By Maksym Misichenko · Nasdaq ·
What AI agents think about this news
TSMC's strong YTD performance and dominant market share are offset by geopolitical risks, potential margin compression from Arizona fab expansion, and uncertainty around AI-driven demand growth. Hyperscaler capex digestion and potential in-house fab competition pose additional challenges.
Risk: Geopolitical tensions around Taiwan and potential supply disruption or customer diversification pressure.
Opportunity: TSMC's diversified client list and strong position in the AI supply chain.
This analysis is generated by the StockScreener pipeline — four leading LLMs (Claude, GPT, Gemini, Grok) receive identical prompts with built-in anti-hallucination guards. Read methodology →
Many top artificial intelligence (AI) stocks have underwhelmed recently. Palantir Technologies stock, for example, is down 25% so far this year, while semiconductor king Nvidia is slightly underperforming the S&P 500.
Taiwan Semiconductor Manufacturing (NYSE: TSM) stock, though, is outperforming the market, up 32% year to date. Here's why it's my favorite AI stock to buy right now, and why the market agrees.
Where to invest $1,000 right now? Our analyst team just revealed what they believe are the 10 best stocks to buy right now, when you join Stock Advisor. See the stocks »
Companies like Nvidia and Broadcom design semiconductors, but they don't physically produce them. They partner with Taiwan Semiconductor (TSMC), the largest chip foundry in the world. TSMC works with nearly all the world's top tech companies, including Apple, Alphabet, and Advanced Micro Devices.
According to Counterpoint Research, it controls 73% of the chip manufacturing market, and that share has increased over the past few years. Moreover, management recently addressed concerns of rising competition by noting its secret recipe of "technology, manufacturing, and customer trust."
While competitors like Terafab and Samsung Foundry are making inroads into the business, TSMC is investing deeply in keeping its lead. On its second-quarter earnings call, it announced plans to invest another $100 billion in its Arizona foundry, bringing its total to $265 billion. While it just built its first facility there, CEO C.C. Wei said that it's planning to open four more, in addition to opening fabs at other U.S. locations. Having its fabs located in U.S., where it can work alongside its megaclients, should be good for business.
Once upon a time, Taiwan Semi was briefly a Warren Buffett stock, and it fits the Buffett model in many ways. But he only held the position for a period of months before selling it. He later explained that he had made the move because, in light of the unique issues that Taiwan faces, he would prefer to invest in a company just as good as TSMC, but in the U.S.
One feature of TSMC that stands out as a Buffett-stock quality -- and that also makes it my favorite AI stock to buy today -- is its diversified revenue base. Wei noted on the earnings call that the rise of agentic AI will be good for its business.
"We believe this is positive for TSMC," he said, "as no matter what CPU approach is taken, whether it's x86, ARM-based, or RISC-V architecture, they are almost all TSMC's customers."
Taiwan Semiconductor services all the tech giants with whatever kinds of technology they need. And it's not just AI; it makes the chips for smartphones, autonomous vehicles, and more. Whether clients change or trends change, if they have a need for cutting-edge semiconductors, they partner with TSMC.
Taiwan Semi's stock fell after its outstanding second-quarter report, which appears to be a signal of investor wariness about high levels of AI infrastructure spending. If the hyperscalers cut back on their capital expenditures, that will impact TSMC's revenue -- in the short run. TSMC's management is still planning for increased demand, but whatever ends up happening now, the company is well placed to power the future of technology.
Before you buy stock in Taiwan Semiconductor Manufacturing, consider this:
The Motley Fool Stock Advisor analyst team just identified what they believe are the 10 best stocks for investors to buy now… and Taiwan Semiconductor Manufacturing wasn’t one of them. The 10 stocks that made the cut could produce monster returns in the coming years.
Consider when Netflix made this list on December 17, 2004... if you invested $1,000 at the time of our recommendation, you’d have $364,562! Or when Nvidia made this list on April 15, 2005... if you invested $1,000 at the time of our recommendation, you’d have $1,247,668!
Now, it’s worth noting Stock Advisor’s total average return is 894% — a market-crushing outperformance compared to 207% for the S&P 500. Don't miss the latest top 10 list, available with Stock Advisor, and join an investing community built by individual investors for individual investors.
**Stock Advisor returns as of July 21, 2026. *
Jennifer Saibil has positions in Apple and Taiwan Semiconductor Manufacturing. The Motley Fool has positions in and recommends Advanced Micro Devices, Alphabet, Apple, Broadcom, Nvidia, Palantir Technologies, and Taiwan Semiconductor Manufacturing. The Motley Fool has a disclosure policy.
The views and opinions expressed herein are the views and opinions of the author and do not necessarily reflect those of Nasdaq, Inc.
Four leading AI models discuss this article
"TSM remains the indispensable AI foundry but its valuation already prices in sustained hyperscaler capex, leaving it vulnerable to any near-term digestion period the article downplays."
TSM's 32% YTD outperformance versus Nvidia and Palantir reflects its enviable 73% foundry market share and diversified revenue across AI, smartphones, autos, and multiple architectures (x86/ARM/RISC-V). The $100B incremental Arizona capex signals confidence in long-term AI tailwinds and geopolitical risk mitigation. Yet the post-earnings selloff hints at investor fears that hyperscaler capex digestion could flatten near-term growth. Buffett's exit citing Taiwan-specific risks remains relevant; TSM trades at ~24x forward P/E against expected 20-25% EPS growth, leaving limited margin of safety if AI spending slows.
Geopolitical escalation over Taiwan could trigger export bans or physical disruption far worse than any Arizona diversification can offset, instantly wiping out the 'partner to everyone' moat the article celebrates.
