Semiconductor horizons: AI, energy, and modular innovation redefining next decade

The semiconductor industry stands at a historic inflection point. As artificial intelligence accelerates across sectors, the demand for computing power and the energy to sustain it is surging at an unprecedented pace. 

For India, this moment offers both opportunity and urgency: to bridge gaps in the value chain, embrace sustainability, and build a resilient ecosystem that can leverage this opportunity to develop a semiconductor manufacturing base, innovate and compete globally.

Inflection 1: AI and energy – From sustainability to sustainable abundance
AI’s appetite for computing is rewriting the global energy demand curve. Today, data centers already consume nearly 2% of global electricity, and forecasts suggest this could rise to 10–12% by 2030, rivaling entire nations’ consumption. Every new Large Language Model (LLM) and multimodal AI workload adds exponential complexity, driving power requirements skyward. 

While the estimate of AI datacenters consuming 10-15% of global electricity might seem too high, it is instructive to consider that according to a 2022 study, lighting alone consumes around 15% of global electricity. If AI is viewed as a significantly more productive technology, then it should be enabled to consume as much or more electricity to raise global productivity.

Traditional sustainability approaches, like offsetting emissions or incremental efficiency gains, are no longer enough. Industry leaders advocate for a new paradigm: sustainable energy abundance. This strategy calls for holistic grid modernization, integrating renewables, nuclear, and transitional conventional sources to ensure AI’s productivity gains are not throttled by energy constraints. 

India’s energy imperative
The demand for data centers in India is being driven by the government’s Digital India and data policies, increased data consumption, and the rollout of 5G, which will enable adoption of data-intensive technologies such as IoT and AI. India has twice the number of internet users compared to Europe, yet its data center capacity remains relatively modest at 1.4 GW versus Europe’s 10 GW. 

With data privacy regulations taking effect and AI adoption accelerating, projections indicate that India’s capacity could triple by 2027 and grow five-fold by 2030, depending on buildout scenarios.

For India, this is a great opportunity to:
• Modernize the grid: Upgrade transmission infrastructure and deploy smart distribution technologies.
• Align policy: Link semiconductor incentives with clean energy adoption.
• Plan regionally: Select fab and data center locations based on carbon intensity of local grids.

Inflection 2: Advanced packaging and chiplets – Modular innovation
The semiconductor industry delivered a 10,000x performance improvement over the last 15 years. Exponential growth in AI computing requirements means another 10,000x improvement in energy-efficient performance is required by 2040. 

As monolithic scaling slows (commonly associated with the limits of Moore’s Law), chiplets and heterogeneous integration are emerging as game-changing technologies to enable systems with higher performance and lower power consumption. 

Open chiplet standards allow designers to mix and match best-in-class dies logic from one vendor, memory from another without re-laying out entire systems. This modular approach reduces costs, accelerates innovation, and enables customization for AI workloads.

Rising demand for AI, data centers, smartphones, and automotive electronics is driving the need for high-bandwidth, low-latency, and power-efficient chips, making advanced packaging essential. It overcomes traditional scaling limits by enabling heterogeneous integration, combining different chip types to enhance performance and energy efficiency. 

Advanced packaging is also a cost-effective alternative to advanced node fabrication, improving chip functionality without relying solely on complex process nodes.

India’s role
India’s semiconductor design ecosystem, already strong in system-on-a-chip (SoC) and IP development, can lead in chiplet-based architectures. However, this requires:
• Pilot packaging lines.
• Shared R&D platforms.
• Partnerships with global OSATs.

The Research Development and Innovation (RDI) Scheme Fund is a positive step towards strengthening India’s R&D ecosystem. Leveraging this scheme in a private-public partnership model will be critical. 

Inflection 3: Power semiconductors and renewables – Enabling AI and clean energy
AI’s growth will drive demand for datacenter power semiconductors, projected to reach $9 billion by 2030, with accelerators accounting for $7 billion. Gallium nitride (GaN) and silicon carbide (SiC) will dominate high-efficiency power conversion, not just in datacenters but in renewable energy systems. 

The renewable energy semiconductor market is projected to have a TAM of approximately $23 billion by 2030 for semiconductor revenues, creating opportunities for India to localize production and reduce import dependency.

India is pushing for an ambitious target of 500 GW renewable power capacity by 2030, while also working to reduce power deficits. Power electronics HVDC, inverters, STATCOMs, and energy storage systems will be vital for grid stability, electric mobility infrastructure, and AI-driven demand. Building a domestic power semiconductor manufacturing base is India’s key to its net-zero goals. 

Road ahead: Talent, collaboration, and India’s moment
As we move into 2026 and beyond, the semiconductor ecosystem will increasingly unite to co-create curricula and deliver hands-on training. Industry and academia will collaborate through joint labs and apprenticeship programs not just here in India, but become a collaborator on critical areas worldwide, ensuring that learning is deeply practical and relevant globally. 

Curricula will need to be co-designed to meet the evolving demands of advanced packaging, materials engineering, and AI hardware. 

This collaborative spirit is not optional; it is essential. The complexity of semiconductors in the AI era calls for a high-velocity model of co-innovation. Shared R&D platforms will accelerate breakthroughs in materials and packaging, while open standards for chiplets and modular architectures will enable interoperability and innovation at scale. 

Joint roadmaps will define energy efficiency and sustainability metrics, ensuring progress is both rapid and responsible. This is not about marketing alliances it is about bridging physics gaps that no single company can overcome alone. For India, collaboration will be the cornerstone of global competitiveness.

Delivering on the promise
2026 will test the semiconductor industry’s ability to innovate at scale and speed while staying within global energy limits. By embracing sustainable energy abundance, investing in advanced packaging, and closing the value chain gap, India can seize this moment.

But, manufacturing alone is not enough. While “Make in India” can help narrow the gap in chip production, “Invent in India” will define leadership. To achieve this, we must commit to deep R&D, indigenous product development, and bold co-innovation models that unite industry and academia. 

Shared research platforms, joint labs, and open standards will accelerate breakthroughs in materials, packaging, and AI hardware—bridging physics gaps that no single company can overcome alone.

The next decade belongs to those who innovate collaboratively and sustainably. India has the talent, ambition, and policy momentum. India has the talent, ambition, and policy momentum. It is time to deliver!

-- Avi Avula, President, Applied Materials India.