Hardware vs. Software: How Global Investors Are Splitting Their AI Bets Between South Korea and India

Much of the commentary around 2026’s capital flows has framed it as a single story: money leaving India and money rushing into South Korea. That framing does not hold up particularly well against the actual data which says that foreign investors have, in fact, been net sellers in Korea too, even as its market surged in value.

The more accurate and, frankly, more useful story is not about money fleeing one country for another. It is about how global fund managers are splitting their AI exposure across two very different layers of the AI value chain, hardware and software, and why that distinction is reshaping capital allocation across two of Asia’s most closely watched equity markets.

Two Countries, Two Very Different Roles in AI

South Korea’s place in the AI story is heavily tied to the physical infrastructure behind it. Samsung Electronics and SK Hynix sit at the heart of global high-bandwidth memory (HBM) production, the specialised memory used in many of the leading AI accelerators produced by Nvidia and its peers. When data centre operators around the world order next-generation AI infrastructure, part of that demand ultimately feeds through to Korean memory production.

This helps explain why SK Hynix and Samsung came to represent more than half of Korea’s stock market capitalisation during the 2026 rally. The country’s public market has, in effect, become highly concentrated around the physical build-out of AI infrastructure.

India’s relationship with AI looks very different. Its listed-market exposure is not primarily about manufacturing the chips themselves; it is about the technology services, engineering talent and enterprise support layer that sits on top of AI infrastructure i.e. IT outsourcing firms, software services companies and, increasingly, AI-adjacent engineering and consulting work for global businesses trying to integrate AI into their operations.

This represents a fundamentally different type of economic exposure. Korea is more directly tied to the capital expenditure being poured into data centres, semiconductor capacity and memory production. India’s IT sector is more closely tied to the ongoing enterprise spending required to deploy, integrate and operate AI-powered systems inside businesses.

Why This Distinction Matters for Fund Managers

This hardware-versus-software distinction explains a lot about how global capital allocators have been behaving, and it offers a more precise lens than simply saying “India is losing and Korea is winning”.

When AI capital spending is accelerating (as it clearly has through 2025 and 2026, with hyperscalers racing to expand data centre capacity), hardware and infrastructure companies are among the most direct and immediate beneficiaries. The relationship is relatively straightforward: more demand for AI creates more demand for advanced memory, and Korea’s public markets offer some of the most liquid, large-cap ways to gain exposure to that part of the AI supply chain.

This is part of why Korean equities have become increasingly important as an early indicator of sentiment around the global AI trade. Bloomberg reported in July that some global fund managers had begun checking Korean stocks before other major markets opened, particularly because movements in Samsung and SK Hynix were increasingly feeding through to global semiconductor sentiment.

Software and services exposure, by contrast, is a slower-moving and more indirect bet. It depends on enterprise IT budgets, corporate spending cycles and the pace at which businesses actually operationalise AI tools, all of which have faced pressure in 2026 as client spending has remained cautious in parts of the global technology sector.

That softness has weighed particularly on India’s large IT services exporters, since a meaningful share of their revenue depends on enterprise technology spending and new project activity.

So when fund managers rotate away from India’s IT sector, it is not necessarily a verdict on India’s AI story. It may instead reflect where we currently are in the AI investment cycle: the infrastructure-building phase is well underway, while the phase in which AI consistently generates measurable enterprise software and services revenue is still developing.

Two Different Investment Theses, Not a Zero-Sum Trade

It is worth being precise here: these are not necessarily substitutes for each other in a portfolio. An investor buying SK Hynix and an investor buying an Indian IT services company are not expressing exactly the same view about AI upside but they are expressing different views about when and how AI value gets captured.

The hardware thesis benefits earlier and more directly from capital expenditure announcements by hyperscalers. The software and services thesis is a longer-duration view that AI adoption will eventually translate into durable demand for the people, platforms and engineering capabilities required to implement it inside real businesses, a process that can lag the infrastructure build-out.

