"The use case lens is too naive": NTT DATA's Jan Wuppermann on why APAC leaders need to stop asking the wrong questions about AI

SVP challenges enterprise fixation on proof-of-concepts, warns current infrastructure investments are already hindering AI adoption, and explains why quantum planning must start now despite 2030 timelines.

Twenty-two years navigating technology transformations across Europe, the Middle East, and Asia-Pacific has given Jan Wuppermann a clear-eyed view of how enterprises approach emerging technologies—and where they consistently go wrong.

As the senior vice president of service assurance and data & AI for NTT DATA's APAC operations, Wuppermann sees the patterns: companies demanding proof-of-concept trials for technology that's already proven, fixating on individual use cases instead of business transformation, and postponing infrastructure investments until capacity constraints become crises.

In a conversation following NTT's R&D Forum announcements—ranging from one million qubit quantum computers by 2030 to IOWN photonics networks controlling factories 300 kilometers away—Wuppermann offered blunt assessments of enterprise AI readiness, infrastructure bottlenecks, and the pragmatic reality of preparing for quantum computing while keeping today's operations running.

His central thesis: stop thinking about AI as a collection of use cases. Start thinking about business transformation.

The infrastructure crunch nobody's talking about

While NTT's quantum computing partnership and tsuzumi 2 LLM captured headlines, Wuppermann identifies a more immediate constraint throttling AI adoption across the region: enterprise network infrastructure designed for yesterday's consumption patterns.

"We don't see the bottleneck necessarily at the data center and backbone part, because that's pretty established," he explains. "We see more of the bottleneck at the enterprise network level. Most network infrastructure has been sized and optimized for current usage patterns. But AI is driving exponential demand growth on networks that weren't built to handle this intensity—that's creating pressure on existing infrastructure."

The numbers support his concern. NTT DATA's whitepaper warns AI workloads will consume 50% of data center power by 2028. But Wuppermann's focus is elsewhere—on the last-mile enterprise networks that connect businesses to cloud infrastructure and data centers.

"Our research, even last year, has already shown that current infrastructure designs and investments are hindering the adoption of AI," he states flatly. "It's not about future capacity—it's already happening."

He acknowledges the exponential growth creates planning challenges. "AI adoption is so exponential that getting ahead of that curve, or in line with that, is tricky. We typically have a bit of headwind where capacity becomes available at 80-85% utilization, then you unlock the next phase. But the growth curve is steeper than those investment cycles."

IOWN adoption: Strategic conversations, not mass market

When pressed on whether enterprises are demanding NTT's IOWN photonics network—demonstrated controlling manufacturing equipment across 300 kilometers with 20-millisecond latency—Wuppermann provides measured assessment rather than marketing spin.

"We do see demand. It is not yet the mass market stable offering. It is quite advanced, and it's very much driven by respective industries and their needs for low latency, high performance networks."

Banking leads adoption, particularly high-speed trading operations where latency determines profitability. Manufacturing follows, especially automated factory control. The public sector shows interest in specific applications. "Others are more strategically relooking at the infrastructure play for the next five to 10 years," he notes.

The objections are predictable: cost and timing. "It's a big commitment to upgrade and change your infrastructure from what you have today to a future-proof investment. It's human nature—we keep kicking the can down the road. These are very strategic conversations, and sometimes they're driven by necessity more than desire."

The necessity question matters because it reveals how infrastructure transformations actually happen. "The biggest driver for infrastructure investments is demand and consumption," Wuppermann explains. "Consumption comes through the right applications and the intensity of what they require in terms of latency or bandwidth. Then you get your normal break-even calculations."

He's skeptical of waiting for a single "killer app" to drive IOWN adoption. "I don't believe there's a single killer app. Certain industries are naturally more prone to investing first, and geographies lean themselves based on how IOWN is being rolled out. It's related to the ecosystem of infrastructure."

Singapore and India show demand. Geographic positioning and existing infrastructure maturity matter more than speculative applications.

Proof of value over proof of concept

Perhaps Wuppermann's sharpest criticism targets the enterprise obsession with proof-of-concept trials—a practice he considers fundamentally misguided for established technologies like AI.

"We've had clients who came to us and said, 'Help me build these five use cases. We need a POC to prove to the board.' I don't like the word POC. We use 'proof of value,' because the technology is working—I don't need to prove that. I need to prove whether I can create value in your environment with those use cases."

The distinction isn't semantic. POCs focus on technical feasibility. POVs focus on business impact, including baseline measurements, ROI calculations, and production deployment planning from the start.

"The POV is not just a technical trial. It's actually looking at impact—how do you do the ROI? Do you know what your baseline is? How do you measure the benefits? Once you're happy with it, how do you bring it into production? Think about it end-to-end, then start the journey."

