Prediction: 2 Unstoppable Artificial Intelligence (AI) Hardware Leaders That Will Join Alphabet in the $4 Trillion Club by 2028

The article argues TSMC and Broadcom are poised to become new members of the $4T market-cap club on continued AI-driven demand. TSMC is projected to generate $24.46 EPS in 2028; at a 30x P/E, that implies ~$733/share and ~71% upside, with management citing strong chip demand through at least 2029–2030 and announcing an additional $100B investment in Arizona capacity. For Broadcom, the piece cites management saying its custom AI chip business will generate over $100B in 2027, with 2028 EPS of $21.28 implying ~$638/share (~52% upside) and potentially understated Street growth.

Analysis

TSM is the cleaner way to express the AI capex upcycle because it sits at the capacity bottleneck, not just the demand side. That makes it a structural beneficiary of every incremental dollar of hyperscaler spend, while also insulating it from some product-cycle risk that hits merchant chip vendors first. AVGO is more levered but also more fragile: its upside depends on custom silicon design wins translating into actual production ramps, so the market will reward it only if the revenue stream proves durable and not one-customer driven.

The near-term risk is that this is a long-duration story being priced as if it were already de-risked. If hyperscaler budgets pause for even one or two quarters, AVGO’s multiple can compress faster than TSM’s because the market will question the conversion from design wins to shipments; TSM’s falsifier is softer utilization or any sign that the multi-year demand visibility is overstated. The Arizona capex is strategically positive, but it also raises the bar for FCF discipline and return on invested capital in the next 12 months.

Contrarian take: the consensus may be underestimating how much of the AI buildout shifts from merchant GPUs to customized silicon and foundry capacity, which helps TSM and AVGO relative to NVDA. But the flip side is that both names already trade like quality compounders, so the easy money is likely in relative value rather than outright longs. The best risk/reward is to buy dips or use defined-risk structures, not chase strength after a narrative-driven upgrade cycle.

Key Decisions for Investors

• Accumulate TSM on 5-8% pullbacks for a 12-18 month hold; it is the highest-conviction AI infrastructure tollbooth. Falsify the thesis if lead times/utilization normalize or 2026 guidance stops implying multi-year growth.
• Prefer AVGO via Jan-2027 call spreads rather than outright stock if initiating now; the catalyst is 12-24 months out and the spread limits damage if custom-chip ramps slip.
• Small relative-value trade: long TSM / short NVDA into the next 1-2 quarters if you expect AI spend to rotate toward foundry and custom ASICs. Cover if NVDA re-accelerates or AVGO adds more hyperscaler wins than expected.
• Set an earnings alert on AVGO AI revenue concentration and TSM FCF/ROIC trends; if either company fails to raise 2026-2027 revenue or margin expectations, trim exposure because the current multiple already prices a long runway.