The AI boom has captured the spotlight: larger models, smarter systems, and generative breakthroughs that promise to remake everything from search to supply chains. But as generative AI applications grow more complex and embedded into daily operations, a more structural challenge is surfacing: most of our digital infrastructure wasn’t built for this.
From fleet optimization to energy management, AI is reshaping operational backbones across industries. Yet, many of these systems are only as good as the networks and compute layers they run on. The performance ceiling isn’t just model quality—it’s bandwidth, latency, compute proximity, and power. In short, it’s infrastructure.
Piyush Jain has been watching this transformation unfold up close. Over nearly two decades, he’s helped structure multi-billion-dollar telecom and digital infrastructure deals across North America, Europe, and Southeast Asia. Piyush notes that while AI has captured the global business imagination, many companies underestimate how important the right network architecture is to AI’s success. “Everyone’s focused on what AI can do, but few are thinking about what it needs to get better,” he says. “AI workloads demand ultra-low latency and massive bandwidth—capabilities that most commercial networks today weren’t designed for.”
This gap is fueling a wave of investment and innovation in connectivity infrastructure—fiber, 5G, data centers, edge computing, and power—all aimed at solving AI’s rising demands.
Fiber, 5G, and Edge: The Building Blocks of AI-Ready Networks
For AI applications to function with minimal delay, data must be processed closer to where it’s generated. This is a departure from the traditional model, where information travels from a device to distant cloud servers and back again. Piyush points to three areas in particular—fiber, 5G, and edge computing—that are becoming the technical foundation for AI-first businesses.
Fiber-Optic Networks
Fiber is the foundational layer. Piyush calls it “the critical path” for AI-ready infrastructure. With its unmatched bandwidth and stability, fiber is fast becoming the standard for high-throughput, scale-driven enterprise and consumer networks.
“In countries like Indonesia and Thailand, where we’ve advised on their leading telecom investments, we’re seeing fiber prioritized over traditional copper or hybrid solutions,” Piyush says. “It’s not just speed. It’s the predictability you need when AI is embedded into real-time systems.”
5G & Network Slicing
Wireless is getting smarter too. 5G’s promise isn’t just speed—it’s customization. Through network slicing, businesses can create dedicated connectivity profiles for different AI applications, isolating workloads based on latency, throughput, or reliability requirements.
“A fully AI-powered city or factory will need different connectivity profiles for different tasks,” Piyush explains. “5G makes that segmentation possible, which is crucial when everything from traffic systems to industrial robots is being guided by AI models.”
Data Centers & Edge Computing
AI’s real-time potential only materializes when compute happens close to the data source. That’s where edge computing comes in. Rather than routing everything back to centralized cloud servers, edge nodes process data locally—reducing round-trip time and enabling faster, context-aware decisions.
“It’s about collapsing the decision loop,” Piyush explains. “We’re optimizing systems where every millisecond counts. That’s what separates a working industrial AI system from one that just looks good in a demo.”
Industries like manufacturing, logistics, and utilities are leading the charge, where an intelligent system’s ability to interpret sensor data instantly can prevent downtime, reduce waste, or avoid safety incidents. The economics are clear: better infrastructure enables better outcomes.
Equally important is the design of the data centers powering AI. Traditional facilities weren’t built to handle the density and power draw of AI workloads. Supporting generative models—whether for large-scale inference or fine-tuning—requires new infrastructure: liquid cooling, high-rack densities, robust redundancy, and direct GPU access.
Power & Energy
All of this runs on power—and lots of it. The energy demands of AI training and inference are outpacing those of traditional data centers. Some estimates suggest electricity consumption from AI-related compute could double global data center energy use by 2030. It will become increasingly important to invest in infrastructure that provides this stable, scalable power capacity, whether through grid modernization or next-gen renewable energy.
Investment & M&A in AI-Optimized Networks
This surge in demand is translating into hard investment, with spending on AI infrastructure projected to exceed $200 billion by 2028. Private equity firms, sovereign wealth funds, and infrastructure giants are channeling billions into fiber deployments, edge platforms, and 5G rollouts—not just as telecom plays, but as bets on AI-readiness.
“AI is becoming a core part of the valuation model,” Piyush says. “We’ve worked on deals where a client’s future earnings—and even its sale price—were tied to how well it can reap benefits through AI, either as a revenue source or to optimize its current business operations.”
In Southeast Asia alone, Piyush has led strategic projects representing $10 billion in value, advising clients on everything from pre-deal structuring to post-deal optimization across over 200 million subscribers. The focus: Aligning technology assets such as fiber, networking, and data centers with the needs of emerging technology, improving EBITDA through infrastructure rationalization and smarter capital allocation.
The flip side of this, Piyush warns, is underinvestment. Clients that delay are at risk of being locked out of the AI economy altogether—unable to meet the technical demands of enterprise clients, or compete for the kinds of partnerships and revenue streams that AI-powered services enable.
The Strategic Shift in AI infrastructure
What’s emerging is a redefinition of the networks, data centers, and other AI infrastructure assets. No longer a utility or a cost center, they are becoming a strategic differentiator—one that can unlock new business models, enable automation at scale, and create defensible moats in competitive industries.
“The people making fiber, 5G, and edge decisions today aren’t just optimizing IT spend—they’re shaping how their companies compete in an AI-driven world,” Piyush says.
For governments, it’s about national productivity and digital resilience. For enterprises, it’s about agility, innovation, and market access. And for investors, it’s about long-term positioning in a rapidly changing value chain.
The views and opinions expressed are solely those of the author and consultant and do not necessarily reflect the views or opinions of the firms mentioned.
