India's stock market weakness is masking one of its strongest pockets of opportunity. While the Nifty 50 has declined 12% in 2026, a Goldman Sachs-screened basket of 42 Indian "AI Enablers" has surged about 60%, making it the best-performing segment by a wide margin. This 72 percentage point gap challenges the dominant view that India has little to gain from the global AI boom. The country has increasingly become the default "anti-AI" trade among major markets because its benchmark indexes have limited exposure to AI-related companies. But beneath the headline index performance, a cluster of companies tied to power, data centres, and semiconductors is benefiting from the infrastructure build-out required to support AI.
The composition explains much of this divergence. The Nifty 50 is dominated by financial services (36.18%), oil & gas (9.65%), and IT services (8.37%)—sectors that face disruption risks from AI rather than being infrastructure beneficiaries. In contrast, the AI enabler basket includes 8 microcaps, 13 smallcaps, 9 midcaps, and only 12 largecaps. This small and midcap skew has enabled the basket to outperform even the Nifty Midcap and Smallcap indexes, suggesting its gains aren't merely a function of broader small-cap momentum but are driven by specific AI-related earnings growth.
The rally has been fundamentally driven, not speculative. Since 2025, the AI enabler basket has returned 53%, with earnings growth contributing 65 percentage points while valuation compression reduced returns by 12 percentage points. This earnings momentum is expected to continue. Consensus estimates indicate earnings for the AI enabler basket could grow 53% in 2026, 39% in 2027, and 29% in 2028. This compares with expected 2027 earnings growth of 16% for MSCI India and 23% for the MSCI India Small and Midcap Index.
The basket trades at approximately 36 times forward earnings, an 85% premium to the MSCI India Index and near the upper end of its five-year historical range. However, when adjusted for earnings growth, the premium narrows significantly. The basket's PEG ratio of 1.3x is actually below MSCI India's 1.4x, suggesting the premium reflects stronger expected earnings rather than excessive valuation alone. This growth-adjusted valuation makes sense given the structural drivers supporting these companies.
Power generation, data centre hardware, and power equipment companies have emerged as the leading sub-sectors within the AI enabler basket, with expected earnings growth of 40-60%. This isn't coincidental—AI workloads are incredibly power-intensive. Global data centre power demand is forecast to hit 106 GW by 2035, a 36% jump from previous outlooks. India's data centre capacity has quadrupled in six years to 1.7 GW, with another 1.3 GW under construction. This build-out requires massive electrical infrastructure.
The earnings outlook is supported by a strong investment cycle. Nifty 500 capex growth is expected to more than double to 16% in 2026 from 7% in 2025, with AI enablers contributing about six percentage points to that increase. Goldman Sachs expects these companies to remain free cash flow positive despite higher capital spending. Power Grid Corporation, for instance, has revised its FY26 capex guidance upward three times and achieved ₹35,540 crore in capex (102% of guidance), with ₹82,000 crore outlined for FY27-28.
The AI theme is appearing more frequently in corporate disclosures. Goldman Sachs found that references to AI in management commentary have risen sharply over the past two years. Earnings calls increasingly include terms such as data centres, power purchase agreements, fibre infrastructure, transformers, switchgear, GPUs, OSAT, and data lakes. This shift in corporate language provides an early indicator of investment and capacity expansion.
The scale of commitments is staggering. Reliance Industries has committed ₹10 lakh crore for multi-gigawatt AI-ready data centres over seven years. Adani Enterprises plans to invest $100 billion by 2035 to build renewable energy-powered AI-ready data centres, scaling capacity from 2 GW to 5 GW. Larsen & Toubro secured a ₹10,000-15,000 crore order from Together AI for an AI data centre with 10,000 NVIDIA B300 chips. These aren't announcements—they're capital deployment.
Data centre operators account for approximately $400 billion of the AI enabler basket's total market value of $670 billion—significantly exceeding power equipment ($100 billion), power generation ($50 billion), and power transmission ($45 billion). This valuation dominance reflects their position at the convergence of multiple powerful trends: market size (India's data centre market growing from $5.55 billion to $13.11 billion by 2034), strategic importance (critical infrastructure for AI and data sovereignty), and ecosystem integration (multiple revenue streams across connectivity, cloud, and AI infrastructure).
Companies are leveraging existing infrastructure to capture this demand. Adani Energy Solutions provides transmission lines, substations, and distribution infrastructure for integrated data-centre campuses within the Adani ecosystem. Power Grid operates under a regulated return framework providing stable earnings visibility for AI-driven electricity demand. Bharti Airtel's Nxtra raised $1 billion to scale capacity from 300 MW to 1 GW, combining AI-ready data centres with Airtel's nationwide fibre network and edge infrastructure.
The relationship between AI infrastructure build-out and power equipment demand is direct and predictable. AI infrastructure requirements create specific equipment demand: high-density power distribution requires low-voltage panels, switchgear, and gas-insulated switchgear; grid connection needs transformers and GIS substations; backup power demands diesel generator sets. This causal relationship creates substantial revenue visibility.
The evidence is in the order books. ABB India reports data-centre orders made up 15-17% of base orders in a recent quarter, with half of India's hyperscale facilities running on ABB solutions. Hitachi Energy India claims roughly one in two Indian hyperscale data centres already runs on its equipment, with a record order backlog of ₹29,555 crore (up 53.5% YoY). CG Power secured a ₹900 crore power-transformer order for a US hyperscale data-center project—its largest export order to date. Cummins India reports data centres accounted for 40% of domestic power-generation revenue in Q1 FY27, up from 23% a year earlier.
Goldman Sachs explicitly acknowledges risks in its screening methodology. The top-down approach can include false positives because companies may reference AI terminology without having substantial revenue-generating exposure. The screen's liquidity and market cap filters may also exclude some potential beneficiaries, particularly emerging AI enablers that are too small or illiquid to qualify. This creates survivorship bias and sector coverage gaps.
Valuation risk is real. The basket trades near the upper end of its five-year historical range, and if the aggressive earnings growth expectations (40-60% for leading sub-sectors) are not met, significant multiple compression could follow. Historical parallels are cautionary—technology booms often overcapitalize the installation phase. The dot-com episode demonstrated that real technology and inflated valuation can coexist. Internet infrastructure ultimately became central to the economy, yet many public internet firms were dramatically overvalued relative to eventual cash flows.
There is significant valuation divergence within the AI enabler group. While some sub-layers trade at demanding valuations, others remain closer to their historical averages. This divergence reflects market differentiation based on revenue quality, competitive positioning, and execution track record. Data centre operators and semiconductor companies with contracted revenue, strong order books, and market leadership command premium multiples. Traditional power generation, transmission utilities, and construction companies trade closer to historical averages due to regulatory constraints, cyclical exposure, and execution risks.
This divergence means stock selection matters significantly. Investors should focus on companies with visible revenue (long-term contracts, strong order books), sustainable competitive advantages (market leadership, technological moats), proven execution capability, and reasonable growth-adjusted valuations. The AI infrastructure theme represents a genuine structural opportunity, but success requires disciplined stock selection, active risk management, and a long-term perspective. The 42-company basket isn't a monolith—it's a collection of individual stories, some compelling, some concerning.
