According to NTT DATA's latest research, more than 96% of organizations believe their current infrastructure is slowing down their AI initiatives. As Dilip Kumar, president and global head of infrastructure solutions at NTT DATA, explains, the constraint on scaling AI was never really the model - it's whether the network, data, security and operating-model layers underneath it were built to support it. The challenge points to a deeper enterprise problem where AI readiness has become harder than AI adoption, with the costliest infrastructure mistakes surfacing only after the model is licensed. This reality contradicts the common perception that AI scaling depends primarily on model size and computational power.
AI infrastructure represents a fundamental misunderstanding in current discussions, as reported by ABP Network. The term 'AI infrastructure' describes two entirely different ecosystems - one that creates Artificial Intelligence and another that deploys AI. The first infrastructure is measured in gigawatts, semiconductor fabs and hyperscale computing, while the second focuses on engineering innovation, embedded electronics, sensors and intelligent products. This distinction changes how countries should approach AI investments and where companies like Unwired Connect fit into the AI economy. NTT DATA's assessment confirms this duality, noting that AI is becoming a global capability that depends on local talent, local ecosystems and local execution, with buying a model alone not helping companies operationalize at scale without the right talent to convert intelligence into tokens.
As reported by ABP Network, training frontier AI models requires a comprehensive physical infrastructure pyramid. Reliable electricity forms the base, as modern AI data centres consume extraordinary amounts of power that must be managed through sophisticated thermal systems including liquid cooling, chillers, precision HVAC systems and water infrastructure. Above power sits semiconductor manufacturing for GPUs, CPUs, NPUs, HBM memory and networking ASICs, which depend on silicon wafers, lithography, deposition, etching, advanced packaging, testing and specialized materials. The final layer involves complex multi-layer printed circuit boards assembled through automated SMT production lines using SPI, pick-and-place systems, reflow soldering, AOI, X-ray inspection and functional testing.
According to ABP Network, India has established itself as a global force in software engineering and semiconductor design, but leading-edge chip fabrication, advanced packaging, AI accelerators and hyperscale AI infrastructure remain concentrated in a handful of countries. The author acknowledges this as engineering realism, noting that semiconductor ecosystems cannot be created overnight and require decades of investment, policy continuity, academic collaboration, manufacturing capability and strategic global partnerships. NTT DATA's research confirms this reality, with the company using sustainability IT assessments and frontier assessments to understand how clients want to modernize and innovate, focusing on energy, carbon or frontier models as a lens to understand business modernization needs.
As reported by ABP Network, deploying AI presents equally important opportunities in embedded processors, sensors, reliable connectivity, efficient power electronics and robust manufacturing. Across India, AI is already transforming fintech, healthcare, logistics, agriculture, manufacturing and public infrastructure, with most systems relying on models created elsewhere but generating enormous value by solving local problems. NTT DATA's approach emphasizes right-sizing based on specific use cases, noting that for visibility purposes, clients don't need 70-billion-parameter models - it's about choosing the right model for the business case. The company's five-layer stack includes core infrastructure (network, data center, security and workplace), data intelligence, identity management, model layer decisions based on ROI rather than model size, and operating model implementation where intelligence is converted into tokens for measurable business outcomes.
According to ABP Network, Unwired Connect has positioned itself to embed intelligence into existing infrastructure rather than building foundation models. The company focuses on lighting systems as one of the largest consumers of electricity in commercial buildings, transforming traditional luminaires into intelligent decision-making systems through IoT connectivity and embedded AI. FMCW radar-based AI and vision-based AI enable luminaires to make decisions locally at the edge without relying on constant cloud connectivity or expensive computing infrastructure, delivering practical outcomes including lower energy consumption, improved occupant comfort and smarter buildings. This approach aligns with NTT DATA's emphasis on agentic operating models and outcome-based deployment, where clients can demonstrate value within 90 days using the right model, right use case and verifiable ROI.
