Back to News
Market Impact: 0.38

Two Taiwanese Stocks With Strong AI Exposure, According To Bernstein

TSMNVDAGOOGLAMZN
Artificial IntelligenceTechnology & InnovationCorporate Guidance & OutlookAnalyst InsightsCompany FundamentalsCorporate Earnings
Two Taiwanese Stocks With Strong AI Exposure, According To Bernstein

Bernstein highlighted Unimicron and TSMC as direct beneficiaries of accelerating AI infrastructure spending, with Unimicron expected to get about 50% of revenue from AI in 2026 and ~35% ABF share in Nvidia high-end GPUs. Bernstein also pointed to stronger pricing, utilization gains, and roughly 50% growth in Unimicron AI HDI/PCB revenue in 2026, while TSMC raised 2026 revenue growth guidance and narrowed capex to the high end of its $52B-$56B range. TSMC earnings are forecast to grow about 40% in 2026, reinforcing a constructive AI-led outlook for Taiwan semiconductor supply chain names.

Analysis

The incremental read-through is not just “AI is strong,” but that the supply chain is shifting from scarcity in logic to scarcity in substrates and packaging. That matters because the next leg of AI capex increasingly accrues to a narrower set of bottleneck vendors with pricing power, while broader semiconductor cyclicals still face end-demand fragility outside AI. TSM is the cleaner expression of that bifurcation: if the AI buildout stays intact, it can compound earnings even if consumer electronics remain soft, while component suppliers like Unimicron offer higher beta but more execution risk. Second-order winners are the companies selling enabling capacity into this constraint regime: ABF, HDI, advanced packaging, and high-end foundry tools/services. The market may still be underestimating how long these bottlenecks persist; even with new fabs and line ramps, the cash conversion is likely to lag demand by 2-4 quarters, which preserves pricing power into 2026. That argues for owning the picks-and-shovels with visible utilization inflection and avoiding names where AI exposure is still aspirational rather than contract-backed. The main risk is that AI capex expectations get too linear from here. If hyperscalers pause after the current build cycle, the “AI half of revenue” narrative for upstream suppliers can de-rate quickly because the stock is already trading on forward utilization and margin expansion. For TSM, the contrarian concern is less demand destruction than expectation saturation: the market already treats it as the default AI winner, so upside likely requires either further capex upside or clear evidence that 2027-28 growth is being pulled forward. The overdone/underdone debate is split: TSM may be somewhat crowded, but the packaging/materials layer is still under-owned relative to its leverage to AI server content per unit. Investors are likely underestimating how much of the next 12 months’ upside comes from pricing, mix, and utilization rather than unit growth alone. That favors a barbell of quality core exposure with a tactically sized higher-beta supplier leg.