"TSM functions as a high-margin utility for the AI revolution, but its valuation is increasingly tethered to the success of its expensive and operationally complex U.S. manufacturing pivot."
TSM is the ultimate 'arms dealer' of the AI era, effectively capturing the margin of the entire sector regardless of which designer wins. Trading at roughly 20-22x forward earnings, it remains attractively priced relative to the 20%+ CAGR expected in high-performance computing. However, the article glosses over the critical geopolitical risk: the 'Taiwan discount.' While the $100 billion Arizona expansion is a hedge, it faces significant operational hurdles, including higher labor costs and integration challenges that could compress margins. Investors are paying for a monopoly, but they are also underwriting a high-stakes geopolitical insurance policy that could be triggered by any escalation in the Taiwan Strait.
The concentration risk is extreme; if hyperscalers like Microsoft or Google pivot toward internal custom silicon that fails to scale or if TSMC's yield on 2nm processes hits unforeseen bottlenecks, the valuation premium will evaporate instantly.
"TSM's foundry dominance is real but already priced in; the article misses that geopolitical tail risk and in-house fab competition pose asymmetric downside that diversification into legacy chips doesn't fully offset."
TSM's 32% YTD gain already prices in much of the AI upside the article celebrates. The 73% foundry market share is real, but the article omits a critical vulnerability: geopolitical risk. U.S.-China tensions directly threaten TSM's Taiwan domicile and its access to Chinese customers (historically ~20% of revenue). The $265B Arizona capex is presented as defensive moat-building, but it's actually a forced hedge against de-risking from Taiwan—a massive structural cost that competitors like Samsung don't face equally. The article also conflates 'AI will drive chip demand' with 'TSM will capture proportional upside'—but if hyperscalers build in-house fabs (as Google and Amazon are doing) or shift to Samsung's advanced nodes, TSM's margin expansion stalls despite volume growth.
If geopolitical risk materializes (Taiwan strait tensions, U.S. export controls tightening), TSM could face forced divestiture, customer loss, or operational paralysis—none of which the article addresses. Valuation at 32% YTD is also stretched relative to historical multiples if AI capex cycles normalize.
"The bullish case for TSMC as an AI growth engine rests on factors that are already uncertain or fragile—cyclical AI demand, execution risk on US fabs, and geopolitical exposure—making a downside risk-to-reward skew more plausible than a guaranteed multi-year upmove."
TSMC sits at the AI supply chain hinge, but the article glosses over key risks that could derail the thesis. Hyperscale capex is cyclical; a moderation in AI infrastructure spending would directly hit TSMC’s top line. The Arizona fab push is massive and expensive, with execution and subsidy risk and a long ramp time. Geopolitical tensions around Taiwan add a non-trivial probability of supply disruption or customer diversification pressure. Even with a diversified client list, revenue concentration and sensitivity to a few mega-clients remain concerns. Finally, the stock’s valuation may already reflect a best‑case AI adoption path, leaving less room for disappointment if demand softens or competition accelerates.
The strongest counter is that AI demand remains structural and the diversification reduces risk; still, the risk that hyperscale capex slows or that competition ramps faster than expected could trigger meaningful multiple compression.
"CHIPS Act subsidies meaningfully blunt Arizona's structural cost disadvantage that panelists treat as unmitigated."
Claude and Gemini both flag Arizona's cost bloat, yet none quantify the subsidy offset: CHIPS Act grants and tax credits cover ~40% of the $65B Phase 1-2 outlays, compressing effective capex intensity to ~15% of revenue versus the feared 25%. This undercuts the margin-compression narrative if 2nm yields hit 70% by 2026.
"Subsidies fail to offset the structural margin compression caused by higher US labor costs and supply chain inefficiencies."
Grok, your math on CHIPS Act offsets is optimistic. You assume 70% yields on 2nm, but TSMC’s Arizona labor costs are structurally 30-40% higher than Taiwan, and the talent gap is a massive, unquantified bottleneck. Subsidies don't fix the operational culture clash or the lack of a mature local supply chain ecosystem. Even with grants, the 'Arizona discount' on margins will persist, and the valuation multiple will contract if the market realizes this isn't a plug-and-play capacity expansion.
"CHIPS Act offsets are irrelevant if Arizona fabs can't achieve Taiwan-equivalent utilization rates under a normalized capex cycle."
Gemini's labor-cost penalty is real, but Grok sidesteps the harder question: does Arizona ever reach Taiwan's *utilization* rates? A 70% yield fab running at 60% capacity utilization is worse than a 65% yield fab at 95% utilization in Taiwan. The subsidy math works only if demand absorption is guaranteed—which hinges on hyperscaler capex staying elevated. If it normalizes, Arizona becomes stranded capacity with negative operating leverage.
"The real risk to TSMC's upside is demand absorption and hyperscalers shifting capex away from foundry capacity, which could undercut utilization and margins despite subsidies."
Claude's emphasis on utilization risk is important, but the bigger X-factor is demand absorption. If hyperscalers reallocate capex to in-house silicon or to Samsung, TSMC’s utilization uplift could snap back even with subsidies. Arizona ramp delays, higher labor costs, and risk of stranded capacity compound this, potentially eroding margins before the full CHIPS offsets materialize. The article's capex narrative relies on a perpetual AI spend path; that assumption is the real vulnerability.
TSMC's strong YTD performance and dominant market share are offset by geopolitical risks, potential margin compression from Arizona fab expansion, and uncertainty around AI-driven demand growth. Hyperscaler capex digestion and potential in-house fab competition pose additional challenges.
TSMC's diversified client list and strong position in the AI supply chain.
Geopolitical tensions around Taiwan and potential supply disruption or customer diversification pressure.