Market news & insights

After three consecutive years in which mega-cap AI-linked names carried the Nasdaq, the mix of winners may be starting to change.

2026 is the "show me the money" year. Any hint of doubt about whether tech companies were correct to spend nearly US$700 billion on AI last year could have a major impact on market sentiment.

‍

Quick facts

• Global AI capex is projected to exceed US$600 billion in 2026.
• The total addressable market (TAM) for AI data centre systems is estimated to exceed US$1.2 trillion by 2030.
• Nvidia, Microsoft and TSMC are all trading below analyst fair value estimates, despite surging revenues.
• Broadcom's AI chip division is targeting US$100 billion in AI revenue by 2027.

‍

What is powering the AI trade?

Multiple macro forces are likely to underpin the AI investment theme through 2026. The direction of US interest rates, the scale of AI infrastructure spending and the geopolitical backdrop are all likely to matter.

Rates and valuations

The Federal Reserve delivered 75 basis points (bps) of rate cuts in 2025, and markets expect another 50 bps in 2026. Lower rates can reduce the discount applied to future tech earnings and typically support growth stocks, including AI-linked names.

Infrastructure spending and earnings expectations

On the spending side, Nvidia CEO Jensen Huang has said data centre operators could spend up to US$4 trillion annually by 2030, and AI capital spending is projected to reach US$571 billion in 2026 alone.

However, markets appear to have already priced in much of this optimism. Analysts are projecting 14% to 16% annual earnings per share (EPS) growth in 2026. That would require S&P 500 stocks outside the Magnificent 7 to roughly double the pace of earnings growth recorded in 2025.

Geopolitics and export controls

Geopolitics could also shape the outlook. US-China export controls on AI chips, along with reduced access to key international buyers, could weigh on data centre growth projections.

‍

Trade the US earnings season

‍

Top AI-linked stocks

Nvidia (NVDA) 

Nvidia remains the clearest expression of the AI trade. It holds a wide economic moat thanks to its market leadership in GPUs, hardware, software, and networking tools. 

Goldman Sachs and Morgan Stanley both carry price targets near $250 on NVDA, with Goldman's call based on a 2027 revenue forecast of over $380 billion. Bank of America sits in the $275 camp, effectively pricing in more AI upside on 2027 earnings.

At 21.6 times forward earnings, Nvidia is now trading below the broader S&P 500's multiple. Key risks include the overhang from US–China export restrictions and any softening in data centre capex guidance from major cloud providers.

Microsoft (MSFT)

Microsoft is down around 25% from its all-time high. During the second quarter of fiscal year 2026, Azure's revenue increased 39% year over year, and the company holds a US$625 billion backlog of contracted usage still to come. 

The gap between the stock's recent performance and its underlying revenue growth has drawn attention from analysts, though elevated valuations across the broader tech sector remain a risk to watch.

Broadcom (AVGO)

While Nvidia makes broad-purpose GPUs, Broadcom is winning business by going bespoke, designing custom AI chips tailored specifically to the needs of individual hyperscalers like Google and Meta.

During Q1 of FY2026, Broadcom's AI semiconductor division grew at a 106% pace to US$8.4 billion, and by the end of 2027 it expects its AI chip revenue to reach more than US$100 billion.

Broadcom trades at a significant premium to the broader market, which could amplify any downside if growth expectations are not met.

TSMC (TSM) 

Almost every major AI chip is manufactured by TSMC. The company holds approximately 70% market share in chip foundry, making it the single most critical piece of infrastructure in the entire AI supply chain. 

TSMC sales are projected to increase by 30% in 2026, with gross margins expected to remain above 60% as new fabrication capacity comes online.

The primary risk is geopolitical: any escalation in Taiwan Strait tensions could weigh heavily on the stock regardless of its underlying fundamentals.

Vertiv (VRT)

Less prominent than the semiconductor giants, Vertiv provides the power management, cooling, and data centre infrastructure that keeps AI hardware running. 

Nvidia, Broadcom, and Vertiv sit at different points in the AI build-out, including compute, custom silicon, networking and physical infrastructure.

Vertiv's revenue is tied to overall AI capex rather than any single chip maker, which gives it a different risk profile to the names above.

Corning (GLW)

Corning's stock rose 84% in 2025 thanks to surging demand from data centres for its fibre optic cables. Its optical communications segment has grown 69% YoY. 

At a Price-to-Earnings (P/E) ratio of roughly 37x, Corning trades at a discount to Nvidia and Broadcom while still carrying direct exposure to AI infrastructure spending. However, its valuation depends heavily on continued capex from the major hyperscalers.

‍

US market drivers for March 2026

‍

AI trades beyond the headline stocks

Energy and utilities

Training large-scale AI models is extraordinarily energy-intensive. A typical 1 gigawatt AI data centre facility requires upwards of US$60 billion in capital expenditure, with roughly half going directly to hardware. Utilities exposed to data centre power demand could also be affected by the AI build-out.

International spillover

South Korea's Kospi surged 76% in 2025 due to AI-linked chipmakers like SK Hynix. Japan's Topix, Germany's DAX, and the UK's FTSE 100 also saw gains of more than 20%. Memory supplier Kioxia was the world's best-performing stock, surging 540%.

Data centre infrastructure

Companies like Emcor, which provides critical electrical, HVAC, and power infrastructure to data centres, reported its contracted backlog surged 29% year over year to a record US$12.6 billion. These companies can offer different exposure to the AI capex cycle, but they carry their own execution, backlog, margin and valuation risks.

‍

What could derail the AI trade?

Valuation compression

Broadcom trades at about 50x earnings and AMD at 56x. Any disappointment in forward guidance could trigger a sharp contraction in multiples.

The return on investment test

Companies are investing today on the assumption that highly profitable business applications of AI will emerge over time. If the timing or scale of those returns disappoints, the AI trade could face pullbacks.

Index concentration

The 10 largest stocks in the S&P 500 account for about 40% of the index's total value. A rotation out of mega-cap tech could disproportionately affect broad indices.

Efficiency disruption

China's DeepSeek recently published research suggesting large language models may be developed more efficiently than previously assumed. If AI can be built with less compute, demand for GPUs and data centre hardware could fall short of current forecasts.

‍

Bottom line for traders

The AI trade is maturing but far from over. 2026 is shaping up to be a more nuanced chapter, spreading across the full AI value chain. 

The US earnings season will be closely watched for evidence that the hundreds of billions being poured into AI infrastructure are beginning to generate the anticipated returns.