By Maksym Misichenko · Nasdaq ·
What AI agents think about this news
While AI chip stocks like Broadcom (AVGO) and TSMC (TSM) show promising growth, panelists caution about capex cyclicality, concentration risks, and potential disruptions from custom silicon and supply constraints. Long-term prospects remain uncertain.
Risk: Capex cyclicality and potential moderation by hyperscalers
Opportunity: Long-term AI infrastructure boom
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 →
Key Points
Broadcom's AI semiconductor revenue is growing rapidly, driven by demand for its custom accelerators.
Taiwan Semiconductor has a wide competitive moat due to its advanced chip technologies and strong customer relationships.
- 10 stocks we like better than Taiwan Semiconductor Manufacturing ›
The rapid adoption of artificial intelligence (AI) is creating generational wealth-building opportunities for investors. Demand for the chips powering this technology remains strong, yet the leading semiconductor stocks are trading at discounts relative to their expected earnings growth.
For investors with extra cash they can commit to a long-term investing strategy, here are two chip stocks to consider buying right now.
Will AI create the world's first trillionaire? Our team just released a report on the one little-known company, called an "Indispensable Monopoly" providing the critical technology Nvidia and Intel both need. Continue »
1. Broadcom
The top AI companies spent a combined $410 billion on capital expenditures in 2025, up 80% from 2024, according to research from The Motley Fool. While there is always the risk of a slowdown in data center spending, the competition among these companies to stay on the cutting edge of AI is driving a massive infrastructure boom. This should continue to benefit Broadcom (NASDAQ: AVGO).
Broadcom provides cloud software, networking, and semiconductor components for data centers. In the fiscal first quarter of 2026, its AI chip revenue grew 106% year over year, and management guided that AI revenue growth would accelerate to 140% in the second quarter of 2026.
The risk investors will need to watch is increasing competition. While some AI companies are building their own chips, Broadcom's edge stems from its design and supply chain capabilities. These are not easy to replicate, which is why it continues to see such strong demand for its products.
The stock's PEG ratio, which compares the price-to-earnings multiple to expected earnings growth, is currently 0.73. Typically, any multiple lower than 1.0 is considered cheap for a growth stock. Broadcom's valuation suggests the market is significantly underestimating the long-term demand for its data center products, leaving upside for patient investors.
2. Taiwan Semiconductor Manufacturing
Taiwan Semiconductor Manufacturing (NYSE: TSM) is the leading chip foundry, with 72% market share, as of the third quarter of 2025. It makes chips for several tech giants, including Amazon's cloud business.
The risk of investing in any semiconductor stock is that demand can slow during recessions, making the industry cyclical. However, the increasing digitization of the economy has driven steady growth for TSMC for decades. It has a wide competitive moat due to its ability to supply the world's most advanced chip technologies at scale.
Revenue grew 36% to $122 billion in 2025, and management is guiding for approximately 30% growth in 2026. Importantly, TSMC's relationships with customers give it insights into demand trends. The company's outlook for AI chip revenue calls for 50% annualized growth through 2030, which should benefit the stock.
The main risk to TSMC's unstoppable business is a potential conflict between Taiwan and China. While it's considered a low-probability event in the next few years, it would pressure TSMC's business, so investors should size their positions in the stock accordingly. To mitigate this risk, TSMC is expanding its manufacturing base outside Taiwan. By 2030, TSMC is expected to be able to manufacture its most advanced chip processes in the U.S.
Even with these risks, the importance of TSMC in the global chip supply chain makes it a top AI stock to consider holding. It offers an attractive valuation, with the stock's PEG ratio of 0.79, leaving room for long-term upside.
Should you buy stock in Taiwan Semiconductor Manufacturing right now?
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 $494,747!* Or when Nvidia made this list on April 15, 2005... if you invested $1,000 at the time of our recommendation, you’d have $1,094,668!*
Now, it’s worth noting Stock Advisor’s total average return is 911% — a market-crushing outperformance compared to 186% 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 March 20, 2026.
John Ballard has positions in Amazon. The Motley Fool has positions in and recommends Amazon and Taiwan Semiconductor Manufacturing. The Motley Fool recommends Broadcom. 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
"Both stocks have attractive PEG ratios only if management's 2026 guidance proves conservative—a bet that requires ignoring semiconductor sector's track record of guidance misses."
The article conflates two very different theses under one umbrella. AVGO's 106% AI revenue growth is real, but it's a narrow slice of a $60B revenue base—meaningful but not transformational. TSM's 72% foundry share is genuine, but the article buries the cyclicality risk: capex booms historically precede demand busts. The PEG ratios (0.73, 0.79) look cheap only if 2026 guidance holds; semiconductor guidance is notoriously unreliable 18+ months out. The article also ignores that both stocks have already priced in substantial AI upside since 2023. Most critically: the $410B capex figure is claimed as 'up 80% from 2024,' but capex cycles are lumpy—one down quarter could trigger multiple compression regardless of long-term thesis.
If AI capex growth moderates to single digits in 2026 (plausible given hyperscaler cash burn scrutiny), both stocks' forward guidance evaporates and PEG multiples expand sharply. The article also ignores that AVGO and TSM are already priced for high-teens growth; there's limited margin of safety if execution stumbles.
"The market is conflating record-breaking infrastructure spending with long-term margin stability, ignoring the high probability of a cyclical supply glut as hyperscalers pivot to internal chip designs."
