🔴 Beyond the AI Behemoths: Market Broadening Signals Diversification Imperative as Tech Concentration Risks Mount

Fourth-quarter valuation jitters and Michael Burry’s depreciation warnings catalyse profit-taking rotation into small-caps, commodities and defence—yet TSMC’s record earnings underscore AI infrastructure boom remains intact despite portfolio rebalancing pressures

After a year dominated by tech sector outperformance that propelled the S&P 500 to consecutive record highs, early 2026 data reveals a pronounced broadening of equity demand across US markets. The rotation—small-caps surging 6% while large-cap tech gains remain subdued at 0.13%—reflects mounting investor concerns over concentration risk in artificial intelligence plays, coupled with compelling valuations in previously overlooked sectors. Yet this diversification impulse coexists uneasily with evidence that the AI infrastructure buildout retains structural momentum, creating strategic tensions for portfolio managers navigating 2026’s evolving landscape.

The Burry Warning That Rattled Fourth-Quarter Confidence

The catalyst for late-2025 tech anxiety originated from an unexpected source: Michael Burry, the investor immortalised for predicting the subprime mortgage crisis, trained his short-selling focus on AI hardware economics. Burry’s central thesis—that hyperscalers including Meta, Alphabet, Microsoft and Oracle systematically understate depreciation by extending GPU useful lives from realistic 2-3 year spans to optimistic 5-6 year schedules—struck at the heart of tech earnings credibility. His calculations suggested the industry could be overstating combined earnings by approximately $176 billion between 2026-2028, with companies like Oracle potentially inflating profits by 27% and Meta by 21%.

Join The European Business Briefing

The daily email on markets, technology, power and money across Europe. Join 10,000+ founders, investors and executives who read EBM every morning.

Subscribe

The depreciation controversy highlighted a fundamental tension in AI capital economics. Nvidia’s GPU architecture refresh cycle operates on 12-18 month cadences—the Blackwell generation launched in 2024 will be superseded by Rubin chips in H2 2026, offering 3x performance improvements. This rapid obsolescence makes 6-year depreciation schedules appear divorced from technological reality. Meta’s decision to extend server useful lives from 4.5 to 5.5 years, reducing 2025 depreciation expense by $2.9 billion, exemplified the accounting manoeuvre Burry characterised as “one of the more common frauds of the modern era.”

While critics noted that these practices comply with GAAP accounting standards and that non-cash depreciation charges differ from actual cash flows, the controversy succeeded in elevating valuation scrutiny. Investors confronting tech multiples approaching historic peaks—the Magnificent Seven trading at 35x forward earnings—began questioning whether AI spending justified current prices or represented a capital misallocation bubble awaiting correction.

The Broadening Rally: Small-Caps and Commodities Ascendant

Market rotation accelerated through late 2025 and early 2026 as capital flowed toward previously lagging sectors. Small-cap companies—trading at 18x forward P/E versus the S&P 500’s 24x, a 22% discount near historic lows—attracted value-seeking investors. The Russell 2000 small-cap index outperformed large-cap benchmarks by widest margins since 2020, driven by multiple catalysts: lower interest rates reducing financing costs, the “One Big Beautiful Bill” Act providing fiscal stimulus, and Trump administration policies favouring domestic-focused businesses.

Sector rotation proved equally dramatic. Basic materials surged 9.05% year-to-date through mid-January 2026, followed by industrials and energy. Gold, metals and mining companies posted “outstanding performances” according to State Street’s Michael Arone, benefiting from geopolitical tensions surrounding Trump’s Greenland annexation threats and resulting commodity price spikes. Defence contractors capitalised on renewed European military spending and deteriorating transatlantic relations, providing fresh growth narratives beyond AI infrastructure.

Within the Magnificent Seven cohort itself, performance diverged sharply. Alphabet gained 6% and Amazon 3% through January, while Apple, Microsoft, Meta and Tesla declined. Nvidia barely edged positive. This intra-tech dispersion reinforced perceptions that the AI rally—which lifted the Magnificent Seven to represent over 40% of S&P 500 market capitalisation—had reached saturation, prompting profit-taking toward cheaper valuations elsewhere.

TSMC’s Earnings: The AI Infrastructure Case Remains Robust

Yet dismissing AI momentum as exhausted would misread the structural forces underpinning semiconductor demand. Taiwan Semiconductor Manufacturing Company’s Q4 2025 results—announced January 2026—delivered resounding confirmation that leading-edge chip demand remains extraordinary. TSMC reported record quarterly revenue of NT$1.046 trillion ($33.7 billion), up 25.5% year-over-year, with net income surging 35% to NT$505.7 billion. The company guided Q1 2026 revenue to $34.6-35.8 billion, representing 38% year-over-year growth at midpoint, dramatically exceeding analyst expectations of $33.2 billion.

