Semiconductor earnings, hyperscaler capital plans and the rising cost of electricity push US benchmarks to fresh highs, prompting closer scrutiny of valuations, concentration risk and the next wave of tech deal-making.

Major US equity benchmarks are setting fresh records in Thursday trading, and Sunnov Investment Pte. Ltd. is tracking whether the artificial intelligence rally is supported by earnings and infrastructure capacity rather than momentum alone. Advanced Micro Devices is jumping 18.6% in the session, while the S&P 500 is closing up 1.5% at 7,365.1 points and the Nasdaq Composite is rising 2.0% to 25,838.9, with the Dow adding 356.0 points.

As of the latest close, the Magnificent Seven account for about 33.3% of the S&P 500’s market capitalisation, giving a small set of balance sheets outsize control over index direction and over the terms of technology financing. Nvidia’s most recent quarter shows revenue of $54.4 billion, up 62.0% from the comparable period a year earlier, and its outlook implies about $62 billion for the coming quarter.

Among those reframing the sector is Thomas Gardner, director of private equity at Sunnov Investment Pte. Ltd., who describes the shift as “a re-rating of nuclear from optionality to necessity as AI turns electricity into the binding constraint for growth”. The energy arithmetic is starting to shape market expectations: in the latest full year dataset, global data centres consume about 415.0 terawatt hours, roughly 1.5% of worldwide electricity demand, with consumption rising around 12.0% a year over the preceding five-year span; accelerated computing linked to AI expands near 30.0% a year over the same span, compared with about 9.0% for conventional server infrastructure.

Projections followed by Gardner put global data-centre demand near 945.0 terawatt hours by the end of the decade and suggest US facilities could reach about 12.0% of national electricity supply within the next two years. Over the forthcoming infrastructure cycle, hyperscalers including Microsoft, Alphabet, Amazon and Meta face projected outlays of about $467.6 billion, up from projections near $400.8 billion in the prior estimate round, and partnership commitments such as OpenAI’s plan to deploy 10.0 gigawatts of Nvidia systems over the coming build-out and Nvidia’s indicated $95.4 billion investment in OpenAI as capacity becomes operational are reinforcing the build-out.

Rotation inside equities is also visible. Over the latest four-week stretch, quality factors lag value by more than 5.0 percentage points, the weakest relative showing in five years, while dividend and buyback exchange-traded funds attract about $6.7 billion of inflows over the same four-week window, feeding the HALO trade of heavy assets and low obsolescence.

In its latest quarter, AMD reports earnings per share of $1.3 versus a $1.2 consensus on revenue of $9.8 billion versus expectations near $9.4 billion, and it guides the coming quarter to $10.4 billion to $11 billion, above estimates around $10 billion; if the move holds, market capitalisation expands by more than $82.1 billion in a single session. The Philadelphia Semiconductor Index sits up roughly 62.0% over the past four months of trading.

On Morningstar’s latest fair value estimates, Nvidia trades about 22.0% below $248.1 and Microsoft sits roughly 30.0% below $572.5, pointing to dispersion that supports selective exposure rather than blanket positioning.

BCA Research outlines a late-cycle surge that could lift indices by 30.0% over the next year and take the S&P 500 beyond 9,200.0, whilst investment-grade technology credit spreads widen versus the broader market over recent weeks. Filings covering the second quarter of the preceding year show Saudi Arabia’s Public Investment Fund trimming US equity exposure from $24.3 billion to $22.7 billion, and prime-brokerage data show hedge funds cutting US technology exposure at the fastest pace in 13.0 years over the latest monthly reporting window.

Over a recent four-quarter span, a high-growth marketing technology company reports revenue growth of 38.0% while costs rise 349.0% as AI infrastructure spending ramps, and in remarks late last year Federal Reserve Chair Jerome Powell characterises US equity prices as “fairly highly valued”, with technology shares falling 1.3% in the following session.

For the next full-year outlook, global AI spending is projected at about $1.9 trillion, and McKinsey’s modelling puts generative AI’s potential economic contribution as high as $4.2 trillion a year. In the latest annual survey cycle, 88.0% of respondents report regular AI use in at least one business function, up from 78.0% in the prior cycle.

Sunnov Investment sees the rally’s durability resting on whether power infrastructure scales as quickly as capital budgets, and whether portfolios are built to survive concentration risk as the trade matures. Gardner’s conclusion is that “AI is moving from a theme to a balance-sheet reality, and portfolios that separate infrastructure winners from valuation excess are better placed to hold exposure through the next phase”.

About Sunnov Investment

Serving accredited investors, foundations and endowments worldwide, Sunnov Investment is a Singapore-based investment manager founded in 2012. The firm runs long-only equity strategies alongside long/short equity, global macro, event-driven and systematic mandates, and develops structured routes for eligible retail participation.

• Website: https://sunnov.com 

• Media enquiries should be directed to Deng Hui at d.hui@sunnov.com 

• The business is registered as Sunnov Investment Pte. Ltd., UEN 201225494E.