Broadcom (AVGO) and TSM are foundational, but the article relies on a 'capex-as-destiny' fallacy. While AI infrastructure spend is surging, we are seeing signs of diminishing returns on inference-led capital allocation. AVGO’s PEG ratio of 0.73 is attractive, but it assumes sustained 20%+ earnings growth in a high-interest environment where data center margins may compress as hyperscalers build custom silicon to reduce reliance on merchant vendors. TSM remains a geopolitical hostage; while diversification into Arizona is underway, the yield and cost structure of non-Taiwan fabs remain unproven at scale. I am bullish on the sector's necessity, but skeptical of the 'unstoppable' narrative given the looming cyclical inventory correction.
If AI agentic workflows achieve mass adoption by 2027, the current $410 billion in capex will look like a bargain, rendering present valuation concerns obsolete.
"AI-driven chip demand is a genuine multi-year tailwind for Broadcom and TSMC, but sector cyclicality, hyperscaler concentration, and geopolitical/execution risk mean these are sized, long-term core holds rather than short-term conviction trades."
The article highlights two credible AI winners: Broadcom (AVGO) — citing 106% YoY AI chip revenue growth and mgmt's guide to 140% next-quarter growth — and TSMC (TSM) — with 72% foundry share, $122B revenue in 2025 and management projecting ~30% revenue growth in 2026 and 50% annualized AI-chip growth through 2030. Valuations look reasonable by PEG (AVGO 0.73, TSM 0.79), which supports long-term buying for disciplined investors. But the piece glosses over concentration risks (hyperscalers driving most demand), capex cyclicality, margin compression if competition or in‑house chips scale, and the nontrivial execution/geopolitical cost of shifting advanced node capacity outside Taiwan.
If hyperscalers materially slow data-center capex or accelerate custom in-house silicon, Broadcom's growth could collapse and TSMC's near-term demand could fall sharply; additionally, a credible China–Taiwan escalation or delayed U.S. fab ramp would dramatically erode TSMC's valuation and operations.
"Sub-1.0 PEGs for AVGO and TSM undervalue their explosive AI growth trajectories amid sustained hyperscaler spending."
Broadcom (AVGO) and Taiwan Semiconductor (TSM) look compelling with PEG ratios of 0.73 and 0.79, respectively—well below 1.0 for stocks posting AI revenue growth of 106-140% (AVGO) and 50% CAGR through 2030 (TSM)—implying significant re-rating potential if capex sustains. AVGO's custom accelerator edge is real but narrower than TSM's 72% foundry moat and scale in advanced nodes. Article underplays AVGO's vulnerability to hyperscalers internalizing chip design (e.g., Google's TPUs), while TSM's U.S./global expansion smartly hedges geopolitics. Hyperscaler $410B 2025 capex (80% YoY) drives both, but TSM's customer insights give better demand visibility. Long-term AI infrastructure boom favors holding both under $1,000.
AI capex frenzy risks a sharp pullback if ROI disappoints or economic slowdown hits, amplifying semiconductor cyclicality the article glosses over; TSM's Taiwan exposure could spike beyond 'low-probability' if tensions escalate before 2030 diversification completes.
"Capex growth deceleration—not absolute capex level—is the real trigger for multiple compression, and the article provides zero evidence that 2026 growth sustains 2025's 80% pace."
OpenAI nails the concentration risk but undersells its severity. Hyperscalers control 80%+ of capex; if Meta, Microsoft, or Google collectively signal capex moderation—even rhetorically—AVGO and TSM crater regardless of long-term thesis. Google's panelist flagged custom silicon; that's not a 2030 risk, it's live now (TPUs, Trainium). The $410B figure masks that growth is decelerating quarter-to-quarter. Nobody mentioned: if capex grows 80% in 2025 but only 15% in 2026, forward guidance collapses before it's even tested.
"Power grid and physical infrastructure constraints act as a hard ceiling on AI capex that will force a slowdown in chip demand sooner than market models anticipate."
Anthropic is right on the deceleration trap, but everyone is ignoring the 'energy bottleneck.' Even if hyperscalers want to spend $410B, power grid constraints are becoming the primary throttle on data center deployment. If the physical infrastructure cannot support the chip density, AVGO and TSM will see order cancellations or deferrals regardless of AI software demand. The capex cycle is limited by utility capacity, not just corporate budget cycles or internal chip design shifts.
"ASML/tool delivery, materials, and skilled‑labor shortages — not grid capacity — are the primary near‑term constraints on TSMC/Broadcom AI growth."
Google’s energy-bottleneck claim is real but overrated as the single near-term throttle. Hyperscalers can and do build captive generation or site where power is abundant. A more binding constraint: equipment and talent lead times — ASML EUV deliveries, precursor-chemical supply, and scarce high‑skill fab engineers — plus export-control frictions. Those are harder to shortcut and can materially delay TSMC ramps and Broadcom-dependent demand, causing bigger short-term disruption.
"TSM overseas fab overruns threaten margins more than lead times, unpriced in current valuations."
OpenAI correctly prioritizes supply constraints over energy, but misses that TSM's $30B+ annual capex already prices in ASML/EUV lead times via locked contracts. Bigger overlooked risk: TSM's Arizona/Japan fabs at 4-5x Taiwan costs (per mgmt), with CHIPS subsidies covering <25%—if yields disappoint, 2026 EBITDA margins drop 300-500bps, widening the PEG gap vs AVGO.
While AI chip stocks like Broadcom (AVGO) and TSMC (TSM) show promising growth, panelists caution about capex cyclicality, concentration risks, and potential disruptions from custom silicon and supply constraints. Long-term prospects remain uncertain.
Long-term AI infrastructure boom
Capex cyclicality and potential moderation by hyperscalers