He describes clients taking opposite approaches. "We had clients that listened to us and said, 'We're going to take a step back,' and we developed for them a 2025-2030 strategy where AI is embedded into the strategy. Other clients still have to deliver something to manage board pressure. We try to balance and say, 'Let's at least pick the right ones that make a difference.'"

The stakeholder question determines use case selection. "If it's a CFO, we've got to show something that has measurable P&L or balance sheet improvements. Otherwise you don't speak the language. So don't just focus on customer experience. What's the portfolio of benefits we want to demonstrate? Pick one or two areas of your business, not a random mix of everything."

The sustainability reality check

On sustainability integration—a topic prominent in NTT's messaging around tsuzumi 2's single-GPU efficiency and IOWN's low power consumption—Wuppermann offers candid assessment of APAC market dynamics.

"It is a priority. But I have to say that in APAC, because of the lack of consistent regulation, sustainability is not really at the forefront. Those driven by European headquarters or dependent on European markets or supply chains absolutely look into it.” He pauses before continuing: "But where concrete reporting and compliance regulations come in, companies are very, very quick to respond."

Singapore's data center moratorium earns his approval. "They said no one gets an additional license. You have to qualify. That was a very courageous step—the only body who's actually done this. It's only right."

The Malaysia contrast is instructive. Johor's data center boom following Singapore's moratorium created pride in the spillover benefits—but Wuppermann questions the sustainability of competing on cheap land and resources alone. "There's going to be natural pressure, whether it's fast enough or not—from my perspective, it can never be fast enough."

NTT is already moving proactively. "We already look at how we weave sustainability into data center design and consumption because we have to provide global clients a consistent footprint," he notes, suggesting regulatory pressure will eventually force the issue regardless of national policies.

Data sovereignty: Fragmented but manageable

Despite APAC's fragmented data sovereignty regulations—China's strict localization, Singapore's openness, Malaysia somewhere in between—Wuppermann doesn't see insurmountable complexity.

"We obviously always comply with national requirements, no different than GDPR or data privacy policies. When it comes to AI, we subscribe to global AI responsibility standards and the EU AI Act. Aligning to that gives us a minimum safeguard—if we meet those standards, we're compliant with pretty much every jurisdiction unless there's unique local developments."

Singapore, Australia, India, and Malaysia have nuances, "but it's not broadly different," he notes. China requires separate consideration.

"We welcome clarity on those standards. It makes many decisions easier, particularly for our customers." He hasn't seen clients changing AI providers by geography due to sovereignty concerns. "There's general enterprise architecture decisions—do I want to go with hyperscalers or certain components as sovereign AI? This is a more fundamental strategic decision, and then they just roll it out. Often the footprint of the providers plays a role."

The quantum question: Plan now, deploy later

On whether APAC CIOs should develop quantum-ready strategies today for technology arriving in 2030, Wuppermann provides pragmatic guidance balancing long-term planning with immediate operational pressures.

"The pragmatic view: CIOs can't have broad conversations about quantum. The question is, how are you going to keep my factory running today? What are you going to do to drive cost and efficiency today? Business pressures won't allow CIOs to jump to that next stage."

His recommendation: "Keep maybe a smaller team aligned with that, understanding as maturity comes closer, being ready in design, architectural, and investment parts. But pragmatically and practically, it's an ask for them to focus on that now."

However, he emphasizes infrastructure planning must extend beyond quarterly thinking. "If you're developing a strategy that doesn't just go three years—infrastructure investments never last only three years, they should last longer. Have a five to 10-year view on your fundamental architecture, then work it backwards into the next two to three years."

What APAC gets wrong about AI

After 22 years across regions, Wuppermann sees consistent patterns. On APAC's AI maturity: "I don't think it's necessarily a geographical maturity level. It's much more company maturity—the maturity of company processes, forward thinking, embracing technology as a differentiator in your core strategy, not looking at technology as a functional strategy at the tail end. Those players will be more successful, and that is independent of the country."

Singapore, Japan, and South Korea tend to lead in technology adoption, "seeing it as a normal part of life." He notes emerging markets' leapfrog advantage: "Mobile phone adoption was much faster because you didn't have infrastructure to replace. There's a leapfrog opportunity."

Governments in Malaysia, India, and Singapore "much more see the responsibility of grasping benefits of new technologies for population development. That's actually more pronounced than in the EU."

The mistake holding enterprises back? "There's still too much conversation on 'what's the use case?' It's less around 'how do I transform my business,' create value, then apply it and drive it in."

His analogy cuts through: "What's the use case for mobile phones? At the beginning, we thought you would talk occasionally. But today, it's my life. Nobody can leave it. That use case lens is still too naive."

The implication: stop asking what AI can do. Start asking how business models must transform.