How design decisions shape supply chain outcomes

This article is authored by Parthiban Arunasalam, chief product officer, L&T Semiconductor Technologies Limited.

The rapid expansion of artificial intelligence, high-performance computing, autonomous systems, and hyperscale networking is pushing semiconductor systems toward far more complex and heterogeneous integration. As AI workloads intensify and chiplet-based architectures become standard, engineering teams must think beyond transistor scaling. They are now required to design systems that simultaneously manage performance, power delivery, latency, thermal density, integration complexity, and manufacturability.

AI (Photo credit: Unsplash)

This evolution is transforming the role of design across the semiconductor value chain and creating a significant opportunity for India.

For many years, design and manufacturing operated as largely independent domains. Design teams focused on functionality and performance, while manufacturing ecosystems concentrated on scale and operational execution. That separation is no longer sustainable. Architectural decisions now influence packaging feasibility, substrate availability, interconnect maturity, and overall supply chain resilience.

This convergence is happening at a moment when India is expanding its semiconductor ambitions. The country’s strength in design, combined with new investments in OSAT, ATMP, and substrate manufacturing, is positioning India to contribute meaningfully to system-level innovation. As India shifts from engineering services toward full-stack product ownership, competitive advantage will depend on how effectively design teams anticipate downstream constraints before they become costly obstacles.

Packaging demonstrates this clearly. In AI accelerators and high-performance computing, packaging choices shape bandwidth, latency, thermal density, and power integrity. For chiplet-based designs or systems using HBM and 2.5D interposer-based integration, early decisions such as die partitioning, bump pitch selection, interconnect standards, hybrid bonding readiness, and thermal stack design determine integration feasibility. These choices also influence signal integrity at high-speed die-to-die links, power delivery network behaviour, and cooling architecture. They further determine whether India’s emerging OSAT ecosystem can support the product.

Materials are equally critical. ABF substrates, silicon interposers, RDL layers, and high-density laminates define routing density, impedance control, warpage behavior, and manufacturability. Substrate availability remains a global constraint, and India’s substrate ecosystem is still developing. When a design depends on specialised interposers or high-density substrates, supplier readiness and regional diversification become essential. Addressing these dependencies early reduces risk and prevents late-stage redesigns that affect schedules and qualification both in India and globally.

Resilience begins at architecture rather than in production. Engineering teams must design products with optionality across foundries, nodes, package stacks, and substrate suppliers. This is particularly important for India, where the ecosystem is expanding but still building capacity in advanced packaging and materials.

This requires asking more rigorous questions at the outset. For example, does the selected node have practical capacity across multiple regions? Can the packaging ecosystem support the required bump pitch, thermal density, and interconnect bandwidth? Are substrate and interposer supply chains diversified enough to avoid geopolitical or capacity-related bottlenecks? Can the design be validated across both domestic and international OSAT partners?

Chiplet-based architectures highlight these considerations. Partitioning a monolithic die into smaller chiplets can improve yield and design flexibility. However, it also introduces new challenges in die-to-die interconnect design, PHY selection, latency budgeting, test coverage, and system-level validation. The trade-off is no longer limited to silicon efficiency. It now involves balancing yield gains against integration complexity, packaging constraints, and ecosystem readiness, including India’s ability to support the product at scale.

These questions are central to modern semiconductor product strategy. In an industry where timelines are tightening and architectures are becoming more complex, the ability to embed flexibility into the design from the outset determines whether a product scales smoothly or encounters friction later.

The next phase of semiconductor innovation will be shaped not only by what can be designed, but also by what can be manufactured, packaged, and qualified reliably. India’s advantage lies in its design leadership, and the country’s ability to convert this strength into system-level products will define its long-term position in the global value chain.

The companies that succeed will be those that minimise friction between design intent and execution reality. Achieving this requires making smarter trade-offs earlier across packaging, yield, materials, interconnect, and integration. It also requires aligning design decisions with the capabilities of the manufacturing and supply chain ecosystems that support them, including those emerging in India.

India’s semiconductor future will be shaped by engineering teams that understand the full system context. The transition from silicon to system is already underway, and India is positioned to contribute significantly to this evolution.

(The views expressed are personal)

This article is authored by Parthiban Arunasalam, chief product officer, L&T Semiconductor Technologies Limited.