This is why it may be more useful to think of global allocators as adjusting the mix between these two exposures within a broader regional or Asia-AI portfolio, rather than simply “moving out of India and into Korea”. While conditions favour infrastructure spending, investors may place greater weight on the more immediate hardware exposure and less on the slower-payoff software story.

It is a rebalancing within a broader thesis about AI, rather than necessarily a rejection of one country in favour of another.

What Could Shift the Balance Back

A few dynamics are worth watching, because this hardware-software split is unlikely to remain static.

First, if AI capital expenditure growth decelerates (a scenario worth considering given how concentrated Korea’s rally has become in two companies) the hardware trade becomes less mechanically attractive, and fund managers may start looking towards the next stage of the AI story: monetisation through software and enterprise deployment, where India has significant capabilities.

Second, as enterprises move from experimenting with AI tools to embedding them at scale, demand for the integration, customisation and engineering services in which Indian IT firms specialise could, in principle, strengthen. That would represent a more lagged, but potentially meaningful, beneficiary effect from AI adoption.

Third, currency and macroeconomic conditions matter independently of the AI story. A more stable rupee and easier global financial conditions could improve foreign appetite for Indian equities, although the impact of currency movements on export-heavy IT companies is more nuanced because exchange rates also affect their revenues and margins.

My Take

The most interesting part of this story, to me, is that it is not really about India versus Korea at all, it is about where we are in the AI investment cycle.

Right now, markets are rewarding the parts of the AI build-out where demand is most immediate and most visible, and that includes semiconductor infrastructure and advanced memory, areas where South Korea is exceptionally strong. That helps explain why fund managers have placed greater emphasis on Korean hardware exposure during this phase of the cycle.

But I would be cautious about interpreting that as evidence that India’s AI positioning is fundamentally weak. It may simply sit at a different point on the payoff curve.

The services and software layer of AI adoption is inherently slower to show up in quarterly numbers than a memory-chip order. Dismissing that opportunity simply because the capital expenditure phase currently attracts more attention risks confusing sequencing with substance.

If previous technology cycles offer any useful lesson, the “picks and shovels” phase of a technology boom is rarely where the story ends. Eventually, more of the debate turns towards who can help the wider economy actually use the technology productively.

That is a race India is still very much in!

Sources:

1. Korea Economic Institute of America: “Explaining South Korea’s Stock Market Boom” https://keia.org/analysis/explaining-south-koreas-stock-market-boom/
2. CSIS: “South Korea’s Market Boom and the Bubble Beneath It” https://www.csis.org/analysis/south-koreas-market-boom-and-bubble-beneath-it
3. Bloomberg: “Korea’s AI-Heavy Market Now Sets the Tone for Global Stocks” https://www.bloomberg.com/news/articles/2026-07-19/korea-s-ai-heavy-market-now-sets-the-tone-for-global-stocks
4. Outlook Money: “FII Sell-off: Why Foreign Investors Sold Rs 2.06 Lakh Crore in 2026” https://www.outlookmoney.com/invest/fii-sell-off-hits-rs-206-lakh-cr-in-2026-whats-behind-the-resumed-exodus
5. Bajaj Finserv: “Why FIIs Are Selling Indian Stocks in 2026” https://www.bajajfinserv.in/why-fii-are-selling
6. Multibagg: “FII outflows from India: what’s driving 2026 selloff” https://www.multibagg.ai/market-pulse/articles/fii-outflows-india-2026-drivers-cmq1wiqzbgfuzps0julkww05w
7. Malik Times: “FII Outflows India 2026: Why $29 Billion Left the Market” https://maliktimes.in/fii-outflows-india-2026-explained/

Note: This article reflects market data and commentary available as of August 2026.

Disclaimer: This article is for educational and informational purposes only and should not be considered financial, investment or trading advice. The views and interpretations expressed are those of the contributor and do not constitute a recommendation to buy, sell or hold any company, security or market exposure discussed. Market conditions and investor positioning can change, and individual circumstances differ. Always conduct your own research before making investment decisions.