI Solutions, Nvidia.