During a closed roundtable at HCLTech’s pavilion during the 2026 World Economic Forum, C-suite and senior leaders explored how innovation, ecosystem collaboration and evolving business models are reshaping the tech industry.

The conversation spanned AI economics, infrastructure constraints, edge and device shifts, services transformation and the next frontier of Physical AI; anchored by a shared question: what does value mean now, and how should leaders measure it as technology becomes embedded everywhere?

Below are seven of the most important takeaways from the discussion.

1. ROI is not the same thing as impact, and leaders need both

Many organizations can see clear impact from AI: time saved, faster cycles, better experiences, improved quality. The harder step is converting that into ROI that shows up in financial outcomes. Leaders returned to a consistent pattern: productivity gains do not automatically become economic gains without deliberate operating model change.

A practical implication emerged: measuring AI success requires a broader scorecard than topline and bottom-line alone. Experience metrics, including employee effectiveness and customer satisfaction, process performance, including cycle time and error rates and innovation velocity matter, especially early, while finance leaders increasingly want traceability from time saved to value captured.

2. The AI cost model is becoming a governance problem, not just a procurement choice

As AI shifts from packaged licenses to usage-based consumption, the cost structure becomes more dynamic, and harder to predict. Leaders described a new reality: token-driven spend can grow faster than expected once AI becomes embedded into core workflows and agent-to-agent automation accelerates usage behind the scenes.

That changes governance. It pushes organizations to treat AI like a managed resource with usage controls, observability and unit economics, rather than a fixed software line item. The takeaway was not avoiding consumption, but designing for it: establishing cost guardrails, measuring yield and building discipline into how agents, models and workflows are orchestrated.

3. Efficiency gains are arriving before productivity gains, especially in regulated systems

A recurring theme was the gap between doing the same work faster and doing more, or delivering more value, with the same resources. In some sectors, AI is already reducing after-hours work and compressing admin burdens, but that does not automatically translate into higher throughput or lower total cost.

Leaders pointed to why: regulation, service expectations and institutional constraints can prevent organizations from converting efficiency into productivity. Without policy, incentive and workflow redesign, the benefits land as less friction rather than more capacity. The leadership challenge is to decide where that is acceptable, and where the organization must push through change to unlock measurable productivity.

4. AI is forcing a services reset: From labor delivery to service redesign

For service providers and enterprise buyers alike, the discussion surfaced a shift in what customers will demand: not simply the same service delivered cheaper, but a redesigned service that produces better outcomes. AI changes both the unit cost and the unit of value.

Leaders described a move away from incremental optimization toward step-change commitments tied to new delivery models, where automation, agents and tooling are built into the service itself. That also changes how buyers evaluate partners: less focus on staffing plans, more on measurable outcomes, adoption and evidence that learning and behavior are changing inside the client organization, not just course completions or dashboards.

5. Infrastructure strategy is becoming a competitiveness strategy

The group returned often to the idea that where compute happens will be shaped by latency, data locality, reliability and economics, not ideology. AI is pushing a more distributed model: cloud, data center, edge and device all matter, and the optimal architecture depends on the use case.

At the same time, leaders flagged that AI economics is not just about compute efficiency. Energy and memory costs, and their uneven global distribution, are increasingly decisive. Regions with structural advantages in energy can attract investment and capacity faster, while constraints in memory supply and ecosystem concentration can create volatility. The upshot: infrastructure planning is no longer a back-office IT topic; it is a board-level lever for resilience, cost control, and speed to value.

6. Sovereignty is less about self-sufficiency and more about resilience

The conversation challenged a simplistic view of sovereign AI as building everything independently. In an interconnected supply chain, complete self-sufficiency is unrealistic for most players. What leaders emphasized instead was resilience: diversified dependencies, clearer risk posture and the ability to operate under constraint.

That reframes decisions around where workloads run, how data is governed, what partnerships are strategic and which parts of the stack require stronger control. Leaders treated sovereignty not as isolation, but as a pragmatic approach to continuity, compliance and competitive positioning in a fragmented geopolitical environment.

7. Physical AI is moving from concept to category, and it will reshape labor questions

Beyond digital workflows, the discussion turned to Physical AI: robotics, sensor-driven autonomy and real-world systems that combine perception, planning and action. Leaders noted that new product categories are emerging, not just incremental upgrades, enabled by smaller on-device models, edge inference and tighter hardware-software integration.

This is where value debates become more complex. Physical AI raises adoption questions around trust, safety, liability and cultural acceptance; often more than technical capability. It also sharpens the workforce impact: once autonomy expands beyond software tasks into physical tasks, the displacement and redesign of work could accelerate in new sectors, such as logistics, hospitality and industrial operations. The leadership imperative is to frame Physical AI around jobs to be done and outcomes; then design systems, safeguards and change programs accordingly.

Redefining value as AI becomes invisible

Across themes, the roundtable converged on one message: AI is becoming embedded rather than optional, and that forces a reset in how value is defined and captured. The winners will be the organizations that treat AI as an operating model shift, linking usage to outcomes, redesigning services and workflows, investing in resilient infrastructure and preparing for the Physical AI wave with safety and governance built in from the start.