Goldman Sachs sees earnings upgrade cycle in India’s AI trade as Nifty estimates get cut

Reports coming in for today mention that India may have been left behind in the global artificial intelligence surge at the index level, but beneath the benchmark a group of firms linked to the AI infrastructure build-out is seeing earnings estimates move sharply elevated, according to Goldman Sachs.

Calendar 2027 earnings-per-share estimates for Goldman’s 42-firm basket of Indian “AI Enablers” have risen 22 percent so far in 2026, even as CY27 EPS estimates for the broader Nifty 500 have been trimmed 2 percent.

Data-centre hardware has led the upgrades, with estimates for the segment boosted 95 percent, followed by power equipment at 15 percent and semiconductor materials at 8 percent. The revisions are not broad-based, that stated: estimates for data-centre developers have been trimmed 15 percent, while those for power generation have fallen 14 percent.

The earnings divergence comes after a striking gap in stock-market performance.

Goldman’s AI Enabler basket has advanced around 60 percent in 2026, while the Nifty has fallen around 12 percent. Goldman describes the cohort as the strongest-performing pocket of the Indian market this year, with healthcare, up around 10 percent, the next best-performing segment.

More importantly, the bank’s analysis suggests the surge has not been fuelled simply by market participants paying elevated valuations for anything associated with AI.

Since 2025, the AI Enablers have returned 53 percent, with earnings expansion contributing 65 percentage points to returns while valuation compression detracted 12 percentage points.

Goldman stated the move has been fuelled by earnings accumulation rather than speculative multiple expansion.

Earnings are catching up with the surge

Consensus estimates compiled by Goldman point to earnings expansion of 53 percent in 2026, 39 percent in 2027 and 29 percent in 2028 for the AI Enabler cohort.

For comparison, MSCI India is anticipated to post around 16 percent earnings expansion in 2027, while the MSCI India Small/Mid index is anticipated to grow earnings around 23 percent.

Goldman anticipates all nine AI-infrastructure sub-layers in its basket to grow earnings faster than the broader Nifty 500 in 2027.

Power generation and data-centre hardware stand out.

Anticipated CY27 earnings expansion is around 62 percent for power generation and 57 percent for data-centre hardware, compared with 39 percent for the overall AI Enabler basket. Power equipment is anticipated to grow around 38 percent, while data-centre developers are anticipated to grow around 31 percent.

Goldman estimates the AI Enablers could together contribute around two percentage points to Nifty 500 earnings expansion in CY27/28. The upgrades are concentrated

The headline 22 percent upward revision in CY27 earnings estimates masks a wide gap across the AI infrastructure chain.

Data-centre hardware has noted the strongest upgrade at 95 percent year-to-date, followed by power equipment at 15 percent and semiconductor materials at 8 percent.

By contrast, data-centre developers have noted estimates trimmed 15 percent, while power-generation estimates have fallen 14 percent.

That distinction matters because Goldman’s AI basket spans businesses at very different stages of the infrastructure investment cycle.

The 42 firms cover three broad areas – power, data centres and semiconductors – and have a combined listed market capitalisation of around $ 670 billion.

India’s AI exposure is hiding below the index

At the benchmark level, India stays one of the least AI-exposed major equity markets.

Goldman estimates that only 16 percent of MSCI India’s market capitalisation is exposed to AI-related themes, compared with 70-80 percent in Korea and Taiwan and 30-50 percent in China and Japan.

That has helped create the perception of India as an “anti-AI” trade.

But Goldman argues that the benchmark misses a different part of the AI opportunity: firms supplying the physical infrastructure required for data centres, semiconductors and electricity-intensive computing.

The basket is heavily tilted towards infrastructure and capital goods rather than the large technology names that dominate AI exposure in other markets.

Goldman additionally notes that hard disclosures on AI-related topline stay limited among Indian firms. Instead, its screening picked up a sharp climb in AI and infrastructure-related language in earnings calls, including references to data centres, transformers, switchgear, substations, GPUs, OSAT and fibre infrastructure.

That makes the earnings revisions particularly important.

While AI-related topline is not yet separately disclosed across much of the basket, consensus earnings estimates for the group are moving in the opposite direction from the broader market.