The foundry giant’s advanced node mix tells the structural story: 3-nanometer processes contributed 28% of Q4 wafer revenue, with chips measuring 7nm or smaller comprising 77% of total sales. High-performance computing—overwhelmingly AI accelerators and data centre processors—accounted for 58% of 2025 net revenue, up from 51% in 2024. Management raised AI accelerator five-year growth outlook to approach mid-to-high 50% CAGR through 2029, reflecting deepening engagement with hyperscaler customers planning multi-year capacity expansions.

Most tellingly, TSMC announced 2026 capital expenditure of $52-56 billion—a substantial increase from $40.9 billion in 2025—with 70-80% allocated to advanced process technologies and packaging. CEO C.C. Wei characterised capacity as “very tight” with engagement cycles extending 2-3 years ahead, underscoring that demand substantially exceeds supply despite record investment. The company’s Arizona gigafab cluster expansion, now encompassing six advanced fabs following additional land purchases, reflects confidence that AI infrastructure buildout will sustain for years rather than quarters.

Strategic Implications: Balancing Concentration Risk Against Generational Opportunity

The divergence between tech sector rotation and semiconductor demand strength creates nuanced portfolio allocation challenges. On one hand, concentration risk merits serious consideration—the top 10 S&P 500 companies commanding over 40% of index market capitalisation creates systemic vulnerability if sentiment turns decisively negative. Burry’s depreciation thesis, while potentially overstated, highlights genuine accounting uncertainties that could trigger earnings disappointments if capital expenditure returns disappoint or hardware obsolescence accelerates beyond management assumptions.

Moreover, broader market fundamentals appear healthy beyond tech. The remaining 493 S&P 500 companies forecast 13% earnings growth for 2026—respectable performance suggesting economic resilience extends across sectors. Small-cap earnings approaching double-digit growth for first time in years supports the broadening thesis. Keith Lerner of Truist Advisory Services articulates the consensus: “I wouldn’t get rid of tech at this point. I think there are more opportunities beyond tech.”

Yet TSMC’s results demonstrate that AI infrastructure investment operates on multi-year capital cycles resistant to short-term sentiment swings. Hyperscalers have committed hundreds of billions to data centre construction, with over $700 billion projected AI infrastructure spending in 2026 alone. The transition from simple text-based chatbots to complex multi-modal AI, agentic systems and robotics applications demands exponentially greater computing power—a secular trend in its infancy rather than maturity.

The Trump Factor: Policy Unpredictability Drives International Diversification

The “unpredictable nature of Donald Trump”—evidenced by erratic tariff threats, territorial expansion demands and alienation of traditional allies—introduces additional portfolio risk beyond tech concentration. The US dollar’s weakness through early 2026, combined with European fiscal stimulus prospects and attractive international equity valuations (trading near 40-year lows at 27.5% of MSCI World Index versus 48.7% historical average), creates compelling case for geographic diversification.

Currency hedging costs climbing amid dollar volatility incentivises international capital reallocation. Trump’s commodity-positive policies—infrastructure spending, protectionism favouring domestic materials producers, geopolitical tensions elevating safe-haven demand—provide fresh narratives for investors seeking non-tech exposure.

Risk Management in a Generational Tech Cycle

The fundamental investment challenge confronting 2026 portfolios centres on mitigating concentration risk without abandoning exposure to a genuine technological revolution. The AI boom’s eventual peak timing remains unknowable—history suggests that transformative technology cycles (railroads, electricity, internet) experience multiple boom-bust oscillations before reaching maturity. Any correction will likely prove transitory as capital gravitates back toward dominant AI infrastructure providers once valuations reset.

The emerging consensus advocates balanced exposure: maintain core AI positions while selectively rotating profits into value stocks, small-caps, commodities and international markets offering superior risk-adjusted returns. The broadening seen through late 2025 and early 2026 likely foreshadows a “more even” market where gains distribute across sectors rather than concentrating in tech behemoths.

Yet abandoning tech entirely poses equal risk to excessive concentration. TSMC’s record earnings underscore that AI infrastructure demand remains structurally robust despite sentiment volatility. The “sheer size of the tech sector” means it “remains the one area that could take the entire market down”—but equally, missing the remaining upside of a generational boom represents opportunity cost many portfolios cannot afford.

Fortunately, the confluence of Trump-driven commodity rallies, defence sector surges and small-cap value opportunities provides multiple avenues for prudent diversification. The imperative for 2026 portfolios combines conviction in AI’s long-term trajectory with tactical flexibility to capture broadening opportunities as markets evolve beyond their 2025 concentration extremes.

Further Reading:

RELATED ARTICLESMORE FROM AUTHOR